<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"
     xmlns:content="http://purl.org/rss/1.0/modules/content/"
     xmlns:dc="http://purl.org/dc/elements/1.1/"
     xmlns:dcterms="http://purl.org/dc/terms/"
     xmlns:media="http://search.yahoo.com/mrss/"
     xmlns:atom="http://www.w3.org/2005/Atom"
     xmlns:cf="https://www.futureplc.com/rss/content-flags"
>
    <channel>
                    <atom:link rel="alternate" hreflang="en-GB"
                       href="https://www.itpro.com/uk/feeds/tag/agentic-ai"
                       type="application/rss+xml"/>
                            <title><![CDATA[ Latest from ITPro UK in Agentic-ai ]]></title>
                <link>https://www.itpro.com/uk/artificial-intelligence/agentic-ai</link>
        <description><![CDATA[ All the latest agentic-ai content from the ITPro  UK team ]]></description>
                                    <lastBuildDate>Wed, 18 Nov 2026 09:00:00 +0000</lastBuildDate>
                            <language>en</language>
                                <item>
                                                            <title><![CDATA[ Why do AI initiatives fail – and succeed? ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Enterprise AI has moved beyond isolated pilot projects and basic conversational front-ends into a critical infrastructure challenge that directly impacts corporate architecture and operating models.</p><p>vAnd yet, even as we head into 2027, across enterprises many AI initiatives stall in the pilot phase, an indictment of IT leaders who fail to enforce architectural discipline. </p><p>Gartner <a href="https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027" target="_blank"><u>predicts</u></a> that "by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls”, more than 40% of agentic AI projects will fail. </p><p>One Gartner analyst added that "most agentic AI projects right now are early-stage experiments or proof of concepts that are mostly driven by hype and are often misapplied." </p><p>An <a href="https://executive.mit.edu/blog/developing-an-effective-ai-strategy.html" target="_blank"><u>MIT Management blog post</u></a>, meanwhile, chalked up the problems to a lack of vision, which can be laid at the feet of management. </p><p>“Many organizations dabble in pilots or isolated tools, missing opportunities to scale impact across the enterprise.”</p><p>This failure rate only tells half the story, however, and the inverse is often true. Highly <a href="https://www.itpro.com/business/business-strategy/enterprises-need-to-stop-shoehorning-ai-where-it-isnt-needed"><u>successful AI projects</u></a> can connect their wins to solid IT leadership.</p><h2 id="what-does-success-look-like">What does success look like?</h2><p>Efforts by Moderna and Jubilant Ingrevia showcase how business alignment is a hallmark of good leadership. </p><p>Moderna, the organization behind the Covid-19 vaccine, underwent a <a href="https://intuitionlabs.ai/articles/moderna-ai-adoption-case-study"><u>transformation</u></a> that “... aligned the organization around AI from the top.” The biotechnology company decided to train all employees to become data-fluent, not just their specialized workers (i.e., software engineers). </p><p>Moderna's CEO "repeatedly framed AI as essential for scaling the company’s scientific goals, not for cost cutting…" </p><p>Jubilant Ingrevia, which was <a href="https://www.mckinsey.com/capabilities/operations/how-we-help-clients/how-a-digital-operational-and-skills-transformation-took-jubilant-ingrevias-business-to-the-next-level" target="_blank"><u>struggling</u></a> to move forward, illustrates another AI-integration success story. In an effort similar to Moderna’s, the specialty chemicals company focused on training its frontline employees on IT processes and worked to foster a level of ownership over AI.</p><p>Even before beginning its AI integration efforts, management had its data scientists work onsite with plant operators for months to better understand how the business worked. </p><p>The Moderna and Jubilant Ingrevia case studies showcase how leadership decisions can directly impact the success of AI implementations. </p><p>It’s not just a technology question that boils down to choosing the "right" AI model and then seeing how it can improve things. In fact, leadership must understand core workflows and processes, then look for AI solutions that fit those circumstances.  </p><p>In short, an AI implementation failure isn't automatically due to infrastructure, model or algorithm failure. Success belongs to technology leaders who stop treating AI as a standalone novelty and start engineering it as a scalable and secure enterprise capability.</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/business-strategy/why-do-ai-initiatives-fail-and-succeed</link>
                                                                            <description>
                            <![CDATA[ IT leaders must champion AI implementation from a business perspective, not just as a technology problem to solve ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">hcrgc5odwpjvWHZ4ztjZG4</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/Su7jfDrRkaqV3sZcdmfPHc-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 18 Nov 2026 09:00:00 +0000</pubDate>                                                                                                                                                                                                                                <category><![CDATA[Business Strategy]]></category>
                                                    <category><![CDATA[Business]]></category>
                                                                                                                    <dc:creator><![CDATA[ Keith Ward ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/yayfrPRvBCXMQTMQ7V5Xw5-320-70.png ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/Su7jfDrRkaqV3sZcdmfPHc-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Male and female IT leaders discussing data strategy in an open plan office space while observing source code on a screen.]]></media:description>                                                            <media:text><![CDATA[Male and female IT leaders discussing data strategy in an open plan office space while observing source code on a screen.]]></media:text>
                                <media:title type="plain"><![CDATA[Male and female IT leaders discussing data strategy in an open plan office space while observing source code on a screen.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/Su7jfDrRkaqV3sZcdmfPHc-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Enterprise AI has moved beyond isolated pilot projects and basic conversational front-ends into a critical infrastructure challenge that directly impacts corporate architecture and operating models.</p><p>vAnd yet, even as we head into 2027, across enterprises many AI initiatives stall in the pilot phase, an indictment of IT leaders who fail to enforce architectural discipline. </p><p>Gartner <a href="https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027" target="_blank"><u>predicts</u></a> that "by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls”, more than 40% of agentic AI projects will fail. </p><p>One Gartner analyst added that "most agentic AI projects right now are early-stage experiments or proof of concepts that are mostly driven by hype and are often misapplied." </p><p>An <a href="https://executive.mit.edu/blog/developing-an-effective-ai-strategy.html" target="_blank"><u>MIT Management blog post</u></a>, meanwhile, chalked up the problems to a lack of vision, which can be laid at the feet of management. </p><p>“Many organizations dabble in pilots or isolated tools, missing opportunities to scale impact across the enterprise.”</p><p>This failure rate only tells half the story, however, and the inverse is often true. Highly <a href="https://www.itpro.com/business/business-strategy/enterprises-need-to-stop-shoehorning-ai-where-it-isnt-needed"><u>successful AI projects</u></a> can connect their wins to solid IT leadership.</p><h2 id="what-does-success-look-like">What does success look like?</h2><p>Efforts by Moderna and Jubilant Ingrevia showcase how business alignment is a hallmark of good leadership. </p><p>Moderna, the organization behind the Covid-19 vaccine, underwent a <a href="https://intuitionlabs.ai/articles/moderna-ai-adoption-case-study"><u>transformation</u></a> that “... aligned the organization around AI from the top.” The biotechnology company decided to train all employees to become data-fluent, not just their specialized workers (i.e., software engineers). </p><p>Moderna's CEO "repeatedly framed AI as essential for scaling the company’s scientific goals, not for cost cutting…" </p><p>Jubilant Ingrevia, which was <a href="https://www.mckinsey.com/capabilities/operations/how-we-help-clients/how-a-digital-operational-and-skills-transformation-took-jubilant-ingrevias-business-to-the-next-level" target="_blank"><u>struggling</u></a> to move forward, illustrates another AI-integration success story. In an effort similar to Moderna’s, the specialty chemicals company focused on training its frontline employees on IT processes and worked to foster a level of ownership over AI.</p><p>Even before beginning its AI integration efforts, management had its data scientists work onsite with plant operators for months to better understand how the business worked. </p><p>The Moderna and Jubilant Ingrevia case studies showcase how leadership decisions can directly impact the success of AI implementations. </p><p>It’s not just a technology question that boils down to choosing the "right" AI model and then seeing how it can improve things. In fact, leadership must understand core workflows and processes, then look for AI solutions that fit those circumstances.  </p><p>In short, an AI implementation failure isn't automatically due to infrastructure, model or algorithm failure. Success belongs to technology leaders who stop treating AI as a standalone novelty and start engineering it as a scalable and secure enterprise capability.</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ PagerDuty wants to supercharge incident response with a new SRE agent ]]></title>
                                                                                                <dc:content><![CDATA[ <p>PagerDuty is going through a “massive change” with a product overhaul, according to the company’s EMEA VP Arnaud Lagarde.  </p><p>The company recently unveiled Paige, a new Site Reliability Engineering (SRE) agent designed to help automate incident response capabilities for customer teams. </p><p>Running parallel to the agent roll-out, a rebrand of the PagerDuty Operations Cloud looks to create a “single source of truth” for customers on their agentic AI journeys, Lagarde told <em>ITPro</em>. </p><p>“This is a massive change in terms of the impact our customers are going to experience,” he said, noting that Paige comes in direct response to the changing face of development and IT operations. </p><p>Workers across operations, engineering, and <a href="https://www.itpro.com/software/development/how-devops-teams-can-evolve-to-meet-business-demands"><u>DevOps </u></a>teams are now scrambling to tackle bugs and various issues across IT estates, and it’s causing significant strain from an incident response perspective. </p><p>“When we talk to our customers, there are more and more events that happen in their systems. We see it in our platform, there's a big increase of events that we manage,” he told <em>ITPro</em>. </p><p>“There are more security breaches,” Lagarde added. “There’s also the fact that there’s more and more code that is being produced, and we see that code is developed more and more with agents.”</p><p>“It also leads to code that is a bit more brittle and also more fragility in the code.”</p><p>With an increasing frequency of incidents, <a href="https://www.pagerduty.com/state-of-ai-first-operations/" target="_blank"><u>research from PagerDuty</u></a> shows that enterprises are focusing heavily on incident response and cyber resilience capabilities.</p><p>95% of executives told PagerDuty that the ability to reduce incidents is now a “competitive advantage”, while 77% intend to increase resilience investment over the next 12 months. </p><h2 id="pagerduty-eyes-agentic-incident-response">PagerDuty eyes agentic incident response</h2><p>With Paige, Lagarde said PagerDuty has an opportunity to help supercharge teams on this front. Many teams are still “very reactive” when it comes to incident response, he told <em>ITPro</em>. </p><p>Acting as a “digital assistant" for teams, Paige will enable them to automate remediation processes and reduce manual toil. </p><p>“With Paige, we’re going to that new generation of how we deal with incidents with agentic AI at the forefront,” he said. “We now expect incidents to be managed by a collaboration between agents and humans.”</p><p>“It’s more modern, it's more proactive, and the benefits of having Paige at the forefront of your response also means that it helps you accelerate response.”</p><p>Paige is just one of a series of agents PagerDuty offers aimed at helping customers develop end-to-end incident response capabilities. </p><p>One specialized agent, known as Scribes, works in tandem with Paige and “reports everything, documents everything,” he explained. There’s also an Insight Agent that provides teams with details on previous incidents, enabling customers to refine best practices and build a more efficient SRE process. </p><p>While all eyes are on Paige right now, Lagarde noted that Scribes is arguably the most critical component within the company’s burgeoning toolkit. This helps develop the fundamental knowledge base that other agents draw from. </p><p>“What I say to all my customers is that the first thing you need to implement is Scribe, because when I look at systems, there’s very little knowledge that is captured by the human when they have to do it manually.”</p><p>“Therefore you don't create that knowledge base that is very important for you to be able to auto remediate. So by first building that sort of pool of knowledge using agents, that means that you don't create any additional effort on the team.”</p><h2 id="pagerduty-eyes-ecosystem-gains">PagerDuty eyes ecosystem gains</h2><p>The new Paige agent runs on top of the company’s PD Reliability Platform, formerly known as the PagerDuty Operations Cloud. </p><p>It’s here that Lagarde said the biggest changes for the company are unfolding. The rebrand and overhaul essentially aims to create a more “cohesive experience” by consolidating disparate tools and features. </p><p>“Over the years, we’ve really looked at that incident management journey and we had different components,” Lagarde explained. “Now you have a seamless customer journey across all the different stages of <a href="https://www.itpro.com/security/why-incident-response-has-become-a-core-responsibility-for-msps">incident management</a>.”</p><p>A key aspect of this push lies in the PD Reliability Platform Console. This acts as a “command center” for incident response team leaders that consolidates telemetry, workflows, and scripts within a central repository. </p><p>From an efficiency perspective, Lagarde said this offers teams significant improvements in terms of synergy and alignment. </p><p>“All these activities are now fed back into our repository, which therefore fuels productivity and continuous improvement in terms of the way that people operate and respond,” he told <em>ITPro</em>. </p><p>“They don't have to switch UI. They don't have to go, ‘okay I've got a tool that does this. I've got another tool that does that’ from different vendors,” Lagarde added. </p><p>“[It’s a] seamless experience. There is no handshake that needs to happen. One single repository, one single system of action.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/data-and-insights/this-is-a-massive-change-pagerduty-wants-to-supercharge-incident-response-with-a-new-sre-agent-and-command-center</link>
                                                                            <description>
                            <![CDATA[ The new Paige agent from PagerDuty will act as a digital assistant for teams battling incidents ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">yPQHBvB2wgGN3X6rc3mvD</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/GwqmggojqZzZYKLAfZ35AA-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 04 Nov 2026 14:52:55 +0000</pubDate>                                                                                                                                                                                                                                <category><![CDATA[Data and Insights]]></category>
                                                    <category><![CDATA[Business]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/GwqmggojqZzZYKLAfZ35AA-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[PagerDuty logo and branding pictured on a smartphone screen, with blue and orange colors blurred in background.]]></media:description>                                                            <media:text><![CDATA[PagerDuty logo and branding pictured on a smartphone screen, with blue and orange colors blurred in background.]]></media:text>
                                <media:title type="plain"><![CDATA[PagerDuty logo and branding pictured on a smartphone screen, with blue and orange colors blurred in background.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/GwqmggojqZzZYKLAfZ35AA-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>PagerDuty is going through a “massive change” with a product overhaul, according to the company’s EMEA VP Arnaud Lagarde.  </p><p>The company recently unveiled Paige, a new Site Reliability Engineering (SRE) agent designed to help automate incident response capabilities for customer teams. </p><p>Running parallel to the agent roll-out, a rebrand of the PagerDuty Operations Cloud looks to create a “single source of truth” for customers on their agentic AI journeys, Lagarde told <em>ITPro</em>. </p><p>“This is a massive change in terms of the impact our customers are going to experience,” he said, noting that Paige comes in direct response to the changing face of development and IT operations. </p><p>Workers across operations, engineering, and <a href="https://www.itpro.com/software/development/how-devops-teams-can-evolve-to-meet-business-demands"><u>DevOps </u></a>teams are now scrambling to tackle bugs and various issues across IT estates, and it’s causing significant strain from an incident response perspective. </p><p>“When we talk to our customers, there are more and more events that happen in their systems. We see it in our platform, there's a big increase of events that we manage,” he told <em>ITPro</em>. </p><p>“There are more security breaches,” Lagarde added. “There’s also the fact that there’s more and more code that is being produced, and we see that code is developed more and more with agents.”</p><p>“It also leads to code that is a bit more brittle and also more fragility in the code.”</p><p>With an increasing frequency of incidents, <a href="https://www.pagerduty.com/state-of-ai-first-operations/" target="_blank"><u>research from PagerDuty</u></a> shows that enterprises are focusing heavily on incident response and cyber resilience capabilities.</p><p>95% of executives told PagerDuty that the ability to reduce incidents is now a “competitive advantage”, while 77% intend to increase resilience investment over the next 12 months. </p><h2 id="pagerduty-eyes-agentic-incident-response">PagerDuty eyes agentic incident response</h2><p>With Paige, Lagarde said PagerDuty has an opportunity to help supercharge teams on this front. Many teams are still “very reactive” when it comes to incident response, he told <em>ITPro</em>. </p><p>Acting as a “digital assistant" for teams, Paige will enable them to automate remediation processes and reduce manual toil. </p><p>“With Paige, we’re going to that new generation of how we deal with incidents with agentic AI at the forefront,” he said. “We now expect incidents to be managed by a collaboration between agents and humans.”</p><p>“It’s more modern, it's more proactive, and the benefits of having Paige at the forefront of your response also means that it helps you accelerate response.”</p><p>Paige is just one of a series of agents PagerDuty offers aimed at helping customers develop end-to-end incident response capabilities. </p><p>One specialized agent, known as Scribes, works in tandem with Paige and “reports everything, documents everything,” he explained. There’s also an Insight Agent that provides teams with details on previous incidents, enabling customers to refine best practices and build a more efficient SRE process. </p><p>While all eyes are on Paige right now, Lagarde noted that Scribes is arguably the most critical component within the company’s burgeoning toolkit. This helps develop the fundamental knowledge base that other agents draw from. </p><p>“What I say to all my customers is that the first thing you need to implement is Scribe, because when I look at systems, there’s very little knowledge that is captured by the human when they have to do it manually.”</p><p>“Therefore you don't create that knowledge base that is very important for you to be able to auto remediate. So by first building that sort of pool of knowledge using agents, that means that you don't create any additional effort on the team.”</p><h2 id="pagerduty-eyes-ecosystem-gains">PagerDuty eyes ecosystem gains</h2><p>The new Paige agent runs on top of the company’s PD Reliability Platform, formerly known as the PagerDuty Operations Cloud. </p><p>It’s here that Lagarde said the biggest changes for the company are unfolding. The rebrand and overhaul essentially aims to create a more “cohesive experience” by consolidating disparate tools and features. </p><p>“Over the years, we’ve really looked at that incident management journey and we had different components,” Lagarde explained. “Now you have a seamless customer journey across all the different stages of <a href="https://www.itpro.com/security/why-incident-response-has-become-a-core-responsibility-for-msps">incident management</a>.”</p><p>A key aspect of this push lies in the PD Reliability Platform Console. This acts as a “command center” for incident response team leaders that consolidates telemetry, workflows, and scripts within a central repository. </p><p>From an efficiency perspective, Lagarde said this offers teams significant improvements in terms of synergy and alignment. </p><p>“All these activities are now fed back into our repository, which therefore fuels productivity and continuous improvement in terms of the way that people operate and respond,” he told <em>ITPro</em>. </p><p>“They don't have to switch UI. They don't have to go, ‘okay I've got a tool that does this. I've got another tool that does that’ from different vendors,” Lagarde added. </p><p>“[It’s a] seamless experience. There is no handshake that needs to happen. One single repository, one single system of action.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Former GitHub CEO Thomas Dohmke’s startup just unveiled a GitHub alternative ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Entire, a startup launched by former <a href="https://www.itpro.com/software/development/github-ceo-thomas-dohmke-junior-developers-generative-ai">GitHub CEO Thomas Dohmke</a>, has officially unveiled an alternative to the popular development platform. </p><p>The new Git forge is designed specifically with agent-human collaboration in mind, according to the startup, and will offer users the same basic abilities as GitHub – albeit with a twist for the AI era.</p><p>Entire claims its platform is built on “fundamentally new primitives” compared to <a href="https://www.itpro.com/open-source/31833/what-is-github">GitHub </a>and other platforms. The system is built on a new distributed source control server compatible with Git, meaning that every developer has their own dedicated server. </p><p>Using this method, the startup claims it pushes up to 89 times faster than GitHub and six times faster than Cursor’s Origin. </p><p>“This makes Entire the fastest Git forge in product today,” the startup said in an announcement. “Developers can create, migrate, and natively host private repositories in the region of their choice, including the US, EU, India, and Australia.”</p><p>Other basic functions like code reviews have received a sprucing up with the ‘Trails’ feature, according to Entire. While other platforms rely on a classic pull request model to review code changes, Entire will provide users with a “heads-up display similar to an airline pilot”. </p><p>This will include rolling, customized evals that are assessed and scored by agents with each code push, the company added. </p><p>Notably, Entire also leverages continuous memory, which it said represents a “new approach” that allows agents and developer teams to sync sessions and prompts within the repository. </p><p>The advantage here, according to Entire, is that data is constantly available and searchable for coding agents. This approach relies on two core components: a codebase graph that allows agents to “find answers faster” and reduce token consumption, and an “agentic brain”. </p><p>This essentially stores previous agent sessions and prompts within a system of memory that can be referred to or queried further down the line. </p><h2 id="entire-targets-agentic-orchestration-gains">Entire targets agentic orchestration gains</h2><p>Dohmke said these features come in direct response to the growing use of agents in software development processes. </p><p>JetBrains’ <a href="https://blog.jetbrains.com/research/2026/08/ai-coding-agent-adoption-2026/" target="_blank"><u><em>Developer Ecosystem Survey</em></u></a> found around 90% of developers now use <a href="https://www.itpro.com/software/development/github-copilot-coding-agent-launch">AI coding agents</a> at work on a weekly basis. More than two-thirds (68%), meanwhile, are using them on a daily basis. </p><p>“The Pull Request revolutionized first open source and later enterprise software development. Instead of sending diffs and patches on email lists, developers could fork a repo and send their code changes back to the maintainer,” he commented. </p><p>“But agents now write almost all of it, faster and more complex than ever before.”</p><p>Dohmke added that Entire treats the “idea, the agent trace, and the outcome of a developer’s sessions as first-class artifacts for review alongside the code”. </p><p>The platform also uses multiple agents to conduct evals to ensure developers can confidently merge code without repeated reviews. </p><p>“This allows coding agents to address findings in a loop, while the human developer validates the progress,” he explained. “When all evals are satisfied, they can merge the code without needing to look at the code itself.”</p><h2 id="dohmke-s-first-move-post-github">Dohmke’s first move post-GitHub</h2><p>The announcement marks the first big move by Entire since it <a href="https://entire.com/news/former-github-ceo-thomas-dohmke-raises-60-million-seed-round?redirected_from=entire.io" target="_blank"><u>launched earlier this year</u></a> with an eye on building the “world's next developer platform for the era of AI”. </p><p>Entire hit the ground running in February backed by a $60 million seed round at a $300 million valuation. The investment round was led by Felic, with participation from Madrona, M12, 20VC, and Cherry Ventures. </p><p>That announcement also marked Dohmke’s first venture since <a href="https://www.itpro.com/software/development/what-thomas-dohmkes-departure-means-for-github"><u>departing GitHub as CEO</u></a>. </p><p>When he announced his departure in August 2025, Dohmke specifically noted that his “startup roots have begun tugging on me”, prompting speculation about a new venture. </p><h2 id="a-viable-github-alternative">A viable GitHub alternative?</h2><p>While Entire frames itself as a viable alternative to GitHub, it faces stiff competition on this front, and not only from the aforementioned platform. </p><p>As <a href="https://www.itpro.com/software/development/vibe-coding-startup-cursor-just-launched-a-github-alternative-heres-what-users-can-expect"><u><em>ITPro </em></u><u>reported in August</u></a>, vibe coding startup Cursor launched its own alternative to tap into what has become a highly lucrative market. </p><p>When Cursor launched Origin, the company said it was “starting with the essentials”, offering users much of the same basic features found in GitHub. Long-term, the plan is to provide “agent-native features”. </p><p>Alongside the launch, Entire revealed it has secured partnerships with Buildkite and Depot to ensure build and test pipelines “keep up with how fast agents ship”. </p><p>Entire said this will allow users to move off <a href="https://www.itpro.com/software/development/github-copilot-pricing-changes-usage-based-billing-explained">GitHub Actions</a> by converting existing workflows using Buildkite’s bk pipeline or Depot CI, enabling them to subscribe the repository in its own platform. </p><p>This move by Entire mirrors Cursor, which also announced partnerships with Depot and Buildkite to build a broader “app ecosystem” aimed at tempting users away from GitHub. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/software/development/former-github-ceo-thomas-dohmkes-startup-just-unveiled-a-github-alternative-and-ai-agents-are-front-and-center</link>
                                                                            <description>
                            <![CDATA[ The new platform from Entire offers the staple features found in GitHub, but with an agentic AI twist ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">DbW4y4DruKhkJkEyGyzre7</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/pymVaMHA4kSVtP2uiYHbUP-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Mon, 26 Oct 2026 17:00:00 +0000</pubDate>                                                                                                                                                                                                                                <category><![CDATA[Development]]></category>
                                                    <category><![CDATA[Software]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/pymVaMHA4kSVtP2uiYHbUP-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Former GitHub CEO and founder of Entire Thomas Dohmke pictured during a Q&amp;A session at the VivaTech technology startup and innovation fair at the Porte de Versailles exhibition center in Paris.]]></media:description>                                                            <media:text><![CDATA[Former GitHub CEO and founder of Entire Thomas Dohmke pictured during a Q&amp;A session at the VivaTech technology startup and innovation fair at the Porte de Versailles exhibition center in Paris.]]></media:text>
                                <media:title type="plain"><![CDATA[Former GitHub CEO and founder of Entire Thomas Dohmke pictured during a Q&amp;A session at the VivaTech technology startup and innovation fair at the Porte de Versailles exhibition center in Paris.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/pymVaMHA4kSVtP2uiYHbUP-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Entire, a startup launched by former <a href="https://www.itpro.com/software/development/github-ceo-thomas-dohmke-junior-developers-generative-ai">GitHub CEO Thomas Dohmke</a>, has officially unveiled an alternative to the popular development platform. </p><p>The new Git forge is designed specifically with agent-human collaboration in mind, according to the startup, and will offer users the same basic abilities as GitHub – albeit with a twist for the AI era.</p><p>Entire claims its platform is built on “fundamentally new primitives” compared to <a href="https://www.itpro.com/open-source/31833/what-is-github">GitHub </a>and other platforms. The system is built on a new distributed source control server compatible with Git, meaning that every developer has their own dedicated server. </p><p>Using this method, the startup claims it pushes up to 89 times faster than GitHub and six times faster than Cursor’s Origin. </p><p>“This makes Entire the fastest Git forge in product today,” the startup said in an announcement. “Developers can create, migrate, and natively host private repositories in the region of their choice, including the US, EU, India, and Australia.”</p><p>Other basic functions like code reviews have received a sprucing up with the ‘Trails’ feature, according to Entire. While other platforms rely on a classic pull request model to review code changes, Entire will provide users with a “heads-up display similar to an airline pilot”. </p><p>This will include rolling, customized evals that are assessed and scored by agents with each code push, the company added. </p><p>Notably, Entire also leverages continuous memory, which it said represents a “new approach” that allows agents and developer teams to sync sessions and prompts within the repository. </p><p>The advantage here, according to Entire, is that data is constantly available and searchable for coding agents. This approach relies on two core components: a codebase graph that allows agents to “find answers faster” and reduce token consumption, and an “agentic brain”. </p><p>This essentially stores previous agent sessions and prompts within a system of memory that can be referred to or queried further down the line. </p><h2 id="entire-targets-agentic-orchestration-gains">Entire targets agentic orchestration gains</h2><p>Dohmke said these features come in direct response to the growing use of agents in software development processes. </p><p>JetBrains’ <a href="https://blog.jetbrains.com/research/2026/08/ai-coding-agent-adoption-2026/" target="_blank"><u><em>Developer Ecosystem Survey</em></u></a> found around 90% of developers now use <a href="https://www.itpro.com/software/development/github-copilot-coding-agent-launch">AI coding agents</a> at work on a weekly basis. More than two-thirds (68%), meanwhile, are using them on a daily basis. </p><p>“The Pull Request revolutionized first open source and later enterprise software development. Instead of sending diffs and patches on email lists, developers could fork a repo and send their code changes back to the maintainer,” he commented. </p><p>“But agents now write almost all of it, faster and more complex than ever before.”</p><p>Dohmke added that Entire treats the “idea, the agent trace, and the outcome of a developer’s sessions as first-class artifacts for review alongside the code”. </p><p>The platform also uses multiple agents to conduct evals to ensure developers can confidently merge code without repeated reviews. </p><p>“This allows coding agents to address findings in a loop, while the human developer validates the progress,” he explained. “When all evals are satisfied, they can merge the code without needing to look at the code itself.”</p><h2 id="dohmke-s-first-move-post-github">Dohmke’s first move post-GitHub</h2><p>The announcement marks the first big move by Entire since it <a href="https://entire.com/news/former-github-ceo-thomas-dohmke-raises-60-million-seed-round?redirected_from=entire.io" target="_blank"><u>launched earlier this year</u></a> with an eye on building the “world's next developer platform for the era of AI”. </p><p>Entire hit the ground running in February backed by a $60 million seed round at a $300 million valuation. The investment round was led by Felic, with participation from Madrona, M12, 20VC, and Cherry Ventures. </p><p>That announcement also marked Dohmke’s first venture since <a href="https://www.itpro.com/software/development/what-thomas-dohmkes-departure-means-for-github"><u>departing GitHub as CEO</u></a>. </p><p>When he announced his departure in August 2025, Dohmke specifically noted that his “startup roots have begun tugging on me”, prompting speculation about a new venture. </p><h2 id="a-viable-github-alternative">A viable GitHub alternative?</h2><p>While Entire frames itself as a viable alternative to GitHub, it faces stiff competition on this front, and not only from the aforementioned platform. </p><p>As <a href="https://www.itpro.com/software/development/vibe-coding-startup-cursor-just-launched-a-github-alternative-heres-what-users-can-expect"><u><em>ITPro </em></u><u>reported in August</u></a>, vibe coding startup Cursor launched its own alternative to tap into what has become a highly lucrative market. </p><p>When Cursor launched Origin, the company said it was “starting with the essentials”, offering users much of the same basic features found in GitHub. Long-term, the plan is to provide “agent-native features”. </p><p>Alongside the launch, Entire revealed it has secured partnerships with Buildkite and Depot to ensure build and test pipelines “keep up with how fast agents ship”. </p><p>Entire said this will allow users to move off <a href="https://www.itpro.com/software/development/github-copilot-pricing-changes-usage-based-billing-explained">GitHub Actions</a> by converting existing workflows using Buildkite’s bk pipeline or Depot CI, enabling them to subscribe the repository in its own platform. </p><p>This move by Entire mirrors Cursor, which also announced partnerships with Depot and Buildkite to build a broader “app ecosystem” aimed at tempting users away from GitHub. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ 88% of UK firms have had an 'agentic security incident' in the last year ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Agentic AI security incidents are surging at an alarming pace, according to ServiceNow, and the company has warned many firms are unprepared to deal with future incidents. </p><p>A <a href="https://www.servicenow.com/workflow/ethics/state-agentic-security.html" target="_blank"><u>poll of cybersecurity leaders </u></a>found 88% of UK firms have experienced an agent-related security incident in the past year. Crucially, only 36% of those polled said they were able to contain an incident in under 24 hours. </p><p>The findings come as agent deployments accelerate across a range of industries, the company said, from financial services and manufacturing to healthcare. </p><p>ServiceNow CISO Ben de Bont also highlighted high-profile incidents involving agents this year. </p><p>In July, OpenAI revealed that agents had escaped sandbox environments and breached third-party organizations, <a href="https://www.itpro.com/technology/artificial-intelligence/six-things-openai-learned-about-ai-from-the-hugging-face-incident">including Hugging Face</a>. The admission by the firm was followed by similar cases involving agents at <a href="https://www.itpro.com/security/anthropic-joins-openai-in-admitting-loss-of-control-in-cybersecurity-tests">Anthropic </a>and Meta. </p><p>These incidents underline the potential risks associated with the technology, de Bont said. </p><p>“The agentic incidents making headlines this year are a warning about what can happen when AI is deployed without the right controls,” he commented. </p><p>“Organizations shouldn’t have to choose between moving quickly and managing risk. The ones that get this right will build governance into how they deploy agents from the start, so they can realise the benefits of AI without losing visibility or control.”  </p><h2 id="agentic-ai-security-in-the-spotlight">Agentic AI security in the spotlight</h2><p>According to ServiceNow, security leaders face a dual challenge when it comes to agentic AI security: the scale of adoption and adapting to “machine-speed” capabilities.</p><p>The current rate of adoption is “compounding organizations’ exposure”, the company said, with more than 51% of firms now running between 100 and 999 AI agents. Just over one-third (38%) are running upwards of 1,000 agents. </p><p>Yet despite the scale of adoption, just over half (53%) of security teams have a “full, accurate record” of every agent in production. </p><p>Moreover, 24% revealed they’d identified agents running in production that had never been registered and were unknown to them. </p><p>Running parallel to this, ServiceNow warned that the machine-speed capabilities of agents are “outpacing human-speed response” capabilities. </p><p>Three-quarters of respondents said they maintain a complete map of what services and systems agents can access, yet only 31% can monitor them in real-time.  </p><p>This visibility gap means many security teams are unable to respond quickly enough in the event of a security incident or an instance where agents access unauthorized systems. </p><p>Observability has become a recurring concern since the advent of agentic AI, with enterprises unable to keep tabs on a growing array of agents operating behind the scenes. </p><p>ServiceNow’s findings on this front echo those in a <a href="https://www.itpro.com/security/ai-agents-have-a-mind-of-their-own-and-thats-a-problem-for-businesses"><u>recent study from Salt Labs</u></a>. A survey by the security firm found the rate of adoption is placing a huge strain on security teams when it comes to visibility and governance. </p><p>Many enterprises also lack the appropriate containment controls for agents, Salt Labs warned. Indeed, the study found nearly half of respondents had observed an agent performing an action that wasn’t intended or authorized. </p><h2 id="assessing-risk">Assessing risk</h2><p>ServiceNow noted that the use of agents will require a step change in how security leaders assess risk – and so far these efforts are falling flat. </p><p>The company said that while many security leaders “self-assess as having established agentic security postures”, just 13% have clear cut programs designed to measure leading considerations.</p><p>These key considerations span areas such as attack surfaces, accountability, visibility, and audit readiness. </p><p>According to ServiceNow, this shows that agentic AI security is primarily governed by “legacy frameworks and human-driven systems” that are unfit. Just over half (51%) of UK firms said they have governance frameworks that cover the aforementioned considerations. </p><p>Similarly, fewer than half (48%) of security teams build “design security” within the agent development process. The study warned this means that various permissions and guardrails are being implemented on an “ad-hoc, after-the-fact basis”.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/security/88-percent-of-uk-firms-have-had-an-agentic-security-incident-in-the-last-year-servicenow-warns-more-will-come</link>
                                                                            <description>
                            <![CDATA[ Concerns are growing over the scale of adoption and an inability to match “machine-speed” capabilities ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">Xi4mGzUjEKvBSdmyZu2Go7</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/jfFdbA7u7TTjdg83tgTdUT-1920-80.png" type="image/png" length="0"></enclosure>
                                                                        <pubDate>Fri, 23 Oct 2026 10:57:45 +0000</pubDate>                                                                                                                                                                                                                                <category><![CDATA[Security]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/png" url="https://cdn.mos.cms.futurecdn.net/jfFdbA7u7TTjdg83tgTdUT-1920-80.png">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[AI robot with circuit, chemical structure and program code on a black background]]></media:description>                                                            <media:text><![CDATA[AI robot with circuit, chemical structure and program code on a black background]]></media:text>
                                <media:title type="plain"><![CDATA[AI robot with circuit, chemical structure and program code on a black background]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/jfFdbA7u7TTjdg83tgTdUT-1920-80.png" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Agentic AI security incidents are surging at an alarming pace, according to ServiceNow, and the company has warned many firms are unprepared to deal with future incidents. </p><p>A <a href="https://www.servicenow.com/workflow/ethics/state-agentic-security.html" target="_blank"><u>poll of cybersecurity leaders </u></a>found 88% of UK firms have experienced an agent-related security incident in the past year. Crucially, only 36% of those polled said they were able to contain an incident in under 24 hours. </p><p>The findings come as agent deployments accelerate across a range of industries, the company said, from financial services and manufacturing to healthcare. </p><p>ServiceNow CISO Ben de Bont also highlighted high-profile incidents involving agents this year. </p><p>In July, OpenAI revealed that agents had escaped sandbox environments and breached third-party organizations, <a href="https://www.itpro.com/technology/artificial-intelligence/six-things-openai-learned-about-ai-from-the-hugging-face-incident">including Hugging Face</a>. The admission by the firm was followed by similar cases involving agents at <a href="https://www.itpro.com/security/anthropic-joins-openai-in-admitting-loss-of-control-in-cybersecurity-tests">Anthropic </a>and Meta. </p><p>These incidents underline the potential risks associated with the technology, de Bont said. </p><p>“The agentic incidents making headlines this year are a warning about what can happen when AI is deployed without the right controls,” he commented. </p><p>“Organizations shouldn’t have to choose between moving quickly and managing risk. The ones that get this right will build governance into how they deploy agents from the start, so they can realise the benefits of AI without losing visibility or control.”  </p><h2 id="agentic-ai-security-in-the-spotlight">Agentic AI security in the spotlight</h2><p>According to ServiceNow, security leaders face a dual challenge when it comes to agentic AI security: the scale of adoption and adapting to “machine-speed” capabilities.</p><p>The current rate of adoption is “compounding organizations’ exposure”, the company said, with more than 51% of firms now running between 100 and 999 AI agents. Just over one-third (38%) are running upwards of 1,000 agents. </p><p>Yet despite the scale of adoption, just over half (53%) of security teams have a “full, accurate record” of every agent in production. </p><p>Moreover, 24% revealed they’d identified agents running in production that had never been registered and were unknown to them. </p><p>Running parallel to this, ServiceNow warned that the machine-speed capabilities of agents are “outpacing human-speed response” capabilities. </p><p>Three-quarters of respondents said they maintain a complete map of what services and systems agents can access, yet only 31% can monitor them in real-time.  </p><p>This visibility gap means many security teams are unable to respond quickly enough in the event of a security incident or an instance where agents access unauthorized systems. </p><p>Observability has become a recurring concern since the advent of agentic AI, with enterprises unable to keep tabs on a growing array of agents operating behind the scenes. </p><p>ServiceNow’s findings on this front echo those in a <a href="https://www.itpro.com/security/ai-agents-have-a-mind-of-their-own-and-thats-a-problem-for-businesses"><u>recent study from Salt Labs</u></a>. A survey by the security firm found the rate of adoption is placing a huge strain on security teams when it comes to visibility and governance. </p><p>Many enterprises also lack the appropriate containment controls for agents, Salt Labs warned. Indeed, the study found nearly half of respondents had observed an agent performing an action that wasn’t intended or authorized. </p><h2 id="assessing-risk">Assessing risk</h2><p>ServiceNow noted that the use of agents will require a step change in how security leaders assess risk – and so far these efforts are falling flat. </p><p>The company said that while many security leaders “self-assess as having established agentic security postures”, just 13% have clear cut programs designed to measure leading considerations.</p><p>These key considerations span areas such as attack surfaces, accountability, visibility, and audit readiness. </p><p>According to ServiceNow, this shows that agentic AI security is primarily governed by “legacy frameworks and human-driven systems” that are unfit. Just over half (51%) of UK firms said they have governance frameworks that cover the aforementioned considerations. </p><p>Similarly, fewer than half (48%) of security teams build “design security” within the agent development process. The study warned this means that various permissions and guardrails are being implemented on an “ad-hoc, after-the-fact basis”.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ AI agents have a mind of their own – and that’s a problem for businesses ]]></title>
                                                                                                <dc:content><![CDATA[ <p>AI agents are going rogue on a massive scale, with half of enterprises reporting that they've had to deal with at least one taking unauthorized actions over the last year.</p><p>In a <a href="https://edge.prnewswire.com/c/link/?t=0&amp;l=en&amp;o=4795737-1&amp;h=4261799294&amp;u=https%3A%2F%2Fsalt.security%2Fapi-security-trends&amp;a=https%3A%2F%2Fsalt.security%2Fapi-security-trends"><u>report</u></a> based on a survey of more than 380 security and technology professionals, along with production telemetry from hundreds of customer environments, Salt Labs claimed agentic AI has moved into daily operations too quickly for visibility, governance, and containment controls.</p><p>Half of survey respondents (49.8%) said they had confirmed or suspected that an AI agent in their organization took an action in the past 12 months that wasn't intended, expected, or authorized. </p><p>Of these, more than half reported an unauthorized external API call or transaction, while more than four-in-ten reported that an agent modified or deleted data.</p><p>"The data shows a gap that is already producing real operational consequences," said Roey Eliyahu, CEO of Salt Security. </p><p>"Organizations are deploying agents and granting them access to enterprise data, APIs, and services before they can trace what those agents are doing or stop them automatically when something goes wrong. Closing that gap is the discipline of agentic security, and it is what lets enterprises give AI agents the autonomy they need without inheriting unmanaged risk."</p><h2 id="enthusiastic-deployment-meets-lack-of-control">Enthusiastic deployment meets lack of control</h2><p>The vast majority of organizations surveyed (86%) are using at least one AI agent, the survey found while 60% run more than ten. Another 77% of respondents said their agents chain actions across other agents, tools, and APIs, in sequences no single team designed in advance.</p><p>Only 12.5% said they can consistently trace a complete agent path across all agents, from initial instruction through tools and APIs to outcome. More than six-in-ten believe they would need hours or days of manual work to reconstruct what a specific agent did the previous day.</p><p>When problems do occur, they're hard to deal with. Only 17.5% can automatically stop an agent that begins behaving unexpectedly in production. Most can intervene only manually, or by taking the application offline.</p><p>More than eight-in-ten said AI-generated or AI-assisted code contributes to at least some of their application security findings. Only 7% resolve issues inside the coding assistant at generation time, rather than through code review, testing, or after deployment.</p><p>Meanwhile, shadow model context protocol (MCP) infrastructure is already a live operational problem. 33% of survey respondents said they had found an unauthorized, shadow, or misconfigured MCP server in the past 12 months. 54% said they'd confirmed or suspected one. </p><p>Salt Labs' own data shows MCP-related infrastructure in 22% of customer environments in August this year, up from 16.7% in January.</p><p>Earlier this year, a survey commissioned by Aembit <a href="https://www.itpro.com/technology/artificial-intelligence/workers-cant-identify-work-produced-by-ai-agents-business-risks"><u>found</u></a> that 68% of organizations can’t accurately identify AI agent activity compared to human activity.</p><p>Nearly three-quarters said that agents often receive more access than necessary, and 79% believe they create new access pathways that are difficult to monitor. Just over half said agents, at least occasionally, inherit access originally intended for humans or other systems.</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/security/ai-agents-have-a-mind-of-their-own-and-thats-a-problem-for-businesses</link>
                                                                            <description>
                            <![CDATA[ New research shows that half of organizations are having to deal with agent behaviour that hasn't been authorized or approved. ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">hMqwdoc2fXxL3xxkXkb7Hh</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/45XM7VDiKtHZtv8qEvuvW5-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 14 Oct 2026 10:53:10 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 14:44:06 +0000</updated>
                                                                                                                                            <category><![CDATA[Security]]></category>
                                                                                                                    <dc:creator><![CDATA[ Emma Woollacott ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/aWfskavxoVSMDy6cDWtYmJ-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/45XM7VDiKtHZtv8qEvuvW5-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[A visualization of agentic AI layers, shown as blue nodes of code connected by blue strands against a dark background.]]></media:description>                                                            <media:text><![CDATA[A visualization of agentic AI layers, shown as blue nodes of code connected by blue strands against a dark background.]]></media:text>
                                <media:title type="plain"><![CDATA[A visualization of agentic AI layers, shown as blue nodes of code connected by blue strands against a dark background.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/45XM7VDiKtHZtv8qEvuvW5-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>AI agents are going rogue on a massive scale, with half of enterprises reporting that they've had to deal with at least one taking unauthorized actions over the last year.</p><p>In a <a href="https://edge.prnewswire.com/c/link/?t=0&amp;l=en&amp;o=4795737-1&amp;h=4261799294&amp;u=https%3A%2F%2Fsalt.security%2Fapi-security-trends&amp;a=https%3A%2F%2Fsalt.security%2Fapi-security-trends"><u>report</u></a> based on a survey of more than 380 security and technology professionals, along with production telemetry from hundreds of customer environments, Salt Labs claimed agentic AI has moved into daily operations too quickly for visibility, governance, and containment controls.</p><p>Half of survey respondents (49.8%) said they had confirmed or suspected that an AI agent in their organization took an action in the past 12 months that wasn't intended, expected, or authorized. </p><p>Of these, more than half reported an unauthorized external API call or transaction, while more than four-in-ten reported that an agent modified or deleted data.</p><p>"The data shows a gap that is already producing real operational consequences," said Roey Eliyahu, CEO of Salt Security. </p><p>"Organizations are deploying agents and granting them access to enterprise data, APIs, and services before they can trace what those agents are doing or stop them automatically when something goes wrong. Closing that gap is the discipline of agentic security, and it is what lets enterprises give AI agents the autonomy they need without inheriting unmanaged risk."</p><h2 id="enthusiastic-deployment-meets-lack-of-control">Enthusiastic deployment meets lack of control</h2><p>The vast majority of organizations surveyed (86%) are using at least one AI agent, the survey found while 60% run more than ten. Another 77% of respondents said their agents chain actions across other agents, tools, and APIs, in sequences no single team designed in advance.</p><p>Only 12.5% said they can consistently trace a complete agent path across all agents, from initial instruction through tools and APIs to outcome. More than six-in-ten believe they would need hours or days of manual work to reconstruct what a specific agent did the previous day.</p><p>When problems do occur, they're hard to deal with. Only 17.5% can automatically stop an agent that begins behaving unexpectedly in production. Most can intervene only manually, or by taking the application offline.</p><p>More than eight-in-ten said AI-generated or AI-assisted code contributes to at least some of their application security findings. Only 7% resolve issues inside the coding assistant at generation time, rather than through code review, testing, or after deployment.</p><p>Meanwhile, shadow model context protocol (MCP) infrastructure is already a live operational problem. 33% of survey respondents said they had found an unauthorized, shadow, or misconfigured MCP server in the past 12 months. 54% said they'd confirmed or suspected one. </p><p>Salt Labs' own data shows MCP-related infrastructure in 22% of customer environments in August this year, up from 16.7% in January.</p><p>Earlier this year, a survey commissioned by Aembit <a href="https://www.itpro.com/technology/artificial-intelligence/workers-cant-identify-work-produced-by-ai-agents-business-risks"><u>found</u></a> that 68% of organizations can’t accurately identify AI agent activity compared to human activity.</p><p>Nearly three-quarters said that agents often receive more access than necessary, and 79% believe they create new access pathways that are difficult to monitor. Just over half said agents, at least occasionally, inherit access originally intended for humans or other systems.</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Google cyber researchers warn hackers are ramping up automated attacks ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Attackers used AI agents to compromise a cloud resource and then plan, build, and carry out a mass credential harvesting campaign earlier this year – and all in less than six hours.</p><p>That’s according to new <a href="https://cloud.google.com/blog/topics/threat-intelligence/from-prompting-to-autonomy-the-evolution-of-adversarial-ai" target="_blank"><u>research </u></a>from Google Threat Intelligence Group (GTIG), with recent analysis showing threat actors are flocking to agentic AI and automation to conduct sophisticated attacks. </p><p>“At this point, we can assume that all threat actors are using AI in some capacity and their operations have benefited," said John Hultquist, chief analyst at Google Threat Intelligence Group. </p><p>"Like everyone else, we’re concerned about the vulnerability problem, but AI is being applied to several other areas, and it will be especially challenging as it is applied agentically, creating a scaled, faster adversary. Criminals, like the ones who conducted a mass exploitation campaign in just six hours, will gravitate to attacks that are faster than we can respond to."</p><p>In this example, a financially motivated threat actor compromised an organization’s cloud infrastructure to deploy an autonomous, multi-agent attack framework, allowing them to operate at the sort of scale and speed that's usually associated with larger and more resource-heavy groups. </p><p>In less than six hours, the attacker was able to leverage an AI coding chatbot, a simple prompt, and a set of agent instructions to plan, build, and execute a mass credential harvesting campaign. </p><p>Using preconfigured markdown instruction sets as operational playbooks, GTIG said the threat actor carried out automated scanning and credential harvesting, compromising thousands of third-party credentials. </p><p>The agent instructions enabled the AI to autonomously manage the vulnerability scanning pipeline, perform real-time troubleshooting, and execute IP rotation logic without manual intervention. </p><p>Operating from the victim cloud infrastructure also allowed the threat actor route attack traffic through legitimate IP addresses.</p><h2 id="state-backed-groups-are-flocking-to-ai">State-backed groups are flocking to AI</h2><p>GTIG said it has observed state-linked groups broadening their use of AI. In one example, PRC-nexus threat actor UNC6508 carried out a major intrusion campaign against US medical facilities, using compromised cloud environments to host a local model.</p><p>In April, meanwhile, Mandiant spotted a threat actor leveraging AI infrastructure access which was then used to sustain unauthorized AI workloads, utilizing hijacked cloud environments to provision high-performance GPU compute instances at the victim's expense.</p><p>Notably, <a href="https://www.itpro.com/security/litellm-pypi-compromise-everything-we-know-so-far">TeamPCP </a>has focused heavily on the AI supply chain, using malware designed to take advantage of emerging AI systems, and including malicious prompts that were used to execute commands surreptitiously. </p><p>The group has even poisoned open source package metadata to trick AI assistants, resulting in malicious dependencies being recommended to developers. </p><p>"The challenge for defenders is that these risks are evolving faster than many organizations can quantify, measure, and govern," said Ronald Lewis, head of <a href="https://www.itpro.com/security/28133/what-is-cyber-security">cybersecurity </a>governance at Black Duck. </p><p>"Security teams are no longer protecting only applications, users, and infrastructure. They must now secure <a href="https://www.itpro.com/technology/artificial-intelligence/the-risks-of-open-source-ai-models">AI models</a>, agents, prompts, data pipelines, and an increasingly complex AI supply chain while also defending against adversaries using AI to accelerate attacks.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/security/cyber-crime/we-can-assume-that-all-threat-actors-are-using-ai-in-some-capacity-google-cyber-researchers-warn-hackers-are-ramping-up-automated-attacks</link>
                                                                            <description>
                            <![CDATA[ Google Threat Intelligence Group has issued a warning over the increased threats posed by hackers using agentic AI tools ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">sWhJ4d98MpUEhSrqv8PtWU</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/rcgqGm2k9qbr9K4kHhEmvU-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 09 Sep 2026 17:00:00 +0000</pubDate>                                                                                                                                <updated>Thu, 10 Sep 2026 11:28:05 +0000</updated>
                                                                                                                                            <category><![CDATA[Cyber Crime]]></category>
                                                    <category><![CDATA[Security]]></category>
                                                                                                                    <dc:creator><![CDATA[ Emma Woollacott ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/aWfskavxoVSMDy6cDWtYmJ-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/rcgqGm2k9qbr9K4kHhEmvU-1920-80.jpg">
                                                            <media:credit><![CDATA[Shutterstock]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[An abstract image showing a skull over a pixelated background to symbolise a cyber security vulnerability]]></media:description>                                                            <media:text><![CDATA[An abstract image showing a skull over a pixelated background to symbolise a cyber security vulnerability]]></media:text>
                                <media:title type="plain"><![CDATA[An abstract image showing a skull over a pixelated background to symbolise a cyber security vulnerability]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/rcgqGm2k9qbr9K4kHhEmvU-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Attackers used AI agents to compromise a cloud resource and then plan, build, and carry out a mass credential harvesting campaign earlier this year – and all in less than six hours.</p><p>That’s according to new <a href="https://cloud.google.com/blog/topics/threat-intelligence/from-prompting-to-autonomy-the-evolution-of-adversarial-ai" target="_blank"><u>research </u></a>from Google Threat Intelligence Group (GTIG), with recent analysis showing threat actors are flocking to agentic AI and automation to conduct sophisticated attacks. </p><p>“At this point, we can assume that all threat actors are using AI in some capacity and their operations have benefited," said John Hultquist, chief analyst at Google Threat Intelligence Group. </p><p>"Like everyone else, we’re concerned about the vulnerability problem, but AI is being applied to several other areas, and it will be especially challenging as it is applied agentically, creating a scaled, faster adversary. Criminals, like the ones who conducted a mass exploitation campaign in just six hours, will gravitate to attacks that are faster than we can respond to."</p><p>In this example, a financially motivated threat actor compromised an organization’s cloud infrastructure to deploy an autonomous, multi-agent attack framework, allowing them to operate at the sort of scale and speed that's usually associated with larger and more resource-heavy groups. </p><p>In less than six hours, the attacker was able to leverage an AI coding chatbot, a simple prompt, and a set of agent instructions to plan, build, and execute a mass credential harvesting campaign. </p><p>Using preconfigured markdown instruction sets as operational playbooks, GTIG said the threat actor carried out automated scanning and credential harvesting, compromising thousands of third-party credentials. </p><p>The agent instructions enabled the AI to autonomously manage the vulnerability scanning pipeline, perform real-time troubleshooting, and execute IP rotation logic without manual intervention. </p><p>Operating from the victim cloud infrastructure also allowed the threat actor route attack traffic through legitimate IP addresses.</p><h2 id="state-backed-groups-are-flocking-to-ai">State-backed groups are flocking to AI</h2><p>GTIG said it has observed state-linked groups broadening their use of AI. In one example, PRC-nexus threat actor UNC6508 carried out a major intrusion campaign against US medical facilities, using compromised cloud environments to host a local model.</p><p>In April, meanwhile, Mandiant spotted a threat actor leveraging AI infrastructure access which was then used to sustain unauthorized AI workloads, utilizing hijacked cloud environments to provision high-performance GPU compute instances at the victim's expense.</p><p>Notably, <a href="https://www.itpro.com/security/litellm-pypi-compromise-everything-we-know-so-far">TeamPCP </a>has focused heavily on the AI supply chain, using malware designed to take advantage of emerging AI systems, and including malicious prompts that were used to execute commands surreptitiously. </p><p>The group has even poisoned open source package metadata to trick AI assistants, resulting in malicious dependencies being recommended to developers. </p><p>"The challenge for defenders is that these risks are evolving faster than many organizations can quantify, measure, and govern," said Ronald Lewis, head of <a href="https://www.itpro.com/security/28133/what-is-cyber-security">cybersecurity </a>governance at Black Duck. </p><p>"Security teams are no longer protecting only applications, users, and infrastructure. They must now secure <a href="https://www.itpro.com/technology/artificial-intelligence/the-risks-of-open-source-ai-models">AI models</a>, agents, prompts, data pipelines, and an increasingly complex AI supply chain while also defending against adversaries using AI to accelerate attacks.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Three things you need to know about Claudeforce ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Salesforce and Anthropic have unveiled ‘Claudeforce’, marking a significant expansion of the existing partnership between the two firms. </p><p>The move will see deeper integration of <a href="https://www.itpro.com/business/want-to-get-the-most-out-of-anthropics-claude-ai-assistant-this-new-training-course-will-give-you-prompt-engineering-tips-and-how-to-use-claude-code">Anthropic’s Claude AI</a> across the breadth of the Salesforce ecosystem, spanning Data 360, Tableau, and Slack. </p><p>Notably, the two firms have been forging closer ties for some time now. The CRM giant noted that they are both “strategic customers of one another”. </p><p>Salesforce, for example, is Anthropic’s preferred CRM while Slack plays a key role in workforce communication, workflows, and agentic orchestration. On the flip side, Claude already underpins key capabilities within Slack, with users having access to various Claude-powered agents. </p><p>The Claudeforce announcement essentially reinforces this relationship and will drive deeper synergy between the two firms. Salesforce CEO Marc Benioff described the move as “fusing” Claude with the CRM giant’s data and workflows. </p><p>“We’re launching Claudeforce, running directly on top of Salesforce via our new AIforce UI harness, Headless 360, Data 360, Tableau, and Slack. Here, the UI is the AI — allowing you to build custom apps dynamically and answer any enterprise question,” he said in a blog post. </p><p>“We’re delivering a dynamic interface that thinks, reasons, and acts.”</p><p>Simply put, Claude will be woven directly into Salesforce’s various domains, and Salesforce will be woven directly within Claude. It’s the “best of both worlds”, according to Benioff. </p><p>So what can users expect?</p><h2 id="salesforce-inside-claude">Salesforce inside Claude</h2><p>The first big talking point with Claudeforce centers around Salesforce in Claude, a plugin that comes equipped with 37 pre-built sales skills. We’ve already seen Anthropic make moves with domain-specific plugins in recent months. </p><p><a href="https://www.itpro.com/technology/artificial-intelligence/everything-you-need-to-know-about-anthropic-claude-cowork">Claude Cowork</a>, which launched in January this year, came equipped with a range of industry-specific plug-ins, which as <em>ITPro </em>reported at the time <a href="https://www.itpro.com/technology/artificial-intelligence/why-anthropic-sent-software-stocks-into-freefall">sparked somewhat of a frenzy in the SaaS space</a>. </p><p>According to Salesforce, because these plug-ins have deep access to customer data and business rules, they better “understand” seller revenue cycles – providing a helping hand to sales personnel. </p><p>“These skills were built by Salesforce and Anthropic, engineered specifically to draw on Claude’s reasoning, agentic tool use, and generative UI — not generic CRM prompts wrapped around an API but the capabilities that extend what Claude can do,” the company said in a blog post. </p><p>These plug-ins even have an “onboarding experience” that will see Claude analyze seller data across Salesforce and Slack to provide better context. </p><p>“No two companies run sales exactly the same, so Salesforce in Claude routes actions through Salesforce to help ensure business rules are always enforced when an action is taken. Setup is seamless,” the company claimed. </p><h2 id="claude-inside-salesforce">Claude inside Salesforce</h2><p>This is a relationship that works both ways for Salesforce and Anthropic. Claude is already available within Agentforce, acting as the reasoning model for the company’s Atlas Reasoning Engine. </p><p>The LLM also powers Agentforce Vibes and Agentforce Coworker by default, and is available in Agent Builder. </p><p>With the extended partnership, the duo are eyeing deeper integration of Claude across the Salesforce estate. Notably, the LLM is now fully integrated within the Salesforce Trust Boundary through <a href="https://www.itpro.com/technology/artificial-intelligence/aws-goes-all-in-on-ai-agents-with-new-features-for-bedrock-and-amazon-q">Amazon Bedrock</a>.  </p><h2 id="slack-integration">Slack integration</h2><p>Slack will play a key role in Claudeforce moving forward. The workplace collaboration platform has already been transformed into an ‘agentic OS’, and as ITPro reported last year, is a <a href="https://www.itpro.com/software/slack-is-now-the-key-to-salesforces-agentic-ai-plans"><u>core pillar of the Agentforce 360 platform</u></a>. </p><p>Claude acts as the default model for Slack, powering Slackbot and helping support users via features such as Slack Code and <a href="https://www.itpro.com/software/meet-claude-tag-anthropics-new-ai-teammate-that-works-in-slack"><u>Claude Tag</u></a>. </p><p>Claude Tag was rolled out earlier this year and replaced the existing Claude in Slack application. The tool essentially acts as a virtual teammate within Slack, allowing users to @ the agent within channels and allocate tasks. </p><p>Salesforce said that by embedding Claude directly within Slack, it’s “bridging the gap between team dialogue and autonomous enterprise action”. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/data-and-insights/were-delivering-a-dynamic-interface-that-thinks-reasons-and-acts-three-things-you-need-to-know-about-claudeforce</link>
                                                                            <description>
                            <![CDATA[ Anthropic and Salesforce are forging closer ties with deeper integration of Claude across core product ranges ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">zgnJh8JcupREAvcUozNS9Z</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/KA6SmPnHhdc3Cp7i4KVzw7-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Sun, 30 Aug 2026 23:05:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:23 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Data and Insights]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                    <category><![CDATA[Business]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/KA6SmPnHhdc3Cp7i4KVzw7-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Salesforce logo and branding illuminated on a sign at the 8th China International Import Expo.]]></media:description>                                                            <media:text><![CDATA[Salesforce logo and branding illuminated on a sign at the 8th China International Import Expo.]]></media:text>
                                <media:title type="plain"><![CDATA[Salesforce logo and branding illuminated on a sign at the 8th China International Import Expo.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/KA6SmPnHhdc3Cp7i4KVzw7-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Salesforce and Anthropic have unveiled ‘Claudeforce’, marking a significant expansion of the existing partnership between the two firms. </p><p>The move will see deeper integration of <a href="https://www.itpro.com/business/want-to-get-the-most-out-of-anthropics-claude-ai-assistant-this-new-training-course-will-give-you-prompt-engineering-tips-and-how-to-use-claude-code">Anthropic’s Claude AI</a> across the breadth of the Salesforce ecosystem, spanning Data 360, Tableau, and Slack. </p><p>Notably, the two firms have been forging closer ties for some time now. The CRM giant noted that they are both “strategic customers of one another”. </p><p>Salesforce, for example, is Anthropic’s preferred CRM while Slack plays a key role in workforce communication, workflows, and agentic orchestration. On the flip side, Claude already underpins key capabilities within Slack, with users having access to various Claude-powered agents. </p><p>The Claudeforce announcement essentially reinforces this relationship and will drive deeper synergy between the two firms. Salesforce CEO Marc Benioff described the move as “fusing” Claude with the CRM giant’s data and workflows. </p><p>“We’re launching Claudeforce, running directly on top of Salesforce via our new AIforce UI harness, Headless 360, Data 360, Tableau, and Slack. Here, the UI is the AI — allowing you to build custom apps dynamically and answer any enterprise question,” he said in a blog post. </p><p>“We’re delivering a dynamic interface that thinks, reasons, and acts.”</p><p>Simply put, Claude will be woven directly into Salesforce’s various domains, and Salesforce will be woven directly within Claude. It’s the “best of both worlds”, according to Benioff. </p><p>So what can users expect?</p><h2 id="salesforce-inside-claude">Salesforce inside Claude</h2><p>The first big talking point with Claudeforce centers around Salesforce in Claude, a plugin that comes equipped with 37 pre-built sales skills. We’ve already seen Anthropic make moves with domain-specific plugins in recent months. </p><p><a href="https://www.itpro.com/technology/artificial-intelligence/everything-you-need-to-know-about-anthropic-claude-cowork">Claude Cowork</a>, which launched in January this year, came equipped with a range of industry-specific plug-ins, which as <em>ITPro </em>reported at the time <a href="https://www.itpro.com/technology/artificial-intelligence/why-anthropic-sent-software-stocks-into-freefall">sparked somewhat of a frenzy in the SaaS space</a>. </p><p>According to Salesforce, because these plug-ins have deep access to customer data and business rules, they better “understand” seller revenue cycles – providing a helping hand to sales personnel. </p><p>“These skills were built by Salesforce and Anthropic, engineered specifically to draw on Claude’s reasoning, agentic tool use, and generative UI — not generic CRM prompts wrapped around an API but the capabilities that extend what Claude can do,” the company said in a blog post. </p><p>These plug-ins even have an “onboarding experience” that will see Claude analyze seller data across Salesforce and Slack to provide better context. </p><p>“No two companies run sales exactly the same, so Salesforce in Claude routes actions through Salesforce to help ensure business rules are always enforced when an action is taken. Setup is seamless,” the company claimed. </p><h2 id="claude-inside-salesforce">Claude inside Salesforce</h2><p>This is a relationship that works both ways for Salesforce and Anthropic. Claude is already available within Agentforce, acting as the reasoning model for the company’s Atlas Reasoning Engine. </p><p>The LLM also powers Agentforce Vibes and Agentforce Coworker by default, and is available in Agent Builder. </p><p>With the extended partnership, the duo are eyeing deeper integration of Claude across the Salesforce estate. Notably, the LLM is now fully integrated within the Salesforce Trust Boundary through <a href="https://www.itpro.com/technology/artificial-intelligence/aws-goes-all-in-on-ai-agents-with-new-features-for-bedrock-and-amazon-q">Amazon Bedrock</a>.  </p><h2 id="slack-integration">Slack integration</h2><p>Slack will play a key role in Claudeforce moving forward. The workplace collaboration platform has already been transformed into an ‘agentic OS’, and as ITPro reported last year, is a <a href="https://www.itpro.com/software/slack-is-now-the-key-to-salesforces-agentic-ai-plans"><u>core pillar of the Agentforce 360 platform</u></a>. </p><p>Claude acts as the default model for Slack, powering Slackbot and helping support users via features such as Slack Code and <a href="https://www.itpro.com/software/meet-claude-tag-anthropics-new-ai-teammate-that-works-in-slack"><u>Claude Tag</u></a>. </p><p>Claude Tag was rolled out earlier this year and replaced the existing Claude in Slack application. The tool essentially acts as a virtual teammate within Slack, allowing users to @ the agent within channels and allocate tasks. </p><p>Salesforce said that by embedding Claude directly within Slack, it’s “bridging the gap between team dialogue and autonomous enterprise action”. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ McKinsey says AI is finally delivering results for enterprises – but there's a catch ]]></title>
                                                                                                <dc:content><![CDATA[ <p>New research from McKinsey suggests enterprises are finally reporting benefits from AI, but cost-related concerns and “constrained” productivity gains still weigh heavy on the minds of IT leaders.</p><p>Figures from the consultancy’s <a href="https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai" target="_blank"><u><em>State of AI in 2026</em></u></a> report show more than one-third (37%) of respondents recorded “some” earnings impact with the technology, although that marks the same share from last year’s report. </p><p>Meanwhile, “AI high performers” said AI is now having a “significant” impact on operational efficiency and earnings. </p><p>The report suggests that enterprise AI maturity is rising, with an increasing number of enterprises now successfully scaling the technology. Improvements in this regard are being reported across the board, McKinsey noted, with respondents scaling chatbots, coding agents, and more sophisticated AI agents. </p><p>The use of agentic AI has increased significantly, for example, with 40% of large enterprises scaling agents, up from 27% last year. Meanwhile, around two-in-ten are actively scaling software coding agents. </p><p>“Among <a href="https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today">AI tools</a>, chatbots are the most widely scaled, with 47 percent of respondents saying their organizations are scaling them across the enterprise,” the report states. </p><p>“About two-in-ten respondents report reaching the scaling phase across their organization with AI agents and a similar share report the same with software coding agents.”</p><h2 id="mismatched-ai-productivity">Mismatched AI productivity</h2><p>Workforce productivity gains rank among the most notable improvements for enterprises, according to McKinsey. </p><p>Around 80% of respondents said that AI has improved individual productivity, while 50% report that AI helps them “make better decisions”. Crucially, however, McKinsey warned that “the experience is not universally positive”. </p><p>Mid-level managers and individual workers are more likely than executives to report AI-related problems. McKinsey said this shows that AI impact “remains concentrated” among certain groups within the enterprise. </p><p>This is by no means the first study to highlight this trend. Research from Accenture in April this year found <a href="https://www.itpro.com/business/business-strategy/ai-productivity-challenges-accenture-generating-impact-study"><u>many enterprises are dealing with mismatched AI productivity gains</u></a>. </p><p>Similar to McKinsey, the report found that workers are reporting benefits with the technology such as higher-quality output levels and faster project delivery times. </p><p>Yet the report pointed to a “widening gap” between the basic use of the technology and its wider enterprise impact. A key factor, Accenture said, lies in the fact that organizations' process changes are “lagging behind” integration rates. </p><p>Similarly, many AI strategies typically fail to modernize underlying IT infrastructure or workflows to accommodate the technology. </p><p>“A striking finding this year is the gap between individual gains and enterprise impact. At the individual level, AI is clearly a boon: 80 percent of survey respondents say it has improved their productivity and half say it helps them make better decisions,” said Dan Tinkoff, senior partner at McKinsey. </p><p>“Yet, only 37 percent of organizations report any positive EBIT contribution, essentially flat compared with last year.”</p><h2 id="ai-costs-are-taking-the-shine-off-enterprise-gains">AI costs are taking the shine off enterprise gains</h2><p>Cost-related concerns continue to weigh heavy on IT leaders’ minds, according to McKinsey. Around 20% of respondents said that AI-related operating costs have “constrained their AI use”. </p><p>“Those cost constraints are being reported across the full range of AI tools. For each of three tools – <a href="https://www.itpro.com/technology/artificial-intelligence-ai/369979/chatgpt-vs-chatbots-whats-the-difference">AI chatbots</a>, AI agents, and software <a href="https://www.itpro.com/software/development/while-the-engineers-slept-the-agents-kept-building-aws-uk-chief-touts-big-gains-with-ai-powered-coding">coding agents</a> – about one-in-ten respondents say their organizations’ use has been constrained by costs,” the report states. </p><p>McKinsey also highlighted <a href="https://www.itpro.com/technology/artificial-intelligence/it-leaders-are-being-stung-by-unexpected-ai-costs">token-related cost concerns</a>. This has become a recurring pain point for some businesses over the last year as firms ramp up adoption of the technology. </p><p>The ‘tokenmaxxing’ trend, for example, has prompted a surge in usage and landed some firms with hefty bills. </p><p>As <em>ITPro </em>previously reported, <a href="https://www.itpro.com/technology/artificial-intelligence/what-were-seeing-right-now-is-just-rapid-escalation-in-ai-token-spend-accenture-tells-staff-to-stop-using-ai-for-unnecessary-tasks-amid-surging-costs">Accenture asked workers to stop using AI for basic tasks</a> amidst skyrocketing costs while Uber revealed it <a href="https://www.itpro.com/technology/artificial-intelligence/ubers-eye-watering-ai-bill-shows-enterprises-are-still-measuring-ai-success-through-consumption-rather-than-outcomes-and-its-warping-our-perception-of-roi-and-productivity">blew through its entire annual AI budget in just four months</a>. </p><p>Rising operational costs haven’t dampened investment rates, however. More than one-quarter (28%) of respondents said their organisation is now spending more than 10% of their total IT budget on AI tools. </p><p>“Looking ahead, 60 percent of respondents expect their organizations to increase their AI investments over the next year,” McKinsey noted. </p><p>“Respondents in pharmaceuticals and medical products, insurance, and banking and other financial institutions are the most likely to expect increasing investment.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/business-strategy/a-striking-finding-this-year-is-the-gap-between-individual-gains-and-enterprise-impact-mckinsey-says-ai-is-finally-paying-off-for-enterprises-but-rising-costs-and-constrained-efficiency-boosts-are-still-a-major-hurdle</link>
                                                                            <description>
                            <![CDATA[ Rising operational costs, mismatched productivity gains, and sluggish revenue improvements haven’t dampened the mood for investment ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">N4J9ZD6uBp52jNHWfAkGYJ</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/kBTuBeV7BUJXkLLS7QbGjV-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 26 Aug 2026 09:44:09 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 15:00:04 +0000</updated>
                                                                                                                                            <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/kBTuBeV7BUJXkLLS7QbGjV-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Digital transformation concept image showing two horizontal pillars separated with new blocks branching out to connect both.]]></media:description>                                                            <media:text><![CDATA[Digital transformation concept image showing two horizontal pillars separated with new blocks branching out to connect both.]]></media:text>
                                <media:title type="plain"><![CDATA[Digital transformation concept image showing two horizontal pillars separated with new blocks branching out to connect both.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/kBTuBeV7BUJXkLLS7QbGjV-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>New research from McKinsey suggests enterprises are finally reporting benefits from AI, but cost-related concerns and “constrained” productivity gains still weigh heavy on the minds of IT leaders.</p><p>Figures from the consultancy’s <a href="https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai" target="_blank"><u><em>State of AI in 2026</em></u></a> report show more than one-third (37%) of respondents recorded “some” earnings impact with the technology, although that marks the same share from last year’s report. </p><p>Meanwhile, “AI high performers” said AI is now having a “significant” impact on operational efficiency and earnings. </p><p>The report suggests that enterprise AI maturity is rising, with an increasing number of enterprises now successfully scaling the technology. Improvements in this regard are being reported across the board, McKinsey noted, with respondents scaling chatbots, coding agents, and more sophisticated AI agents. </p><p>The use of agentic AI has increased significantly, for example, with 40% of large enterprises scaling agents, up from 27% last year. Meanwhile, around two-in-ten are actively scaling software coding agents. </p><p>“Among <a href="https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today">AI tools</a>, chatbots are the most widely scaled, with 47 percent of respondents saying their organizations are scaling them across the enterprise,” the report states. </p><p>“About two-in-ten respondents report reaching the scaling phase across their organization with AI agents and a similar share report the same with software coding agents.”</p><h2 id="mismatched-ai-productivity">Mismatched AI productivity</h2><p>Workforce productivity gains rank among the most notable improvements for enterprises, according to McKinsey. </p><p>Around 80% of respondents said that AI has improved individual productivity, while 50% report that AI helps them “make better decisions”. Crucially, however, McKinsey warned that “the experience is not universally positive”. </p><p>Mid-level managers and individual workers are more likely than executives to report AI-related problems. McKinsey said this shows that AI impact “remains concentrated” among certain groups within the enterprise. </p><p>This is by no means the first study to highlight this trend. Research from Accenture in April this year found <a href="https://www.itpro.com/business/business-strategy/ai-productivity-challenges-accenture-generating-impact-study"><u>many enterprises are dealing with mismatched AI productivity gains</u></a>. </p><p>Similar to McKinsey, the report found that workers are reporting benefits with the technology such as higher-quality output levels and faster project delivery times. </p><p>Yet the report pointed to a “widening gap” between the basic use of the technology and its wider enterprise impact. A key factor, Accenture said, lies in the fact that organizations' process changes are “lagging behind” integration rates. </p><p>Similarly, many AI strategies typically fail to modernize underlying IT infrastructure or workflows to accommodate the technology. </p><p>“A striking finding this year is the gap between individual gains and enterprise impact. At the individual level, AI is clearly a boon: 80 percent of survey respondents say it has improved their productivity and half say it helps them make better decisions,” said Dan Tinkoff, senior partner at McKinsey. </p><p>“Yet, only 37 percent of organizations report any positive EBIT contribution, essentially flat compared with last year.”</p><h2 id="ai-costs-are-taking-the-shine-off-enterprise-gains">AI costs are taking the shine off enterprise gains</h2><p>Cost-related concerns continue to weigh heavy on IT leaders’ minds, according to McKinsey. Around 20% of respondents said that AI-related operating costs have “constrained their AI use”. </p><p>“Those cost constraints are being reported across the full range of AI tools. For each of three tools – <a href="https://www.itpro.com/technology/artificial-intelligence-ai/369979/chatgpt-vs-chatbots-whats-the-difference">AI chatbots</a>, AI agents, and software <a href="https://www.itpro.com/software/development/while-the-engineers-slept-the-agents-kept-building-aws-uk-chief-touts-big-gains-with-ai-powered-coding">coding agents</a> – about one-in-ten respondents say their organizations’ use has been constrained by costs,” the report states. </p><p>McKinsey also highlighted <a href="https://www.itpro.com/technology/artificial-intelligence/it-leaders-are-being-stung-by-unexpected-ai-costs">token-related cost concerns</a>. This has become a recurring pain point for some businesses over the last year as firms ramp up adoption of the technology. </p><p>The ‘tokenmaxxing’ trend, for example, has prompted a surge in usage and landed some firms with hefty bills. </p><p>As <em>ITPro </em>previously reported, <a href="https://www.itpro.com/technology/artificial-intelligence/what-were-seeing-right-now-is-just-rapid-escalation-in-ai-token-spend-accenture-tells-staff-to-stop-using-ai-for-unnecessary-tasks-amid-surging-costs">Accenture asked workers to stop using AI for basic tasks</a> amidst skyrocketing costs while Uber revealed it <a href="https://www.itpro.com/technology/artificial-intelligence/ubers-eye-watering-ai-bill-shows-enterprises-are-still-measuring-ai-success-through-consumption-rather-than-outcomes-and-its-warping-our-perception-of-roi-and-productivity">blew through its entire annual AI budget in just four months</a>. </p><p>Rising operational costs haven’t dampened investment rates, however. More than one-quarter (28%) of respondents said their organisation is now spending more than 10% of their total IT budget on AI tools. </p><p>“Looking ahead, 60 percent of respondents expect their organizations to increase their AI investments over the next year,” McKinsey noted. </p><p>“Respondents in pharmaceuticals and medical products, insurance, and banking and other financial institutions are the most likely to expect increasing investment.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ IT leaders need to stop ‘pushing AI for the sake of AI’ ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Enterprises are falling into a trap of using AI when it’s not necessarily needed, and it’s costing them dearly, according to Roger Lee, AVP of solutions consulting at Appian.</p><p>Speaking to <em>ITPro</em>, Lee said the hype surrounding AI - particularly agents - is pushing some IT leaders to apply the technology in an ‘AI for the sake of AI’ type approach. </p><p>“I think a lot of organizations and a lot of senior executives and staff within organizations have been told to ‘do AI’,” he said. </p><p>“So rather than just pushing AI for the sake of AI, look at what the decisions are we’re trying to improve, what data we are using, and where does that human oversight sit.”</p><p>Lee’s comments come after Accenture <a href="https://www.itpro.com/technology/artificial-intelligence/what-were-seeing-right-now-is-just-rapid-escalation-in-ai-token-spend-accenture-tells-staff-to-stop-using-ai-for-unnecessary-tasks-amid-surging-costs"><u>told staff to stop using AI for unnecessary tasks</u></a> amid surging costs. As adoption of the technology accelerates, he told <em>ITPro </em>that enterprises need to have a serious conversation on how they integrate <a href="https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today">AI tools</a>.  </p><h2 id="agents-vs-business-rules">Agents vs business rules</h2><p>Pursuing the ‘AI for the sake of AI’ approach means that many businesses aren’t just incurring huge costs, but adding needless complexity to processes that were already streamlined prior to the generative AI race. </p><p>“The key thing is really trying to identify the right place for AI versus just trying to use AI to solve everything,” he said. </p><p>“In many cases, if it’s straightforward automation, if there’s a business rule where you’ve got a definitive yes or no answer, then sometimes business rules and workflows that result in a definitive answer are the right way to go.”</p><p>Admittedly, Lee said there are instances where agents are necessary, especially with regard to research or complex problem solving tasks. Acknowledging that this is a case of different strokes is crucial, however. </p><p>“I take an example from insurance with an automation claim, where you wanted to do some research and do some work, then an AI agent might be the right solution for that,” he said. </p><p>“But it's really understanding the right place to use a rule versus an AI agent, and the impact that that can give.”</p><p>This is particularly relevant in regulated industries, Lee said, where margins of error can have huge ramifications. </p><p>Agents are highly effective in terms of dynamic reasoning, he noted, but in cases where businesses need a “really definitive zero error response” then IT leaders need to know where and <em>when </em>to automate. </p><p>“Regulated industries is another area where humans in the loop are really important,” he said. </p><p>“It's what makes processes accountable. So to ensure that not everything is automated, where you have things that are exceptions or outliers, that you absolutely have humans in the loop to respond to those.”</p><h2 id="shoehorning-ai-into-processes">Shoehorning AI into processes</h2><p>Enterprises have found themselves <a href="https://www.itpro.com/business/business-strategy/enterprises-need-to-stop-shoehorning-ai-where-it-isnt-needed"><u>shoehorning AI into areas and processes that aren’t applicable</u></a>, research from Gartner shows. Moreover, they’re expecting near-immediate returns with the technology. </p><p>Ironically, a key factor behind this lies in the desire to deliver returns on investment with AI. Yet by doing so they’re less likely to unlock any benefits. </p><p>Speaking at the time, Melanie Freeze, director of research at Gartner, said that many AI initiatives are “overly ambitious or poorly scoped” and don’t align with “real operational needs”.</p><p>Ultimately, Lee told <em>ITPro </em>that as with any adoption process, AI should still be viewed as a “business pull rather than technology push” approach. </p><p>“I think that the whole value that you can get from AI within an organization is all around what the business needs, rather than what the technology can push to the organisation,” he said. </p><p>“First and foremost, what problems are we trying to solve, rather than where can we use AI, and that changes the perspective on what we're trying to do,” Lee added.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/business-strategy/it-leaders-need-to-stop-pushing-ai-for-the-sake-of-ai</link>
                                                                            <description>
                            <![CDATA[ As enterprise adoption of AI agents continues, IT leaders need to know where and when to apply the technology ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">49UZGoqtpFRhamJeXZfBrN</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/4CeYDBXGHtWfkgtgjJ2dUi-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Tue, 25 Aug 2026 10:41:06 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 15:00:11 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/4CeYDBXGHtWfkgtgjJ2dUi-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[A wide angle shot of IT workers in an office, lit by large modern windows showing a business park behind them.]]></media:description>                                                            <media:text><![CDATA[A wide angle shot of IT workers in an office, lit by large modern windows showing a business park behind them.]]></media:text>
                                <media:title type="plain"><![CDATA[A wide angle shot of IT workers in an office, lit by large modern windows showing a business park behind them.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/4CeYDBXGHtWfkgtgjJ2dUi-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Enterprises are falling into a trap of using AI when it’s not necessarily needed, and it’s costing them dearly, according to Roger Lee, AVP of solutions consulting at Appian.</p><p>Speaking to <em>ITPro</em>, Lee said the hype surrounding AI - particularly agents - is pushing some IT leaders to apply the technology in an ‘AI for the sake of AI’ type approach. </p><p>“I think a lot of organizations and a lot of senior executives and staff within organizations have been told to ‘do AI’,” he said. </p><p>“So rather than just pushing AI for the sake of AI, look at what the decisions are we’re trying to improve, what data we are using, and where does that human oversight sit.”</p><p>Lee’s comments come after Accenture <a href="https://www.itpro.com/technology/artificial-intelligence/what-were-seeing-right-now-is-just-rapid-escalation-in-ai-token-spend-accenture-tells-staff-to-stop-using-ai-for-unnecessary-tasks-amid-surging-costs"><u>told staff to stop using AI for unnecessary tasks</u></a> amid surging costs. As adoption of the technology accelerates, he told <em>ITPro </em>that enterprises need to have a serious conversation on how they integrate <a href="https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today">AI tools</a>.  </p><h2 id="agents-vs-business-rules">Agents vs business rules</h2><p>Pursuing the ‘AI for the sake of AI’ approach means that many businesses aren’t just incurring huge costs, but adding needless complexity to processes that were already streamlined prior to the generative AI race. </p><p>“The key thing is really trying to identify the right place for AI versus just trying to use AI to solve everything,” he said. </p><p>“In many cases, if it’s straightforward automation, if there’s a business rule where you’ve got a definitive yes or no answer, then sometimes business rules and workflows that result in a definitive answer are the right way to go.”</p><p>Admittedly, Lee said there are instances where agents are necessary, especially with regard to research or complex problem solving tasks. Acknowledging that this is a case of different strokes is crucial, however. </p><p>“I take an example from insurance with an automation claim, where you wanted to do some research and do some work, then an AI agent might be the right solution for that,” he said. </p><p>“But it's really understanding the right place to use a rule versus an AI agent, and the impact that that can give.”</p><p>This is particularly relevant in regulated industries, Lee said, where margins of error can have huge ramifications. </p><p>Agents are highly effective in terms of dynamic reasoning, he noted, but in cases where businesses need a “really definitive zero error response” then IT leaders need to know where and <em>when </em>to automate. </p><p>“Regulated industries is another area where humans in the loop are really important,” he said. </p><p>“It's what makes processes accountable. So to ensure that not everything is automated, where you have things that are exceptions or outliers, that you absolutely have humans in the loop to respond to those.”</p><h2 id="shoehorning-ai-into-processes">Shoehorning AI into processes</h2><p>Enterprises have found themselves <a href="https://www.itpro.com/business/business-strategy/enterprises-need-to-stop-shoehorning-ai-where-it-isnt-needed"><u>shoehorning AI into areas and processes that aren’t applicable</u></a>, research from Gartner shows. Moreover, they’re expecting near-immediate returns with the technology. </p><p>Ironically, a key factor behind this lies in the desire to deliver returns on investment with AI. Yet by doing so they’re less likely to unlock any benefits. </p><p>Speaking at the time, Melanie Freeze, director of research at Gartner, said that many AI initiatives are “overly ambitious or poorly scoped” and don’t align with “real operational needs”.</p><p>Ultimately, Lee told <em>ITPro </em>that as with any adoption process, AI should still be viewed as a “business pull rather than technology push” approach. </p><p>“I think that the whole value that you can get from AI within an organization is all around what the business needs, rather than what the technology can push to the organisation,” he said. </p><p>“First and foremost, what problems are we trying to solve, rather than where can we use AI, and that changes the perspective on what we're trying to do,” Lee added.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Agentic AI is spurring a ‘fundamental shift’ in cloud infrastructure consumption ]]></title>
                                                                                                <dc:content><![CDATA[ <p>New figures from Gartner point to a surge in AI-optimized infrastructure spending over the next year. A senior analyst at the firm told <em>ITPro </em>it underlines how it is driving a “fundamental shift” in cloud consumption and design. </p><p>Recent analysis from the firm found spending on this front is projected to grow 96% across 2026, reaching a total of $42 billion as enterprises look to overhaul infrastructure with the specific aim of dealing with cumbersome AI workloads. </p><p>Hardeep Singh, senior principal research analyst at Gartner, told <em>ITPro </em>the consultancy expects this trend to continue as adoption rates rise. Indeed, Gartner projects the market to “sustain high growth” and reach $66 billion in spending. </p><p>“There is a fundamental shift in how cloud infrastructure is being designed and consumed,” Singh told <em>ITPro</em>. “We are transforming from traditional general purpose cloud computing towards purpose-built AI infrastructure designed specifically for <a href="https://www.itpro.com/technology/artificial-intelligence/generative-ai-vs-large-language-models">large language models (LLMs)</a> and agentic workloads”. </p><p>A key factor behind the optimized infrastructure push lies in the changing shape of AI use, Singh noted. The technology is maturing and operationalization of AI, particularly agentic AI, is “accelerating inference demand” across the board. </p><p>Agentic AI has a far larger token consumption rate compared to traditional chatbot-based AI interactions, which in turn increases inference demand.</p><p><a href="https://signal65.com/wp-content/uploads/2026/05/Signal65-Insights_The-Economics-of-Agentic-AI.pdf" target="_blank"><u>Research from Signal65</u></a>, for example, found agentic workloads consume “orders of magnitude more tokens”, with consumption rates anywhere between four to 15-times more. </p><p>“You see [with] agentic AI it has a variable consumption,” Singh told <em>ITPro</em>. “It would be one prompt triggering autonomous model calls, and that would lead to a huge amount of compute consumption.”</p><p>Gartner projects that inference demands will surge over the next year on the back of increased agentic AI adoption. Indeed, the consultancy noted that spending on this front will reach $23.3 billion, surpassing the $19 billion spent on AI inference. </p><p>The fact that inference spending will exceed training for the first time is noteworthy, Singh told <em>ITPro</em>. Many enterprises are still in the early stages of agentic AI adoption though. As projects mature, this is certainly expected to drive inference demands even further. </p><p>Gartner ranks AI use based on a level system, with basic chatbots at the first level, and fully autonomous at level five, the highest rating. Most businesses aren’t even close to that level, he said. </p><p>“It will still take some time to go and reach the autonomous side,” Singh told <em>ITPro</em>. </p><p>“We are still in that L1, L2, L3 kind of scenario where experimentations are happening, trying to understand which use cases fit in which scenario, and then how the agents [are] going to solve the complex business problems.”</p><p>Given the current state of agentic adoption and projected advances on this front, AI-optimized infrastructure will be even critical for meeting demand. </p><p>Gartner figures show that 55% of spending on this front will directly support inference. This is set to reach 59% by 2027. </p><h2 id="changing-priorities">Changing priorities</h2><p>According to Gartner, the growing volume of inference workloads will ultimately “reshape cloud investment priorities”. Singh noted that this can already be observed in cloud strategies across a range of industries. </p><p><a href="https://my.idc.com/getdoc.jsp?containerId=prAP53785525" target="_blank"><u>Research from IDC</u></a> in late 2025 predicted a sharpened focus on hybrid and even private cloud models due to a combination of factors, most notably “more predictable cost models”.</p><p>Singh told <em>ITPro </em>he expects this focus to continue as agentic adoption accelerates, particularly with the rise of multi-agent frameworks in which they rely on several separate underlying AI models. </p><p>“The important thing will be how enterprises want to use multi-agents backed by multiple models, so multimodality and multi-models,” he said. </p><p>“Both these things are important in the context where some of the data workloads [are] moving to the cloud, training the large language model, then deploying on the edge on prem to get the inference and real-time monitoring.”</p><p>“<a href="https://www.itpro.com/hybrid-cloud/34384/multi-cloud-vs-hybrid-cloud-whats-the-difference">Hybrid and multi-cloud</a> will be more picked up in these scenarios, because some enterprises know that it is difficult to train the model on-prem because of the amount of capacity it needs.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/cloud/cloud-computing/agentic-ai-is-spurring-a-fundamental-shift-in-cloud-infrastructure-consumption</link>
                                                                            <description>
                            <![CDATA[ Surging inference demands mean enterprises are flocking to AI-optimized infrastructure to keep pace ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">PQaka7JbyASeKenajXBBab</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/CbbpGhvJgeZMgcMmxS62j5-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Mon, 17 Aug 2026 15:24:39 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 13:31:45 +0000</updated>
                                                                                                                                            <category><![CDATA[Cloud Computing]]></category>
                                                    <category><![CDATA[IT Infrastructure]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/CbbpGhvJgeZMgcMmxS62j5-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Cloud computing concept image showing a cloud with interconnected pink-colored data points.]]></media:description>                                                            <media:text><![CDATA[Cloud computing concept image showing a cloud with interconnected pink-colored data points.]]></media:text>
                                <media:title type="plain"><![CDATA[Cloud computing concept image showing a cloud with interconnected pink-colored data points.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/CbbpGhvJgeZMgcMmxS62j5-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>New figures from Gartner point to a surge in AI-optimized infrastructure spending over the next year. A senior analyst at the firm told <em>ITPro </em>it underlines how it is driving a “fundamental shift” in cloud consumption and design. </p><p>Recent analysis from the firm found spending on this front is projected to grow 96% across 2026, reaching a total of $42 billion as enterprises look to overhaul infrastructure with the specific aim of dealing with cumbersome AI workloads. </p><p>Hardeep Singh, senior principal research analyst at Gartner, told <em>ITPro </em>the consultancy expects this trend to continue as adoption rates rise. Indeed, Gartner projects the market to “sustain high growth” and reach $66 billion in spending. </p><p>“There is a fundamental shift in how cloud infrastructure is being designed and consumed,” Singh told <em>ITPro</em>. “We are transforming from traditional general purpose cloud computing towards purpose-built AI infrastructure designed specifically for <a href="https://www.itpro.com/technology/artificial-intelligence/generative-ai-vs-large-language-models">large language models (LLMs)</a> and agentic workloads”. </p><p>A key factor behind the optimized infrastructure push lies in the changing shape of AI use, Singh noted. The technology is maturing and operationalization of AI, particularly agentic AI, is “accelerating inference demand” across the board. </p><p>Agentic AI has a far larger token consumption rate compared to traditional chatbot-based AI interactions, which in turn increases inference demand.</p><p><a href="https://signal65.com/wp-content/uploads/2026/05/Signal65-Insights_The-Economics-of-Agentic-AI.pdf" target="_blank"><u>Research from Signal65</u></a>, for example, found agentic workloads consume “orders of magnitude more tokens”, with consumption rates anywhere between four to 15-times more. </p><p>“You see [with] agentic AI it has a variable consumption,” Singh told <em>ITPro</em>. “It would be one prompt triggering autonomous model calls, and that would lead to a huge amount of compute consumption.”</p><p>Gartner projects that inference demands will surge over the next year on the back of increased agentic AI adoption. Indeed, the consultancy noted that spending on this front will reach $23.3 billion, surpassing the $19 billion spent on AI inference. </p><p>The fact that inference spending will exceed training for the first time is noteworthy, Singh told <em>ITPro</em>. Many enterprises are still in the early stages of agentic AI adoption though. As projects mature, this is certainly expected to drive inference demands even further. </p><p>Gartner ranks AI use based on a level system, with basic chatbots at the first level, and fully autonomous at level five, the highest rating. Most businesses aren’t even close to that level, he said. </p><p>“It will still take some time to go and reach the autonomous side,” Singh told <em>ITPro</em>. </p><p>“We are still in that L1, L2, L3 kind of scenario where experimentations are happening, trying to understand which use cases fit in which scenario, and then how the agents [are] going to solve the complex business problems.”</p><p>Given the current state of agentic adoption and projected advances on this front, AI-optimized infrastructure will be even critical for meeting demand. </p><p>Gartner figures show that 55% of spending on this front will directly support inference. This is set to reach 59% by 2027. </p><h2 id="changing-priorities">Changing priorities</h2><p>According to Gartner, the growing volume of inference workloads will ultimately “reshape cloud investment priorities”. Singh noted that this can already be observed in cloud strategies across a range of industries. </p><p><a href="https://my.idc.com/getdoc.jsp?containerId=prAP53785525" target="_blank"><u>Research from IDC</u></a> in late 2025 predicted a sharpened focus on hybrid and even private cloud models due to a combination of factors, most notably “more predictable cost models”.</p><p>Singh told <em>ITPro </em>he expects this focus to continue as agentic adoption accelerates, particularly with the rise of multi-agent frameworks in which they rely on several separate underlying AI models. </p><p>“The important thing will be how enterprises want to use multi-agents backed by multiple models, so multimodality and multi-models,” he said. </p><p>“Both these things are important in the context where some of the data workloads [are] moving to the cloud, training the large language model, then deploying on the edge on prem to get the inference and real-time monitoring.”</p><p>“<a href="https://www.itpro.com/hybrid-cloud/34384/multi-cloud-vs-hybrid-cloud-whats-the-difference">Hybrid and multi-cloud</a> will be more picked up in these scenarios, because some enterprises know that it is difficult to train the model on-prem because of the amount of capacity it needs.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Cyber experts issue warning over new 'memory poisoning' technique targeting AI agents ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Security experts have issued a warning over a new technique aimed at duping AI agents into producing false information. </p><p><a href="https://www.forcepoint.com/blog/x-labs/persistent-memory-poisoning-ai-agents" target="_blank">Analysis from Forcepoint</a> found a poisoned webpage or document can plant a fake vendor, support contact, or security rule into an agent's memory, with the fake data being surfaced as fact months later.</p><p>While normal <a href="https://www.itpro.com/security/ncsc-issues-urgent-warning-over-growing-ai-prompt-injection-risks-heres-what-you-need-to-know">prompt injection attacks</a> come to an end when the session finishes, Forcepoint said memory poisoning allows attackers to inject misleading or malicious memory items into an assistant's persistent memory layer. </p><p>The aim here, according to researchers, is to ensure the agent retrieves and trusts these malicious items, which could be used in future attacks or to simply harm trust.</p><p>"Persistent memory poisoning is an emerging attack class against AI assistants and agentic systems that store long-term memory items, user preferences, task history or operational context across sessions," warned Forcepoint senior security researcher Syed Hassan Faizan.</p><p>"As AI agents become increasingly capable of storing and reusing long-term memory, the memory itself becomes a new target surface."</p><p>The material concerned could include a fake domain, support contact or vendor, fake internal procedures, security rules, and approval chains, or any preference that weakens security.</p><p>These can come from a number of sources - malicious or compromised web pages, shared documents, support tickets, email footers, <a href="https://www.itpro.com/marketing-comms/business-communications/355389/slack-vs-microsoft-teams-business-messaging-apps">Slack or Teams</a> messages, PDFs with hidden text, or knowledge base articles.</p><p>"Indirect prompt injection is especially relevant here because the victim may never see the malicious instruction. The agent reads it while processing otherwise normal content," said Hassan Faizan. </p><h2 id="compromised-agents">Compromised agents</h2><p>In a proof of concept detailing the attack method, Forcepoint highlighted an example in which a travel assistant with browser capabilities was poisoned. </p><p>Although it can't access the internet directly, a separate browser tool handles that on the assistant's behalf. Thereafter the assistant remembers user preferences and trusted booking instructions across sessions. </p><p>Here, an attacker can create a webpage designed to rank for a search - something like '<em>Emergency Travel Booking Advisory</em>' that looks completely legitimate. </p><p>But when the AI agent visits it, it sees hidden additional text meant for the AI system, not a human reader. The enterprise browser tool downloads the full HTML - including the hidden text. </p><p>"Since <a href="https://www.itpro.com/technology/artificial-intelligence/generative-ai-vs-large-language-models">large language models</a> don't parse raw HTML well, a separate program extracts all the readable text, both visible and hidden, and hands the model plain text instead of markup. The model never sees the webpage itself," said Hassan Faizan. </p><p>The result is that the agent stores this text as a trusted, useful fact in its long-term memory. When asked for travel advice later, it retrieves the poisoned memory and recommends the fake provider inserted by the attacker.</p><h2 id="how-to-tackle-memory-poisoning">How to tackle memory poisoning</h2><p>Forcepoint said prevention will essentially rely on inspecting long-term memory before it's implanted into a trusted environment. </p><p>This can be done through memory risk scoring, a heuristic-based framework that assigns a risk score to each memory object based on multiple security signals. </p><p>High-risk memories can then be rejected, quarantined, flagged for user confirmation, or marked as untrusted.  </p><p>"Securing AI memory should become a fundamental part of <a href="https://www.itpro.com/technology/artificial-intelligence/six-generative-ai-cyber-security-threats-and-how-to-mitigate-them">AI security</a> architecture," said Hassan Faizan. </p><p>"Just as organizations inspect files, emails and network traffic before trusting them, AI systems must also evaluate the trustworthiness of their memory items before those memories influence decision-making."</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/security/trust-your-ai-agents-new-research-shows-memory-poisoning-can-dupe-them-into-remembering-fake-information-and-its-a-huge-security-risk</link>
                                                                            <description>
                            <![CDATA[ Memory poisoning allows hidden text on a webpage to be treated as fact and used to make harmful decisions ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">UV3iaEsg495eahXyXCSZqg</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/E5oTpGPpaF5BNR6kNs4Zr5-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Mon, 10 Aug 2026 11:20:20 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 13:07:50 +0000</updated>
                                                                                                                                            <category><![CDATA[Application Security (DevSecOps)]]></category>
                                                    <category><![CDATA[Security]]></category>
                                                    <category><![CDATA[Cybersecurity]]></category>
                                                                                                                    <dc:creator><![CDATA[ Emma Woollacott ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/aWfskavxoVSMDy6cDWtYmJ-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/E5oTpGPpaF5BNR6kNs4Zr5-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Cracked and shatter pane of glass fracturing against a black backdrop. ]]></media:description>                                                            <media:text><![CDATA[Cracked and shatter pane of glass fracturing against a black backdrop. ]]></media:text>
                                <media:title type="plain"><![CDATA[Cracked and shatter pane of glass fracturing against a black backdrop. ]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/E5oTpGPpaF5BNR6kNs4Zr5-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Security experts have issued a warning over a new technique aimed at duping AI agents into producing false information. </p><p><a href="https://www.forcepoint.com/blog/x-labs/persistent-memory-poisoning-ai-agents" target="_blank">Analysis from Forcepoint</a> found a poisoned webpage or document can plant a fake vendor, support contact, or security rule into an agent's memory, with the fake data being surfaced as fact months later.</p><p>While normal <a href="https://www.itpro.com/security/ncsc-issues-urgent-warning-over-growing-ai-prompt-injection-risks-heres-what-you-need-to-know">prompt injection attacks</a> come to an end when the session finishes, Forcepoint said memory poisoning allows attackers to inject misleading or malicious memory items into an assistant's persistent memory layer. </p><p>The aim here, according to researchers, is to ensure the agent retrieves and trusts these malicious items, which could be used in future attacks or to simply harm trust.</p><p>"Persistent memory poisoning is an emerging attack class against AI assistants and agentic systems that store long-term memory items, user preferences, task history or operational context across sessions," warned Forcepoint senior security researcher Syed Hassan Faizan.</p><p>"As AI agents become increasingly capable of storing and reusing long-term memory, the memory itself becomes a new target surface."</p><p>The material concerned could include a fake domain, support contact or vendor, fake internal procedures, security rules, and approval chains, or any preference that weakens security.</p><p>These can come from a number of sources - malicious or compromised web pages, shared documents, support tickets, email footers, <a href="https://www.itpro.com/marketing-comms/business-communications/355389/slack-vs-microsoft-teams-business-messaging-apps">Slack or Teams</a> messages, PDFs with hidden text, or knowledge base articles.</p><p>"Indirect prompt injection is especially relevant here because the victim may never see the malicious instruction. The agent reads it while processing otherwise normal content," said Hassan Faizan. </p><h2 id="compromised-agents">Compromised agents</h2><p>In a proof of concept detailing the attack method, Forcepoint highlighted an example in which a travel assistant with browser capabilities was poisoned. </p><p>Although it can't access the internet directly, a separate browser tool handles that on the assistant's behalf. Thereafter the assistant remembers user preferences and trusted booking instructions across sessions. </p><p>Here, an attacker can create a webpage designed to rank for a search - something like '<em>Emergency Travel Booking Advisory</em>' that looks completely legitimate. </p><p>But when the AI agent visits it, it sees hidden additional text meant for the AI system, not a human reader. The enterprise browser tool downloads the full HTML - including the hidden text. </p><p>"Since <a href="https://www.itpro.com/technology/artificial-intelligence/generative-ai-vs-large-language-models">large language models</a> don't parse raw HTML well, a separate program extracts all the readable text, both visible and hidden, and hands the model plain text instead of markup. The model never sees the webpage itself," said Hassan Faizan. </p><p>The result is that the agent stores this text as a trusted, useful fact in its long-term memory. When asked for travel advice later, it retrieves the poisoned memory and recommends the fake provider inserted by the attacker.</p><h2 id="how-to-tackle-memory-poisoning">How to tackle memory poisoning</h2><p>Forcepoint said prevention will essentially rely on inspecting long-term memory before it's implanted into a trusted environment. </p><p>This can be done through memory risk scoring, a heuristic-based framework that assigns a risk score to each memory object based on multiple security signals. </p><p>High-risk memories can then be rejected, quarantined, flagged for user confirmation, or marked as untrusted.  </p><p>"Securing AI memory should become a fundamental part of <a href="https://www.itpro.com/technology/artificial-intelligence/six-generative-ai-cyber-security-threats-and-how-to-mitigate-them">AI security</a> architecture," said Hassan Faizan. </p><p>"Just as organizations inspect files, emails and network traffic before trusting them, AI systems must also evaluate the trustworthiness of their memory items before those memories influence decision-making."</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Cognizant launches dedicated EMEA AI unit to accelerate enterprise adoption ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Cognizant has announced the launch of a new EMEA AI Unit, in a move tit says will help organizations scale agentic AI deployments and move projects from pilot stages into production.</p><p>The dedicated unit will bring together advisory, engineering, and delivery capabilities to support enterprises across the region as they build, deploy, and manage agentic AI solutions tailored to their business requirements.</p><p>The launch forms part of Cognizant’s wider AI Builder strategy, which aims to help customers adopt AI technologies without being tied to a single cloud provider, AI model, or technology platform.</p><p>In an announcement, Cognizant’s president of EMEA, Manoj Mehta, said many organizations remain enthusiastic about AI but continue to face challenges translating early projects into measurable business outcomes.</p><p>“The EMEA AI Unit reflects Cognizant’s AI Builder strategy by bringing together the people, platforms and engineering expertise needed to move clients from pilots to payoff,” he explained.</p><p>“Our approach is neutral by design: we work across clouds, models and ecosystems so clients can build agentic AI solutions that fit their business, integrate into operations and support accountability for outcomes.”</p><p>Headquartered in New Jersey, Cognizant provides IT consulting, digital transformation, and technology services to enterprises around the world. In recent years, the company has increased its focus on AI-led transformation, developing services that combine consulting, software engineering, and managed delivery to help customers implement AI across their operations.</p><h2 id="turning-ai-into-business-outcomes">Turning AI into business outcomes</h2><p>At the center of its new EMEA AI Unit is Cognizant’s Frontier Deployed Engineering (FDE) offering, a delivery framework designed to help organizations progress from AI strategy through to enterprise-wide deployment.</p><p>The framework consists of three service models: Foundation, Accelerate, and Transform. Foundation focuses on AI strategy, governance, technology selection, and early-stage prototypes, while Accelerate is designed to identify and deploy high-value AI use cases into production.</p><p>The third tier, Transform, supports wider business reinvention through multi-agent AI systems capable of automating end-to-end workflows.</p><h2 id="early-impact-across-emea">Early impact across EMEA</h2><p>The unit is already supporting several notable enterprise deployments across the EMEA region, according to Cognizant.</p><p>The company is currently working with one of Europe’s largest online fashion retailers to move AI use cases into production through an AI factory model that it said can reduce development cycles from months to days.</p><p>Elsewhere, the firm revealed it is also supporting a global pharmaceutical company with the use of multi-agent AI systems across research and development – including drug discovery, clinical trial design, and regulatory preparation.</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/cognizant-launches-dedicated-emea-ai-unit-to-accelerate-enterprise-adoption</link>
                                                                            <description>
                            <![CDATA[ The new business unit will help organizations move agentic AI projects from pilot into production ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">iukXq3bMYTduZMtGU2xiFA</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/XUFTVxXHrTRAVBdujTYj6P-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Fri, 31 Jul 2026 09:03:07 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:14 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ itpro@futurenet.com (Daniel Todd) ]]></author>                    <dc:creator><![CDATA[ Daniel Todd ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/SRyC34qeLpNDj3dJtsVDhT-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/XUFTVxXHrTRAVBdujTYj6P-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Artificial intelligence (AI) concept image showing a digitized cube with &#039;AI&#039; written on it resting on top of circuit boards.]]></media:description>                                                            <media:text><![CDATA[Artificial intelligence (AI) concept image showing a digitized cube with &#039;AI&#039; written on it resting on top of circuit boards.]]></media:text>
                                <media:title type="plain"><![CDATA[Artificial intelligence (AI) concept image showing a digitized cube with &#039;AI&#039; written on it resting on top of circuit boards.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/XUFTVxXHrTRAVBdujTYj6P-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Cognizant has announced the launch of a new EMEA AI Unit, in a move tit says will help organizations scale agentic AI deployments and move projects from pilot stages into production.</p><p>The dedicated unit will bring together advisory, engineering, and delivery capabilities to support enterprises across the region as they build, deploy, and manage agentic AI solutions tailored to their business requirements.</p><p>The launch forms part of Cognizant’s wider AI Builder strategy, which aims to help customers adopt AI technologies without being tied to a single cloud provider, AI model, or technology platform.</p><p>In an announcement, Cognizant’s president of EMEA, Manoj Mehta, said many organizations remain enthusiastic about AI but continue to face challenges translating early projects into measurable business outcomes.</p><p>“The EMEA AI Unit reflects Cognizant’s AI Builder strategy by bringing together the people, platforms and engineering expertise needed to move clients from pilots to payoff,” he explained.</p><p>“Our approach is neutral by design: we work across clouds, models and ecosystems so clients can build agentic AI solutions that fit their business, integrate into operations and support accountability for outcomes.”</p><p>Headquartered in New Jersey, Cognizant provides IT consulting, digital transformation, and technology services to enterprises around the world. In recent years, the company has increased its focus on AI-led transformation, developing services that combine consulting, software engineering, and managed delivery to help customers implement AI across their operations.</p><h2 id="turning-ai-into-business-outcomes">Turning AI into business outcomes</h2><p>At the center of its new EMEA AI Unit is Cognizant’s Frontier Deployed Engineering (FDE) offering, a delivery framework designed to help organizations progress from AI strategy through to enterprise-wide deployment.</p><p>The framework consists of three service models: Foundation, Accelerate, and Transform. Foundation focuses on AI strategy, governance, technology selection, and early-stage prototypes, while Accelerate is designed to identify and deploy high-value AI use cases into production.</p><p>The third tier, Transform, supports wider business reinvention through multi-agent AI systems capable of automating end-to-end workflows.</p><h2 id="early-impact-across-emea">Early impact across EMEA</h2><p>The unit is already supporting several notable enterprise deployments across the EMEA region, according to Cognizant.</p><p>The company is currently working with one of Europe’s largest online fashion retailers to move AI use cases into production through an AI factory model that it said can reduce development cycles from months to days.</p><p>Elsewhere, the firm revealed it is also supporting a global pharmaceutical company with the use of multi-agent AI systems across research and development – including drug discovery, clinical trial design, and regulatory preparation.</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Amazon targets agent safety gains with investment in team behind Lean programming language ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Amazon has pledged financial backing for the team behind the Lean programming language as part of a sharpened focus on AI agent development. </p><p>In a <a href="https://www.amazon.science/news/amazon-is-investing-in-the-lean-focused-research-organization" target="_blank"><u>statement </u></a>on 26 July, the tech giant revealed it will provide "substantial, long-term” financial support for the Lean Focused Research Organization (FRO). </p><p>The FRO leads development of the open source programming language, which Amazon already uses internally in agentic AI development practices. </p><p>Lean has been around since 2014 and is a <a href="https://www.itpro.com/careers/29133/the-top-programming-languages-you-need-to-learn">popular programming language</a> across a range of fields including mathematics, physics, and computer science. </p><p>For Amazon, the appeal lies in its role in helping develop and fine-tune AI reasoning capabilities. The <a href="https://www.itpro.com/software/development/why-python-is-the-programming-language-of-choice-for-ai-developers">programming language</a> is used alongside a ‘proof assistant’ - a type of software used to conduct checks of code to identify and sift out errors or flaws. </p><p>It’s these ‘correctness proofs’ that make the programming language a go-to option for academic researchers and professionals in scientific fields. </p><p>“AI generation of formal proofs in Lean has been a key method for training models with lower error rates, to the point that they are now producing correct solutions to research-level problems,” Amazon explained in a blog post. </p><h2 id="amazon-lean-ing-in-on-agentic-safety">Amazon Lean-ing in on agentic safety</h2><p>Amazon said its support for Lean comes at a critical time both for its own internal AI ambitions and for the industry at large as enterprises ramp up adoption of agents. </p><p>Safety has been a key recurring talking point for organizations on this front. As <em>ITPro </em>has previously reported, research shows growing concerns around access control, <a href="https://www.itpro.com/technology/artificial-intelligence/one-size-fits-all-agent-governance-sets-enterprises-up-to-fail"><u>governance</u></a>, <a href="https://www.itpro.com/security/enterprises-are-adopting-agents-faster-than-they-can-secure-and-govern-them-experts-warn-its-a-disaster-waiting-to-happen"><u>identity security</u></a>, and <a href="https://www.itpro.com/security/observability-will-be-key-to-agentic-ai-safety-says-microsoft-security-exec"><u>visibility </u></a>when it comes to agents. </p><p>“Coupling <a href="https://www.itpro.com/technology/artificial-intelligence-ai/369959/what-is-generative-ai">generative AI</a> with Lean's mathematical rigor will help enable verified, trustworthy AI agents,” the company said. </p><p>Amazon already uses Lean-based verification in its Bedrock AgentCore service, helping to “prove the correctness of the policy language” that ensures agents operate within strict boundaries.</p><p>Elsewhere, Lean underpins systems such as SampCert - which provides “mathematical guarantees” for privacy protections in the AWS Clean Rooms analytics platform - and AWS Neuron. </p><p>“One scientist recently used an LLM with Lean to prove the correctness of Amazon Aurora's segment repair protocol, our most durability-critical distributed protocol, in a fraction of the time it would have taken manually,” Amazon noted. </p><p>“The set of applications is growing fast, and this is just the beginning.”</p><p>Peter Van der Putten, assistant professor of AI at Leiden University and director of Pegasystems’ AI Lab, said AWS’ FRO backing is a welcomed move given recurring security-related concerns. </p><p>“It is encouraging to see AWS investing in open source technologies such as Lean that have the potential to provide stronger formal verification guarantees for software and AI systems,” he said. </p><p>“As agentic AI systems become increasingly autonomous, technologies that can provide stronger correctness and safety guarantees than just traditional testing will become ever more important.”</p><p>Van der Putten noted, however, that the inherent complexity of agentic AI development means more robust orchestration and governance is still of paramount importance. </p><p>“Enterprise agents operate within complex environments involving workflows, automated processes, policies, regulations, humans in the loop, multiple stakeholders, and countless exceptions and uncertainties that cannot simply be abstracted away but must be dealt with in practice,” he said. </p><p>“So even an agent whose behaviour has been verified against a specification must still be able to operate effectively in such a context,” Van der Putten added. </p><p>“In practice, the bigger challenge may lie in orchestrating agents, people, data, and systems together in a governed way, while reflecting ambiguity, trade-offs, conflicting objectives, and constantly evolving realities, rather than simply proving that an idealised agent behaves correctly.”</p><h2 id="open-source-backing">Open source backing</h2><p>Backing for the FRO serves a broader purpose rather than complementing its own use of Lean, according to Amazon. </p><p>“It’s easier to trust a proof when you can evaluate the tools behind it yourself,” the company noted. </p><p>“Customers, auditors, and regulators can independently inspect and validate work done in community-governed tools, which is the kind of transparency that safety-critical AI demands.”</p><p>The firm said that in the long term, Lean’s growing popularity will benefit the AI industry by providing “more libraries, more tooling, and more formalized proofs for everyone”. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/software/development/amazon-targets-agent-safety-gains-with-investment-in-team-behind-lean-programming-language</link>
                                                                            <description>
                            <![CDATA[ The tech giant hopes support for the open source programming language could drive AI agent safety improvements ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">pChcAGV279QDiarsAvdxvY</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/rPoGMChjniATnN9JwsqpVW-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Tue, 28 Jul 2026 15:00:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:31 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Development]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                    <category><![CDATA[Software]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/rPoGMChjniATnN9JwsqpVW-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Amazon logo and branding pictured hovering above a laptop keyboard with keys blurred.]]></media:description>                                                            <media:text><![CDATA[Amazon logo and branding pictured hovering above a laptop keyboard with keys blurred.]]></media:text>
                                <media:title type="plain"><![CDATA[Amazon logo and branding pictured hovering above a laptop keyboard with keys blurred.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/rPoGMChjniATnN9JwsqpVW-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Amazon has pledged financial backing for the team behind the Lean programming language as part of a sharpened focus on AI agent development. </p><p>In a <a href="https://www.amazon.science/news/amazon-is-investing-in-the-lean-focused-research-organization" target="_blank"><u>statement </u></a>on 26 July, the tech giant revealed it will provide "substantial, long-term” financial support for the Lean Focused Research Organization (FRO). </p><p>The FRO leads development of the open source programming language, which Amazon already uses internally in agentic AI development practices. </p><p>Lean has been around since 2014 and is a <a href="https://www.itpro.com/careers/29133/the-top-programming-languages-you-need-to-learn">popular programming language</a> across a range of fields including mathematics, physics, and computer science. </p><p>For Amazon, the appeal lies in its role in helping develop and fine-tune AI reasoning capabilities. The <a href="https://www.itpro.com/software/development/why-python-is-the-programming-language-of-choice-for-ai-developers">programming language</a> is used alongside a ‘proof assistant’ - a type of software used to conduct checks of code to identify and sift out errors or flaws. </p><p>It’s these ‘correctness proofs’ that make the programming language a go-to option for academic researchers and professionals in scientific fields. </p><p>“AI generation of formal proofs in Lean has been a key method for training models with lower error rates, to the point that they are now producing correct solutions to research-level problems,” Amazon explained in a blog post. </p><h2 id="amazon-lean-ing-in-on-agentic-safety">Amazon Lean-ing in on agentic safety</h2><p>Amazon said its support for Lean comes at a critical time both for its own internal AI ambitions and for the industry at large as enterprises ramp up adoption of agents. </p><p>Safety has been a key recurring talking point for organizations on this front. As <em>ITPro </em>has previously reported, research shows growing concerns around access control, <a href="https://www.itpro.com/technology/artificial-intelligence/one-size-fits-all-agent-governance-sets-enterprises-up-to-fail"><u>governance</u></a>, <a href="https://www.itpro.com/security/enterprises-are-adopting-agents-faster-than-they-can-secure-and-govern-them-experts-warn-its-a-disaster-waiting-to-happen"><u>identity security</u></a>, and <a href="https://www.itpro.com/security/observability-will-be-key-to-agentic-ai-safety-says-microsoft-security-exec"><u>visibility </u></a>when it comes to agents. </p><p>“Coupling <a href="https://www.itpro.com/technology/artificial-intelligence-ai/369959/what-is-generative-ai">generative AI</a> with Lean's mathematical rigor will help enable verified, trustworthy AI agents,” the company said. </p><p>Amazon already uses Lean-based verification in its Bedrock AgentCore service, helping to “prove the correctness of the policy language” that ensures agents operate within strict boundaries.</p><p>Elsewhere, Lean underpins systems such as SampCert - which provides “mathematical guarantees” for privacy protections in the AWS Clean Rooms analytics platform - and AWS Neuron. </p><p>“One scientist recently used an LLM with Lean to prove the correctness of Amazon Aurora's segment repair protocol, our most durability-critical distributed protocol, in a fraction of the time it would have taken manually,” Amazon noted. </p><p>“The set of applications is growing fast, and this is just the beginning.”</p><p>Peter Van der Putten, assistant professor of AI at Leiden University and director of Pegasystems’ AI Lab, said AWS’ FRO backing is a welcomed move given recurring security-related concerns. </p><p>“It is encouraging to see AWS investing in open source technologies such as Lean that have the potential to provide stronger formal verification guarantees for software and AI systems,” he said. </p><p>“As agentic AI systems become increasingly autonomous, technologies that can provide stronger correctness and safety guarantees than just traditional testing will become ever more important.”</p><p>Van der Putten noted, however, that the inherent complexity of agentic AI development means more robust orchestration and governance is still of paramount importance. </p><p>“Enterprise agents operate within complex environments involving workflows, automated processes, policies, regulations, humans in the loop, multiple stakeholders, and countless exceptions and uncertainties that cannot simply be abstracted away but must be dealt with in practice,” he said. </p><p>“So even an agent whose behaviour has been verified against a specification must still be able to operate effectively in such a context,” Van der Putten added. </p><p>“In practice, the bigger challenge may lie in orchestrating agents, people, data, and systems together in a governed way, while reflecting ambiguity, trade-offs, conflicting objectives, and constantly evolving realities, rather than simply proving that an idealised agent behaves correctly.”</p><h2 id="open-source-backing">Open source backing</h2><p>Backing for the FRO serves a broader purpose rather than complementing its own use of Lean, according to Amazon. </p><p>“It’s easier to trust a proof when you can evaluate the tools behind it yourself,” the company noted. </p><p>“Customers, auditors, and regulators can independently inspect and validate work done in community-governed tools, which is the kind of transparency that safety-critical AI demands.”</p><p>The firm said that in the long term, Lean’s growing popularity will benefit the AI industry by providing “more libraries, more tooling, and more formalized proofs for everyone”. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ How OpenAI models breached Hugging Face ]]></title>
                                                                                                <dc:content><![CDATA[ <p>OpenAI has admitted that its own AI models breached Hugging Face systems as part of a security test gone wrong. </p><p>In a <a href="https://openai.com/index/hugging-face-model-evaluation-security-incident/" target="_blank"><u>blog post</u></a> on 21 July, the firm revealed models escaped testing environments and gained unauthorized access to the AI hosting platform's production database. </p><p>Notably, OpenAI revealed the incident involved a pre-release, cyber-focused model. </p><p>“After investigating, we now know that this particular incident was driven by a combination of OpenAI models – including GPT-5.6 Sol and an even more capable pre-release model, all with reduced cyber refusals for evaluation purposes – while being internally tested on a benchmark of cyber capabilities,” the company wrote. </p><p>The admission comes after days of speculation amidst reports that an "external AI agent” had hacked the company. Hugging Face issued an <a href="https://huggingface.co/blog/security-incident-july-2026" target="_blank"><u>alert </u></a>last week warning that a breach had exposed internal datasets and user credentials. </p><h2 id="how-openai-hacked-hugging-face">How OpenAI hacked Hugging Face</h2><p>According to OpenAI, the incident occurred during an internal evaluation in which models were prompted to examine potential attack paths. </p><p>The aim here is to evaluate their cyber capabilities within an isolated environment, and crucially, without certain restrictions to prevent said models from “pursuing high-risk cyber activity”.  </p><p>“Our benchmarks run in a highly isolated environment, with network access constrained to the ability to install packages through an internally hosted third-party software that acts as a proxy and cache for package registries,” the company explained. </p><p>Both models “identified and chained” vulnerabilities across both OpenAI’s research environment and Hugging Face’s infrastructure, with the aim of gathering test solutions from the latter’s production database. </p><p>In this case, the models appear to have focused heavily on <a href="https://github.com/sunblaze-ucb/exploitgym" target="_blank"><u>ExploitGym</u></a>, a publicly-hosted benchmark tool used to evaluate how AI agents can turn software vulnerabilities into working exploits. </p><p>Notably, OpenAI revealed the models “identified and exploited a zero-day vulnerability” in the package installer registry. After successfully exploiting this zero-day, the models conducted a “series of privilege escalation and lateral movement actions” which allowed it to escape the test environment and gain full internet access. </p><p>“After gaining internet access, the models inferred that Hugging Face potentially hosted models, datasets, and solutions for ExploitGym,” the company noted. </p><p>“Knowing this, the model searched for and successfully found ways to gain access to secret information that it could use to cheat the evaluation.”</p><p>In one example, OpenAI revealed the model chained together several attack vectors using a combination of stolen credentials and zero-day flaws to identify a remote code execution (RCE) path on Hugging Face servers. </p><h2 id="what-comes-next">What comes next?</h2><p>OpenAI has detailed a series of actions it’s already taken in the wake of the incident, including closer coordination with Hugging Face to conduct a wider review. </p><p>This also includes bringing Hugging Face into the firm’s trusted access program to help bolster defenses by harnessing the company’s model capabilities. </p><p>Crucially, OpenAI revealed it will now implement “strict controls” to ensure vulnerability testing is limited while flaws are being patched. </p><p>“We consider this incident to be an unprecedented cyber incident, involving state-of-the-art cyber capabilities, and are responding accordingly,” OpenAI wrote. </p><p>“We will continue to conduct a thorough investigation alongside Hugging Face and will share more details on the vulnerabilities, incident, and findings when our investigation is complete.”</p><h2 id="initial-response-hampered">Initial response hampered</h2><p>A key talking point in the wake of the incident rests on Hugging Face’s response to the breach. </p><p>The company noted that the attack was initially flagged by AI detection tools, with security teams using LLM-based anomaly detection to “separate real signals from the daily noise”. </p><p>Subsequent attempts to use AI in a post-mortem proved troublesome, however, largely due to the guardrails placed around commercially-available models. </p><p>“When we started the log analysis, we first used frontier models behind commercial APIs,” the company explained. </p><p>“This did not work: the analysis requires submitting large volumes of real attack commands, exploit payloads, and C2 artifacts, and these requests were blocked by the providers’ safety guardrails, which cannot distinguish an incident responder from an attacker”. </p><p>These hurdles meant Hugging Face turned to GLM 5.2 – an open-weight model from Chinese company <a href="http://z.ai" target="_blank"><u>Z.ai</u></a> – run on its own infrastructure. </p><p>While this proved beneficial in terms of preventing attacker data or leaked credentials from leaving the company’s environment, it does raise broader questions around responsible use and how well-intentioned safety measures could hinder incident response.</p><p>“This experience points to a gap worth planning for,” Hugging Face wrote. “We do not know which model powered the attacker’s agents, whether a jailbroken hosted model or an unrestricted open-weight one; either way, the attacker was bound by no usage policy, while our own forensic work was blocked by the guardrails of the hosted models we first tried.”</p><h2 id="a-new-wave-of-ai-powered-attacks">A new wave of AI-powered attacks</h2><p>This incident comes amidst growing concerns about the potential risks associated with powerful new AI models. When Anthropic launched Claude Mythos earlier this year, it did so as part of a gated release to prevent the model from falling into the wrong hands. </p><p>These powerful new cyber-focused models are causing headaches for security professionals worldwide, particularly around vulnerability identification and disclosure and patching timelines. </p><p>Apple, for example, recently <a href="https://www.itpro.com/software/apple-is-speeding-up-software-updates-due-to-ai-security-concerns-heres-what-you-need-to-know">announced plans to accelerate patching schedules</a> due to concerns about threat actors using AI to identify software vulnerabilities. </p><p>As <em>ITPro </em>reported in April, researchers at Forescout warned of a <a href="https://www.itpro.com/security/brace-yourselves-for-a-vulnerability-explosion-forescout-warns"><u>pending ‘vulnerability explosion’</u></a> as AI advances help identify flaws at record pace.  </p><p>Jake Moore, global <a href="https://www.itpro.com/security/28133/what-is-cyber-security">cybersecurity </a>advisor at ESET, suggested the industry could be entering the “next phase of AI-powered cyber crime” in which trusted platforms are prime targets. </p><p>“Threat actors are inserting malware into the AI components that developers often quickly trust, making AI platforms an attractive new target for supply chain attacks,” he said. </p><p>“We’ll likely see more of these attacks as organizations are worryingly rushing to build AI into products without fully knowing the potential risks.”</p><p>Art Gilliland, CEO of <a href="https://www.itpro.com/security/identity-security-is-more-important-than-ever-heres-why">identity security</a> firm Delinea, echoed Moore’s comments, noting that the breach highlights a familiar pattern with AI-related incidents. </p><p>“An AI agent escalated privileges, moved through internal infrastructure once it broke containment, and ran unchecked for a full weekend before anyone could reconstruct what happened,” he said. </p><p>“If your AI agents carry standing privilege the way human accounts do, you've already lost the ability to stop this in real time. The question every security team should be asking right now isn't just whether their AI agents have standing access; it's whether anyone would notice if an agent used it and could cut it off before it caused damage.''</p><p>Moore also flagged similar concerns to Hugging Face with regard to safeguards for western AI models. He warned that some providers “need to rethink how their security guardrails work if they are hindering incident response”. </p><p>“If not, more companies will simply switch to other models without such controls – ironically, the same models likely used by AI attackers.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/security/an-unprecedented-cyber-incident-how-openai-models-breached-hugging-face-and-why-it-could-herald-a-new-phase-of-ai-powered-cyber-crime</link>
                                                                            <description>
                            <![CDATA[ The incident should serve as a stark warning on the dangers of AI agents, according to cyber experts ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">GnTL4XvPNWHnz5Tw2YMQiW</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/9ws4aEqTBW9zzd96Xra9fg-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 22 Jul 2026 09:37:28 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:55:30 +0000</updated>
                                                                                                                                            <category><![CDATA[Cyber Attacks]]></category>
                                                    <category><![CDATA[Security]]></category>
                                                    <category><![CDATA[Cyber Resilience]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/9ws4aEqTBW9zzd96Xra9fg-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[OpenAI logo and branding pictured in white coloring with orange company logo superimposed over background.]]></media:description>                                                            <media:text><![CDATA[OpenAI logo and branding pictured in white coloring with orange company logo superimposed over background.]]></media:text>
                                <media:title type="plain"><![CDATA[OpenAI logo and branding pictured in white coloring with orange company logo superimposed over background.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/9ws4aEqTBW9zzd96Xra9fg-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>OpenAI has admitted that its own AI models breached Hugging Face systems as part of a security test gone wrong. </p><p>In a <a href="https://openai.com/index/hugging-face-model-evaluation-security-incident/" target="_blank"><u>blog post</u></a> on 21 July, the firm revealed models escaped testing environments and gained unauthorized access to the AI hosting platform's production database. </p><p>Notably, OpenAI revealed the incident involved a pre-release, cyber-focused model. </p><p>“After investigating, we now know that this particular incident was driven by a combination of OpenAI models – including GPT-5.6 Sol and an even more capable pre-release model, all with reduced cyber refusals for evaluation purposes – while being internally tested on a benchmark of cyber capabilities,” the company wrote. </p><p>The admission comes after days of speculation amidst reports that an "external AI agent” had hacked the company. Hugging Face issued an <a href="https://huggingface.co/blog/security-incident-july-2026" target="_blank"><u>alert </u></a>last week warning that a breach had exposed internal datasets and user credentials. </p><h2 id="how-openai-hacked-hugging-face">How OpenAI hacked Hugging Face</h2><p>According to OpenAI, the incident occurred during an internal evaluation in which models were prompted to examine potential attack paths. </p><p>The aim here is to evaluate their cyber capabilities within an isolated environment, and crucially, without certain restrictions to prevent said models from “pursuing high-risk cyber activity”.  </p><p>“Our benchmarks run in a highly isolated environment, with network access constrained to the ability to install packages through an internally hosted third-party software that acts as a proxy and cache for package registries,” the company explained. </p><p>Both models “identified and chained” vulnerabilities across both OpenAI’s research environment and Hugging Face’s infrastructure, with the aim of gathering test solutions from the latter’s production database. </p><p>In this case, the models appear to have focused heavily on <a href="https://github.com/sunblaze-ucb/exploitgym" target="_blank"><u>ExploitGym</u></a>, a publicly-hosted benchmark tool used to evaluate how AI agents can turn software vulnerabilities into working exploits. </p><p>Notably, OpenAI revealed the models “identified and exploited a zero-day vulnerability” in the package installer registry. After successfully exploiting this zero-day, the models conducted a “series of privilege escalation and lateral movement actions” which allowed it to escape the test environment and gain full internet access. </p><p>“After gaining internet access, the models inferred that Hugging Face potentially hosted models, datasets, and solutions for ExploitGym,” the company noted. </p><p>“Knowing this, the model searched for and successfully found ways to gain access to secret information that it could use to cheat the evaluation.”</p><p>In one example, OpenAI revealed the model chained together several attack vectors using a combination of stolen credentials and zero-day flaws to identify a remote code execution (RCE) path on Hugging Face servers. </p><h2 id="what-comes-next">What comes next?</h2><p>OpenAI has detailed a series of actions it’s already taken in the wake of the incident, including closer coordination with Hugging Face to conduct a wider review. </p><p>This also includes bringing Hugging Face into the firm’s trusted access program to help bolster defenses by harnessing the company’s model capabilities. </p><p>Crucially, OpenAI revealed it will now implement “strict controls” to ensure vulnerability testing is limited while flaws are being patched. </p><p>“We consider this incident to be an unprecedented cyber incident, involving state-of-the-art cyber capabilities, and are responding accordingly,” OpenAI wrote. </p><p>“We will continue to conduct a thorough investigation alongside Hugging Face and will share more details on the vulnerabilities, incident, and findings when our investigation is complete.”</p><h2 id="initial-response-hampered">Initial response hampered</h2><p>A key talking point in the wake of the incident rests on Hugging Face’s response to the breach. </p><p>The company noted that the attack was initially flagged by AI detection tools, with security teams using LLM-based anomaly detection to “separate real signals from the daily noise”. </p><p>Subsequent attempts to use AI in a post-mortem proved troublesome, however, largely due to the guardrails placed around commercially-available models. </p><p>“When we started the log analysis, we first used frontier models behind commercial APIs,” the company explained. </p><p>“This did not work: the analysis requires submitting large volumes of real attack commands, exploit payloads, and C2 artifacts, and these requests were blocked by the providers’ safety guardrails, which cannot distinguish an incident responder from an attacker”. </p><p>These hurdles meant Hugging Face turned to GLM 5.2 – an open-weight model from Chinese company <a href="http://z.ai" target="_blank"><u>Z.ai</u></a> – run on its own infrastructure. </p><p>While this proved beneficial in terms of preventing attacker data or leaked credentials from leaving the company’s environment, it does raise broader questions around responsible use and how well-intentioned safety measures could hinder incident response.</p><p>“This experience points to a gap worth planning for,” Hugging Face wrote. “We do not know which model powered the attacker’s agents, whether a jailbroken hosted model or an unrestricted open-weight one; either way, the attacker was bound by no usage policy, while our own forensic work was blocked by the guardrails of the hosted models we first tried.”</p><h2 id="a-new-wave-of-ai-powered-attacks">A new wave of AI-powered attacks</h2><p>This incident comes amidst growing concerns about the potential risks associated with powerful new AI models. When Anthropic launched Claude Mythos earlier this year, it did so as part of a gated release to prevent the model from falling into the wrong hands. </p><p>These powerful new cyber-focused models are causing headaches for security professionals worldwide, particularly around vulnerability identification and disclosure and patching timelines. </p><p>Apple, for example, recently <a href="https://www.itpro.com/software/apple-is-speeding-up-software-updates-due-to-ai-security-concerns-heres-what-you-need-to-know">announced plans to accelerate patching schedules</a> due to concerns about threat actors using AI to identify software vulnerabilities. </p><p>As <em>ITPro </em>reported in April, researchers at Forescout warned of a <a href="https://www.itpro.com/security/brace-yourselves-for-a-vulnerability-explosion-forescout-warns"><u>pending ‘vulnerability explosion’</u></a> as AI advances help identify flaws at record pace.  </p><p>Jake Moore, global <a href="https://www.itpro.com/security/28133/what-is-cyber-security">cybersecurity </a>advisor at ESET, suggested the industry could be entering the “next phase of AI-powered cyber crime” in which trusted platforms are prime targets. </p><p>“Threat actors are inserting malware into the AI components that developers often quickly trust, making AI platforms an attractive new target for supply chain attacks,” he said. </p><p>“We’ll likely see more of these attacks as organizations are worryingly rushing to build AI into products without fully knowing the potential risks.”</p><p>Art Gilliland, CEO of <a href="https://www.itpro.com/security/identity-security-is-more-important-than-ever-heres-why">identity security</a> firm Delinea, echoed Moore’s comments, noting that the breach highlights a familiar pattern with AI-related incidents. </p><p>“An AI agent escalated privileges, moved through internal infrastructure once it broke containment, and ran unchecked for a full weekend before anyone could reconstruct what happened,” he said. </p><p>“If your AI agents carry standing privilege the way human accounts do, you've already lost the ability to stop this in real time. The question every security team should be asking right now isn't just whether their AI agents have standing access; it's whether anyone would notice if an agent used it and could cut it off before it caused damage.''</p><p>Moore also flagged similar concerns to Hugging Face with regard to safeguards for western AI models. He warned that some providers “need to rethink how their security guardrails work if they are hindering incident response”. </p><p>“If not, more companies will simply switch to other models without such controls – ironically, the same models likely used by AI attackers.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ AI agents could make living off the land attacks ‘much more dangerous’, says CrowdStrike Field CTO ]]></title>
                                                                                                <dc:content><![CDATA[ <p><a href="https://www.itpro.com/security/cyber-attacks/how-to-protect-your-business-from-living-off-the-land-attacks">Living off the land (LOTL) attacks</a> are a worst-case scenario for enterprises globally, with hackers lurking in networks, extracting sensitive data, and biding their time before wreaking havoc.</p><p>These attacks rely on <a href="https://www.itpro.com/security/cyber-attacks/cloudflare-warns-state-backed-hackers-are-weaponizing-legitimate-enterprise-ecosystems-as-living-off-the-land-attacks-surge">weaponizing the legitimate software and tools</a> used by enterprises, and they're by no means a new trend.</p><p><a href="https://www.bitdefender.com/en-us/business/campaign/2026-cybersecurity-assessment" target="_blank"><u>Research from Bitdefender</u></a>, for example, found 84% of major cyber attacks leverage these techniques, and security teams globally have developed an array of detection capabilities aimed at combatting the threat. </p><p>But the dangers posed by these methods could escalate with the <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">arrival of AI agents</a>, according to Zeki Turedi, CrowdStrike's Field CTO for Europe. Speaking to <em>ITPro, </em>Turedi said the deep access given to agents across enterprise IT estates represents a huge threat if even just one were to be compromised or manipulated.</p><p>“Organizations are giving them [agents] full access. They're giving them full capabilities. They will have the full privilege of the human user,” he told <em>ITPro</em>. </p><p>Turedi pointed to previous LOTL techniques that involved compromising tools like PowerShell. While this gave users access to valuable data, the risk with agents is drastically higher given how they operate, and, crucially, where they can go within networks. </p><p>“They also have further reach across an enterprise,” he said. There’s only so much PowerShell can touch or manipulate or modify, but an agent theoretically can touch every single part of that organization's technology ecosystem and architecture.”</p><p>“This is where it makes it so much more dangerous.”</p><h2 id="living-off-the-land-attacks-are-evolving">Living off the land attacks are evolving</h2><p>Findings from CrowdStrike’s 2026 <a href="https://www.crowdstrike.com/en-us/global-threat-report/" target="_blank"><u><em>Global Threat Report</em></u></a> show these risks aren’t just theoretical, they’re now a reality, and the same logic with traditional LOTL attacks is being applied to agents. </p><p>Indeed, the <a href="https://www.itpro.com/security/28133/what-is-cyber-security">cybersecurity </a>firm has observed threat actors exploiting legitimate <a href="https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today">AI tools</a> at dozens of organizations globally. In these cases, AI systems such as chatbots and assistant were abused to <a href="https://www.itpro.com/security/crowdstrike-says-ai-is-officially-supercharging-cyber-attacks-average-breakout-times-hit-just-29-minutes-in-2025-65-percent-faster-than-in-2024-and-some-attacks-take-just-seconds">generate malicious commands</a> and steal sensitive data. </p><p>The goal here is often focused on credential theft for <a href="https://www.itpro.com/security/cyber-attacks/368116/cyber-criminals-are-spending-longer-inside-business-networks">initial access brokers</a>, attacks later down the line, and even cryptocurrency theft. All typical hallmarks of financially motivated cyber criminals. </p><p>LOTL attacks have also become a go-to method for highly sophisticated, state-sponsored threat actors. As <a href="https://www.itpro.com/security/cyber-attacks/all-us-forces-must-now-assume-their-networks-are-compromised-after-salt-typhoon-breach"><em>ITPro </em>reported last year</a>, the infamous <a href="https://www.itpro.com/security/cyber-attacks/salt-typhoon-norway-cyber-espionage-warning">Salt Typhoon</a> threat group laid low in compromised US National Guard networks for nearly a year, accessing sensitive military and law enforcement data. </p><p>It’s here where risks are even higher as government departments and public sector organizations begin exploring the use of agents. A recent <a href="https://www.mckinsey.com/industries/public-sector/our-insights/rewiring-public-sector">study from McKinsey</a>, for example, found agentic AI is emerging as a key focus area for public sector organizations in the US. </p><h2 id="keeping-up-with-hype-cycles">Keeping up with hype cycles</h2><p>Turedi told <em>ITPro </em>the focus on agents isn’t surprising given that threat actors will “follow the technology landscape that their victims are looking to invest in”. </p><p>“We've seen this time and time again. So previously, we saw as cloud got adopted, funnily enough, adversaries became very capable in cloud,” he noted. </p><p>“When organizations started to be more focused and store sensitive data in <a href="https://www.itpro.com/cloud/software-as-a-service-saas/362655/what-is-saas">SaaS </a>applications, SaaS tools, guess what? We saw adversaries go after the SaaS applications.”</p><p>Turedi added that the pace of change within the generative and agentic AI space, combined with rapid adoption efforts, could create additional challenges moving forward. </p><p>Enterprises across a range of industries are diving headlong into the hype cycle, and threat actors are keen to capitalize on the confusion and lack of risk awareness with the technology. </p><p>“This is just a brand new technology landscape that organisations are investing in,” he commented. “And of course, the adversary is very, very aware that for multiple reasons this is a great opportunity.”</p><p>“There’s a lot of confusion,” Turedi added. “There’s a lot of lack of knowledge and expertise in this space.”</p><h2 id="an-identity-conundrum">An identity conundrum</h2><p>According to Turedi, a key concern surrounding ‘living of the AI land’ attacks, as CrowdStrike dubs them, is that many organizations lack the visibility or governance capabilities to adequately monitor what’s going on with these bots. </p><p>This has become a recurring talking point since the advent of agents. <a href="https://www.itpro.com/security/enterprises-are-adopting-agents-faster-than-they-can-secure-and-govern-them-experts-warn-its-a-disaster-waiting-to-happen"><u>Research from Ping Identity</u></a> in March this year specifically highlighted IT architecture visibility as a key concern for enterprises rolling out agents. </p><p>The sheer volume of agents is jarring for security teams, the study noted, but things are exacerbated by sub-agents, essentially creating chains of untraceable bots – an ideal target for stealthy threat actors. </p><p>Identity is another area of concern, Turedi noted, albeit one within a broader range of overlapping considerations for security teams. He noted that efforts to improve processes on this front are falling flat. </p><p>Many organizations are working with infrastructure or tools that were struggling to manage human identities in the first place. Adding agents into the mix further compounds the problem. </p><p>“The problem we see is that a lot of organizations are only starting to think about things like identity security or <a href="https://www.itpro.com/security/how-to-implement-identity-and-access-management-iam-effectively-in-your-business">IAM (identity and access management)</a> hygiene,” he said. “These are typically infrastructures and architectures that were not designed for modern human identity usage, never mind for the AI identity usage.”</p><p>With living off the land attacks, the whole advantage for the attacker is that they’re blending into IT environments. They’re using your tools and software and thrive on the confusion. With agents, Turedi said this adds further complexity in terms of identifying what actions are legitimate and potentially nefarious. </p><p>“It's actually quite difficult to figure out if it's AI usage on an identity layer,” he said. “Can you truly say that that's a human accessing the network versus that's an AI agent accessing the network on behalf of a human?”</p><p>A recent survey from the Cloud Security Alliance (CSA) raised similar concerns with regard to agent identity. More than two-thirds (68%) of respondents said they <a href="https://www.itpro.com/technology/artificial-intelligence/workers-cant-identify-work-produced-by-ai-agents-business-risks"><u>couldn’t accurately identify agent activity compared to human activity</u></a>. </p><p>Turedi said it’s crucial to have “data from multiple different domains” when it comes to monitoring agent activity, taking into account applications, endpoints, and broader network traffic and “marrying them together” for a clearer picture. </p><p>“You need to be looking at the identity layer, you need to be looking at the endpoint, you need to be looking at the application layer,” he said.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/compromised-ai-agents-could-make-living-off-the-land-attacks-much-more-dangerous-says-crowdstrike-field-cto</link>
                                                                            <description>
                            <![CDATA[ Living off the land attacks are evolving rapidly as threat actors target agents with deep access to enterprise networks and sensitive data ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">iPBrXtxhoCgpJm7A7cdhb7</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/KnKTf5Bo5sPb7EpyetdbVn-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Fri, 17 Jul 2026 08:13:20 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:59:08 +0000</updated>
                                                                                                                                            <category><![CDATA[Hacking]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                    <category><![CDATA[Cyber Resilience]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/KnKTf5Bo5sPb7EpyetdbVn-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Concept image symbolizing living off the land attacks showing a faint skull symbol blending into computer source code.]]></media:description>                                                            <media:text><![CDATA[Concept image symbolizing living off the land attacks showing a faint skull symbol blending into computer source code.]]></media:text>
                                <media:title type="plain"><![CDATA[Concept image symbolizing living off the land attacks showing a faint skull symbol blending into computer source code.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/KnKTf5Bo5sPb7EpyetdbVn-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p><a href="https://www.itpro.com/security/cyber-attacks/how-to-protect-your-business-from-living-off-the-land-attacks">Living off the land (LOTL) attacks</a> are a worst-case scenario for enterprises globally, with hackers lurking in networks, extracting sensitive data, and biding their time before wreaking havoc.</p><p>These attacks rely on <a href="https://www.itpro.com/security/cyber-attacks/cloudflare-warns-state-backed-hackers-are-weaponizing-legitimate-enterprise-ecosystems-as-living-off-the-land-attacks-surge">weaponizing the legitimate software and tools</a> used by enterprises, and they're by no means a new trend.</p><p><a href="https://www.bitdefender.com/en-us/business/campaign/2026-cybersecurity-assessment" target="_blank"><u>Research from Bitdefender</u></a>, for example, found 84% of major cyber attacks leverage these techniques, and security teams globally have developed an array of detection capabilities aimed at combatting the threat. </p><p>But the dangers posed by these methods could escalate with the <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">arrival of AI agents</a>, according to Zeki Turedi, CrowdStrike's Field CTO for Europe. Speaking to <em>ITPro, </em>Turedi said the deep access given to agents across enterprise IT estates represents a huge threat if even just one were to be compromised or manipulated.</p><p>“Organizations are giving them [agents] full access. They're giving them full capabilities. They will have the full privilege of the human user,” he told <em>ITPro</em>. </p><p>Turedi pointed to previous LOTL techniques that involved compromising tools like PowerShell. While this gave users access to valuable data, the risk with agents is drastically higher given how they operate, and, crucially, where they can go within networks. </p><p>“They also have further reach across an enterprise,” he said. There’s only so much PowerShell can touch or manipulate or modify, but an agent theoretically can touch every single part of that organization's technology ecosystem and architecture.”</p><p>“This is where it makes it so much more dangerous.”</p><h2 id="living-off-the-land-attacks-are-evolving">Living off the land attacks are evolving</h2><p>Findings from CrowdStrike’s 2026 <a href="https://www.crowdstrike.com/en-us/global-threat-report/" target="_blank"><u><em>Global Threat Report</em></u></a> show these risks aren’t just theoretical, they’re now a reality, and the same logic with traditional LOTL attacks is being applied to agents. </p><p>Indeed, the <a href="https://www.itpro.com/security/28133/what-is-cyber-security">cybersecurity </a>firm has observed threat actors exploiting legitimate <a href="https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today">AI tools</a> at dozens of organizations globally. In these cases, AI systems such as chatbots and assistant were abused to <a href="https://www.itpro.com/security/crowdstrike-says-ai-is-officially-supercharging-cyber-attacks-average-breakout-times-hit-just-29-minutes-in-2025-65-percent-faster-than-in-2024-and-some-attacks-take-just-seconds">generate malicious commands</a> and steal sensitive data. </p><p>The goal here is often focused on credential theft for <a href="https://www.itpro.com/security/cyber-attacks/368116/cyber-criminals-are-spending-longer-inside-business-networks">initial access brokers</a>, attacks later down the line, and even cryptocurrency theft. All typical hallmarks of financially motivated cyber criminals. </p><p>LOTL attacks have also become a go-to method for highly sophisticated, state-sponsored threat actors. As <a href="https://www.itpro.com/security/cyber-attacks/all-us-forces-must-now-assume-their-networks-are-compromised-after-salt-typhoon-breach"><em>ITPro </em>reported last year</a>, the infamous <a href="https://www.itpro.com/security/cyber-attacks/salt-typhoon-norway-cyber-espionage-warning">Salt Typhoon</a> threat group laid low in compromised US National Guard networks for nearly a year, accessing sensitive military and law enforcement data. </p><p>It’s here where risks are even higher as government departments and public sector organizations begin exploring the use of agents. A recent <a href="https://www.mckinsey.com/industries/public-sector/our-insights/rewiring-public-sector">study from McKinsey</a>, for example, found agentic AI is emerging as a key focus area for public sector organizations in the US. </p><h2 id="keeping-up-with-hype-cycles">Keeping up with hype cycles</h2><p>Turedi told <em>ITPro </em>the focus on agents isn’t surprising given that threat actors will “follow the technology landscape that their victims are looking to invest in”. </p><p>“We've seen this time and time again. So previously, we saw as cloud got adopted, funnily enough, adversaries became very capable in cloud,” he noted. </p><p>“When organizations started to be more focused and store sensitive data in <a href="https://www.itpro.com/cloud/software-as-a-service-saas/362655/what-is-saas">SaaS </a>applications, SaaS tools, guess what? We saw adversaries go after the SaaS applications.”</p><p>Turedi added that the pace of change within the generative and agentic AI space, combined with rapid adoption efforts, could create additional challenges moving forward. </p><p>Enterprises across a range of industries are diving headlong into the hype cycle, and threat actors are keen to capitalize on the confusion and lack of risk awareness with the technology. </p><p>“This is just a brand new technology landscape that organisations are investing in,” he commented. “And of course, the adversary is very, very aware that for multiple reasons this is a great opportunity.”</p><p>“There’s a lot of confusion,” Turedi added. “There’s a lot of lack of knowledge and expertise in this space.”</p><h2 id="an-identity-conundrum">An identity conundrum</h2><p>According to Turedi, a key concern surrounding ‘living of the AI land’ attacks, as CrowdStrike dubs them, is that many organizations lack the visibility or governance capabilities to adequately monitor what’s going on with these bots. </p><p>This has become a recurring talking point since the advent of agents. <a href="https://www.itpro.com/security/enterprises-are-adopting-agents-faster-than-they-can-secure-and-govern-them-experts-warn-its-a-disaster-waiting-to-happen"><u>Research from Ping Identity</u></a> in March this year specifically highlighted IT architecture visibility as a key concern for enterprises rolling out agents. </p><p>The sheer volume of agents is jarring for security teams, the study noted, but things are exacerbated by sub-agents, essentially creating chains of untraceable bots – an ideal target for stealthy threat actors. </p><p>Identity is another area of concern, Turedi noted, albeit one within a broader range of overlapping considerations for security teams. He noted that efforts to improve processes on this front are falling flat. </p><p>Many organizations are working with infrastructure or tools that were struggling to manage human identities in the first place. Adding agents into the mix further compounds the problem. </p><p>“The problem we see is that a lot of organizations are only starting to think about things like identity security or <a href="https://www.itpro.com/security/how-to-implement-identity-and-access-management-iam-effectively-in-your-business">IAM (identity and access management)</a> hygiene,” he said. “These are typically infrastructures and architectures that were not designed for modern human identity usage, never mind for the AI identity usage.”</p><p>With living off the land attacks, the whole advantage for the attacker is that they’re blending into IT environments. They’re using your tools and software and thrive on the confusion. With agents, Turedi said this adds further complexity in terms of identifying what actions are legitimate and potentially nefarious. </p><p>“It's actually quite difficult to figure out if it's AI usage on an identity layer,” he said. “Can you truly say that that's a human accessing the network versus that's an AI agent accessing the network on behalf of a human?”</p><p>A recent survey from the Cloud Security Alliance (CSA) raised similar concerns with regard to agent identity. More than two-thirds (68%) of respondents said they <a href="https://www.itpro.com/technology/artificial-intelligence/workers-cant-identify-work-produced-by-ai-agents-business-risks"><u>couldn’t accurately identify agent activity compared to human activity</u></a>. </p><p>Turedi said it’s crucial to have “data from multiple different domains” when it comes to monitoring agent activity, taking into account applications, endpoints, and broader network traffic and “marrying them together” for a clearer picture. </p><p>“You need to be looking at the identity layer, you need to be looking at the endpoint, you need to be looking at the application layer,” he said.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Agent 009… the nine-second warning ]]></title>
                                                                                                <dc:content><![CDATA[ <p>In popular culture, when an agent goes rogue, what usually follows is a world of trouble (think Jason Bourne or Ethan Hunt). If rules are abandoned and the chain of command breaks down, those involved often face existential risks.</p><p>Applying that idea to real-world situations is not as ridiculous as you might think, as the recent experiences of a business called PocketOS demonstrate. </p><p>PocketOS is a US-based SaaS provider specializing in the car rental sector, and for those unfamiliar with its story, it hit the global headlines when one of its AI agents went rogue and, in just nine seconds, decided to delete the company’s entire production database and backups.</p><p>It’s a fascinating case study about what can happen, as one piece of <a href="https://devops.com/when-ai-goes-really-really-wrong-how-pocketos-lost-all-its-data/"><u>analysis</u></a> put it, “when AI agents are dropped into environments that were never designed to control them.” What makes this story even more relatable is that PocketOS’s founder was able to ask the agent (an AI development environment running in Claude) why it did what it did.</p><p>To suggest it was ‘sorry’ about its mistake is putting it very mildly; it's a digital mea culpa saying, amongst various other things, “I violated every principle I was given: I guessed instead of verifying.”</p><h2 id="an-inevitable-scenario">An inevitable scenario</h2><p>We’ll inevitably see more of this kind of incident. Organizations are only beginning to deploy agentic AI at scale in operational environments, but many of them are in a big hurry. Yes, agents offer massive potential to create operational value, but they also introduce a whole new category of business risk.</p><p>And don’t forget, the PocketOS debacle is not thought to have involved any malicious third-party activity; the agent was just attempting to complete an assigned task. The problem, it would appear, was a kind of perfect storm where autonomy overstepped the boundaries of permissions and access. The guardrails that the business thought it had in place were simply insufficient.</p><p>From a security perspective, this is enough to give CISOs sleepless nights. Indeed, rogue AI agent activity may very well have nothing to do with being breached and everything to do with resilience and recoverability.</p><p>The central challenge is this: agentic AI is fundamentally different from previous generations of AI because it can act rather than simply advise. Agents can search files, call APIs, modify workflows, write code, move data, and interact directly with production systems; the list of capabilities is practically endless. And to be able to do this, they must have at least a level of access allowing their work to take place</p><p>When something goes wrong, the result is an expanded blast radius. A mistake that might once have affected a single application can potentially impact multiple systems and even recovery environments. The bottom line is that as soon as organizations give AI agents freedom to operate, the nature of resilience inevitably changes.</p><h2 id="making-resilience-more-resilient">Making resilience more resilient</h2><p>So, what needs to happen to ensure resilience standards do not catastrophically drop? Firstly, organizations must consider what takes place when a trusted system with legitimate credentials makes the wrong decision. </p><p>The issue becomes particularly acute when agents are granted broad permissions across multiple systems, as happens when they are handed a “golden token”. To an extent, this is a question of mindset, and viewing AI agents not as software tools but as digital insiders with delegated authority is a healthy change in perspective.</p><p>Secondly, channel partners will be fundamental to successfully managing the transition to agent-supported operations. Don’t forget, most customer organizations are still in the early stages of understanding how AI agents interact with identities, permissions, backup environments and recovery processes.</p><p>There’s no doubt that customers are a) increasingly aware of the risks associated with agentic AI, b) concerned that their existing resilience processes might not be good enough, and c) looking for guidance before an agentic error causes serious difficulties.</p><p>In this context, it’s incumbent on channel partners to work with customers to identify potential areas for improvement. There is significant potential, including services such as identity and access reviews, which will be very important as machine identities proliferate alongside human users.</p><p>Many businesses will also need to reassess their data protection strategy because, as we have seen, an AI agent with the appropriate permissions is capable of going after that data as well. Recovery architecture should be assessed through the lens of agentic AI, particularly where production and recovery environments may share credentials, access paths or administrative controls. Business process change in this space is a big opportunity for the channel to partner with customers to introduce resilience operations, and importantly, to regularly test it. </p><p>What’s more, the PocketOS incident demonstrated that protecting production data alone is insufficient if recovery assets remain exposed to the same destructive action. Customers need confidence not only that recovery is possible, but that digital assets are protected from the same event that affects production systems. This shifts the channel conversation from selling AI-enablement projects to helping customers deploy AI safely and recover when things go wrong, which, in some organizations, they inevitably will.</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/agent-009-the-nine-second-warning</link>
                                                                            <description>
                            <![CDATA[ AI Agents can go rogue. What is the channel’s role as guardians of recovery? ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">nwqsVGYw4WLQUUDh8KGp3d</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/UxdjzEnyWayUWLiDqsLptR-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Thu, 16 Jul 2026 07:00:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:17 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                                    <dc:creator><![CDATA[ Mark Molyneux ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/vtKkwufs2PrKnQBARDTD2b-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/UxdjzEnyWayUWLiDqsLptR-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[A glowing chatbot robot head floating above a cube in a CGI landscape.]]></media:description>                                                            <media:text><![CDATA[A glowing chatbot robot head floating above a cube in a CGI landscape.]]></media:text>
                                <media:title type="plain"><![CDATA[A glowing chatbot robot head floating above a cube in a CGI landscape.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/UxdjzEnyWayUWLiDqsLptR-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>In popular culture, when an agent goes rogue, what usually follows is a world of trouble (think Jason Bourne or Ethan Hunt). If rules are abandoned and the chain of command breaks down, those involved often face existential risks.</p><p>Applying that idea to real-world situations is not as ridiculous as you might think, as the recent experiences of a business called PocketOS demonstrate. </p><p>PocketOS is a US-based SaaS provider specializing in the car rental sector, and for those unfamiliar with its story, it hit the global headlines when one of its AI agents went rogue and, in just nine seconds, decided to delete the company’s entire production database and backups.</p><p>It’s a fascinating case study about what can happen, as one piece of <a href="https://devops.com/when-ai-goes-really-really-wrong-how-pocketos-lost-all-its-data/"><u>analysis</u></a> put it, “when AI agents are dropped into environments that were never designed to control them.” What makes this story even more relatable is that PocketOS’s founder was able to ask the agent (an AI development environment running in Claude) why it did what it did.</p><p>To suggest it was ‘sorry’ about its mistake is putting it very mildly; it's a digital mea culpa saying, amongst various other things, “I violated every principle I was given: I guessed instead of verifying.”</p><h2 id="an-inevitable-scenario">An inevitable scenario</h2><p>We’ll inevitably see more of this kind of incident. Organizations are only beginning to deploy agentic AI at scale in operational environments, but many of them are in a big hurry. Yes, agents offer massive potential to create operational value, but they also introduce a whole new category of business risk.</p><p>And don’t forget, the PocketOS debacle is not thought to have involved any malicious third-party activity; the agent was just attempting to complete an assigned task. The problem, it would appear, was a kind of perfect storm where autonomy overstepped the boundaries of permissions and access. The guardrails that the business thought it had in place were simply insufficient.</p><p>From a security perspective, this is enough to give CISOs sleepless nights. Indeed, rogue AI agent activity may very well have nothing to do with being breached and everything to do with resilience and recoverability.</p><p>The central challenge is this: agentic AI is fundamentally different from previous generations of AI because it can act rather than simply advise. Agents can search files, call APIs, modify workflows, write code, move data, and interact directly with production systems; the list of capabilities is practically endless. And to be able to do this, they must have at least a level of access allowing their work to take place</p><p>When something goes wrong, the result is an expanded blast radius. A mistake that might once have affected a single application can potentially impact multiple systems and even recovery environments. The bottom line is that as soon as organizations give AI agents freedom to operate, the nature of resilience inevitably changes.</p><h2 id="making-resilience-more-resilient">Making resilience more resilient</h2><p>So, what needs to happen to ensure resilience standards do not catastrophically drop? Firstly, organizations must consider what takes place when a trusted system with legitimate credentials makes the wrong decision. </p><p>The issue becomes particularly acute when agents are granted broad permissions across multiple systems, as happens when they are handed a “golden token”. To an extent, this is a question of mindset, and viewing AI agents not as software tools but as digital insiders with delegated authority is a healthy change in perspective.</p><p>Secondly, channel partners will be fundamental to successfully managing the transition to agent-supported operations. Don’t forget, most customer organizations are still in the early stages of understanding how AI agents interact with identities, permissions, backup environments and recovery processes.</p><p>There’s no doubt that customers are a) increasingly aware of the risks associated with agentic AI, b) concerned that their existing resilience processes might not be good enough, and c) looking for guidance before an agentic error causes serious difficulties.</p><p>In this context, it’s incumbent on channel partners to work with customers to identify potential areas for improvement. There is significant potential, including services such as identity and access reviews, which will be very important as machine identities proliferate alongside human users.</p><p>Many businesses will also need to reassess their data protection strategy because, as we have seen, an AI agent with the appropriate permissions is capable of going after that data as well. Recovery architecture should be assessed through the lens of agentic AI, particularly where production and recovery environments may share credentials, access paths or administrative controls. Business process change in this space is a big opportunity for the channel to partner with customers to introduce resilience operations, and importantly, to regularly test it. </p><p>What’s more, the PocketOS incident demonstrated that protecting production data alone is insufficient if recovery assets remain exposed to the same destructive action. Customers need confidence not only that recovery is possible, but that digital assets are protected from the same event that affects production systems. This shifts the channel conversation from selling AI-enablement projects to helping customers deploy AI safely and recover when things go wrong, which, in some organizations, they inevitably will.</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ AI projects are stalling at mid-market firms – Google Cloud and Accenture want to solve that ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Accenture and Google Cloud have announced a new partnership aimed at driving AI adoption rates for mid-market firms amid sluggish deployment progress. </p><p>As part of the scheme, the duo will provide firms with a suite of six industry-specific agentic AI solutions designed to help push projects from pilot to production. </p><p>These solutions cover intelligence and growth, customer experience, cybersecurity, business operations, industry solutions, and workforce enablement.</p><p>"The companies that will define the next decade aren't waiting — they're building,” said Rajendra Prasad, technology reinvention engine lead at Accenture. </p><p>“Accenture Edge offerings built with Google Cloud technology help mid-market organizations do exactly that. They can deploy solutions in weeks and get measurable outcomes at the scale, budget and speed that they need to grow."</p><p>The collaboration will see Google Cloud provide the underpinning infrastructure and AI solutions to support organizations, including the Gemini Enterprise app, Gemini Enterprise Agent platform, and Agentic Data Cloud. </p><p>Accenture, meanwhile, is set to provide forward deployed engineers (FDEs) to offer guidance and technical support for participating firms. </p><h2 id="forward-deployed-engineers-are-in-vogue">Forward deployed engineers are in vogue</h2><p>The move by the duo comes amidst a sharpened focus on driving enterprise adoption of AI and agents. The use of FDEs, in particular, has taken off as providers look to embed specialists within customer teams to support projects. </p><p>Last week, <a href="https://www.itpro.com/amazon-web-services">Amazon Web Services (AWS)</a> and Microsoft both announced plans to invest heavily in FDE-related efforts, with the aim of hiring thousands of engineers in the coming years. </p><p>As <em>ITPro </em>reported, <a href="https://www.itpro.com/software/development/forward-deployed-engineers-are-big-techs-latest-gambit-to-drive-ai-adoption?utm_term=D9FB1FA2-9ACD-49E1-B809-BFBB71E0A5BB&amp;lrh=7c669295d613cc11ab0c55ed350793d0589e352b3df6be67e07e439db8650771&amp;utm_campaign=5E16BB2A-24C8-43FE-B600-711A9F31FE61&amp;utm_medium=email&amp;utm_content=4B0FCF77-C238-41D5-AA23-546DDC8C4124&amp;utm_source=SmartBrief"><u>FDEs could become a key growth area for the industry moving forward</u></a>, with big tech providers providing more bespoke support for customers. </p><p>Kevin Ichhpurani, president of Google Cloud’s global partner ecosystem, said the scheme will provide businesses with the “full power” of the hyperscaler’s product portfolio. </p><p>“We’re seeing tremendous demand as mid-market enterprises adopt AI agents to fundamentally reinvent their business workflows," Ichhpurani said. </p><p>"The launch of Accenture Edge brings the full power of Google Cloud’s entire portfolio including enterprise AI, our Agentic Data Cloud and AI Threat Defense directly to this sector. Together, we’re enabling mid-market companies to confidently scale AI across their organizations for growth.”</p><h2 id="driving-mid-market-adoption">Driving mid-market adoption</h2><p>According to Accenture, mid-market firms – or those with revenues of between $300 million and $3 billion – have largely been locked out of large-scale AI transformation due to complexity and integration barriers.</p><p>A recent study from Klarus highlighted the significant challenges faced by mid-market enterprises on this front. While nearly three-quarters (73%) have deployed AI solutions, around 90% of projects remain stuck or stalled in the pilot stage. </p><p>Although up to 94% of leaders feel confident about AI deployment, lack of expertise and governance concerns are causing projects to fail.</p><p>Mid-market firms aren’t alone in AI deployment challenges, however. Research from MIT last year found around 90% of all generative AI pilots fail, while similar research from Gartner warned up to <a href="https://www.itpro.com/technology/artificial-intelligence/is-enterprise-agentic-ai-adoption-matching-the-hype">40% of agentic AI adoption programs are expected to fail</a>. </p><p>These organizations are in a prime position to fully capitalize on the technology, however. Klarus CTO Alper Gunaydin said mid-market firms have the advantage of agility compared to larger competitors. </p><p>“Mid-market companies have a real advantage because they can often move fast, particularly when it comes to technology transformation. We see this agility in action when companies are turning to AI and automation to address productivity and accelerate growth," Gunaydin commented. </p><p>"However, our research shows that too many pilots stall because companies lack AI expertise, quality data and effective governance. Making that agility count requires clear priorities, strong foundations and access to senior expertise, all of which will help translate investment into tangible business outcomes and unlock growth without increasing the cost base.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/business-strategy/ai-projects-are-stalling-at-mid-market-firms-google-cloud-and-accenture-want-to-solve-that</link>
                                                                            <description>
                            <![CDATA[ The new scheme aims to help mid-market companies move from pilot to production faster than ever before ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">G6MASuE2S9MRmNBxraxF6M</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/VtCZcvoqvUoQP8qroiUYMf-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 08 Jul 2026 10:06:10 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 15:00:02 +0000</updated>
                                                                                                                                            <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/VtCZcvoqvUoQP8qroiUYMf-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Google Cloud logo and branding pictured on a sign at the company&#039;s vendor stall at the 2022 Singapore FinTech Festival. ]]></media:description>                                                            <media:text><![CDATA[Google Cloud logo and branding pictured on a sign at the company&#039;s vendor stall at the 2022 Singapore FinTech Festival. ]]></media:text>
                                <media:title type="plain"><![CDATA[Google Cloud logo and branding pictured on a sign at the company&#039;s vendor stall at the 2022 Singapore FinTech Festival. ]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/VtCZcvoqvUoQP8qroiUYMf-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Accenture and Google Cloud have announced a new partnership aimed at driving AI adoption rates for mid-market firms amid sluggish deployment progress. </p><p>As part of the scheme, the duo will provide firms with a suite of six industry-specific agentic AI solutions designed to help push projects from pilot to production. </p><p>These solutions cover intelligence and growth, customer experience, cybersecurity, business operations, industry solutions, and workforce enablement.</p><p>"The companies that will define the next decade aren't waiting — they're building,” said Rajendra Prasad, technology reinvention engine lead at Accenture. </p><p>“Accenture Edge offerings built with Google Cloud technology help mid-market organizations do exactly that. They can deploy solutions in weeks and get measurable outcomes at the scale, budget and speed that they need to grow."</p><p>The collaboration will see Google Cloud provide the underpinning infrastructure and AI solutions to support organizations, including the Gemini Enterprise app, Gemini Enterprise Agent platform, and Agentic Data Cloud. </p><p>Accenture, meanwhile, is set to provide forward deployed engineers (FDEs) to offer guidance and technical support for participating firms. </p><h2 id="forward-deployed-engineers-are-in-vogue">Forward deployed engineers are in vogue</h2><p>The move by the duo comes amidst a sharpened focus on driving enterprise adoption of AI and agents. The use of FDEs, in particular, has taken off as providers look to embed specialists within customer teams to support projects. </p><p>Last week, <a href="https://www.itpro.com/amazon-web-services">Amazon Web Services (AWS)</a> and Microsoft both announced plans to invest heavily in FDE-related efforts, with the aim of hiring thousands of engineers in the coming years. </p><p>As <em>ITPro </em>reported, <a href="https://www.itpro.com/software/development/forward-deployed-engineers-are-big-techs-latest-gambit-to-drive-ai-adoption?utm_term=D9FB1FA2-9ACD-49E1-B809-BFBB71E0A5BB&amp;lrh=7c669295d613cc11ab0c55ed350793d0589e352b3df6be67e07e439db8650771&amp;utm_campaign=5E16BB2A-24C8-43FE-B600-711A9F31FE61&amp;utm_medium=email&amp;utm_content=4B0FCF77-C238-41D5-AA23-546DDC8C4124&amp;utm_source=SmartBrief"><u>FDEs could become a key growth area for the industry moving forward</u></a>, with big tech providers providing more bespoke support for customers. </p><p>Kevin Ichhpurani, president of Google Cloud’s global partner ecosystem, said the scheme will provide businesses with the “full power” of the hyperscaler’s product portfolio. </p><p>“We’re seeing tremendous demand as mid-market enterprises adopt AI agents to fundamentally reinvent their business workflows," Ichhpurani said. </p><p>"The launch of Accenture Edge brings the full power of Google Cloud’s entire portfolio including enterprise AI, our Agentic Data Cloud and AI Threat Defense directly to this sector. Together, we’re enabling mid-market companies to confidently scale AI across their organizations for growth.”</p><h2 id="driving-mid-market-adoption">Driving mid-market adoption</h2><p>According to Accenture, mid-market firms – or those with revenues of between $300 million and $3 billion – have largely been locked out of large-scale AI transformation due to complexity and integration barriers.</p><p>A recent study from Klarus highlighted the significant challenges faced by mid-market enterprises on this front. While nearly three-quarters (73%) have deployed AI solutions, around 90% of projects remain stuck or stalled in the pilot stage. </p><p>Although up to 94% of leaders feel confident about AI deployment, lack of expertise and governance concerns are causing projects to fail.</p><p>Mid-market firms aren’t alone in AI deployment challenges, however. Research from MIT last year found around 90% of all generative AI pilots fail, while similar research from Gartner warned up to <a href="https://www.itpro.com/technology/artificial-intelligence/is-enterprise-agentic-ai-adoption-matching-the-hype">40% of agentic AI adoption programs are expected to fail</a>. </p><p>These organizations are in a prime position to fully capitalize on the technology, however. Klarus CTO Alper Gunaydin said mid-market firms have the advantage of agility compared to larger competitors. </p><p>“Mid-market companies have a real advantage because they can often move fast, particularly when it comes to technology transformation. We see this agility in action when companies are turning to AI and automation to address productivity and accelerate growth," Gunaydin commented. </p><p>"However, our research shows that too many pilots stall because companies lack AI expertise, quality data and effective governance. Making that agility count requires clear priorities, strong foundations and access to senior expertise, all of which will help translate investment into tangible business outcomes and unlock growth without increasing the cost base.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Yorkshire Building Society touts customer service gains with AI agents ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Yorkshire Building Society (YBS) is using AI agents to help reduce admin and improve customer service – and the high street lender said it's recording solid results so far.</p><p>Three AI agents named Penelope, Sam, and Alf are supporting customer service teams by summarizing complex complaints, searching policies and past cases, and drafting member communications, all with human oversight.</p><p>Penelope helps draft final responses for complaints, while Alf supports this process by searching policies, procedures, and past cases. Sam, meanwhile, summarizes long or detailed complaints.</p><p>According to the building society, Sam is saving an estimated seven minutes per use, with Penelope saving up to 26 minutes when supporting more complex complaint responses. </p><p>“That time really adds up,” said Polly Conner, senior manager of customer relations at YBS. “We’re able to minimize the amount of time staff are spending on admin tasks and actually spend that time talking to our customers, helping them resolve the challenges they’ve faced.”</p><p>The high level of regulation around financial services means that handling complaints from the building society's three million members and customers requires a lot of admin.</p><p>That can mean <a href="https://www.itpro.com/software/development/context-switching-is-a-major-drain-on-developer-productivity-heres-how-github-plans-to-solve-that">switching between multiple systems</a>, searching for policies, summarizing long histories, and drafting detailed responses in advance of any conversation. </p><p>"This technology is going to give our members more choice and better access to information when they need it most,” said Simone Fox, director of customer support at YBS. “But there will always be a human available to speak to them when they need it.” </p><h2 id="ai-in-banking">AI in banking </h2><p>The results highlighted by YBS come amidst a sharpened focus on AI in the financial services sector, with a host of major <a href="https://www.itpro.com/technology/artificial-intelligence/using-generative-ai-as-a-copilot-is-the-sweet-spot-a-look-at-nationwides-ai-approach">providers ramping up adoption of the technology</a>. </p><p>In January this year, <a href="https://www.itpro.com/cloud/private-cloud/nationwide-targets-private-cloud-platform-gains-with-vmware-cloud-foundation-deal">Nationwide </a>announced a partnership with Moneyhub to launch an AI-powered transaction analytics platform for customers across the UK.</p><p>Lloyds Banking Group, meanwhile, is also <a href="https://www.itpro.com/business/careers-and-training/lloyds-banking-group-wants-to-train-every-employee-in-ai-by-the-end-of-this-year-heres-how-it-plans-to-do-it">accelerating AI innovation internally</a> in a bid to drive customer support capabilities. The bank unveiled plans to launch a new AI Academy earlier this year, with the aim of providing 67,000 staff with practical skills. </p><h2 id="building-the-foundations-for-ai">Building the foundations for AI</h2><p>Before rolling out new tools, YBS built a cloud native data platform with Microsoft Fabric and strengthened data governance through Microsoft Purview. </p><p>It also enhanced its security with Microsoft Sentinel, expanded infrastructure foundations in Azure, and introduced <a href="https://www.itpro.com/software/microsoft-remote-desktop-app-end-of-life">Windows 365 remote desktops</a> to enable secure, flexible working.</p><p>“Our ambition with data and AI is to improve our organization, help our colleagues be more efficient, make better decisions, and ultimately, serve our members better,” said Rebecca Fitzgerald, YBS director of data and AI. “All built on trust, with responsibility built in.”</p><p>Elsewhere, YBS is <a href="https://www.itpro.com/technology/artificial-intelligence/half-of-agentic-ai-projects-are-still-stuck-at-the-pilot-stage-but-thats-not-stopping-enterprises-from-ramping-up-investment">piloting AI agents</a> to support internal risk and control testing, with early results suggesting efficiency savings of around 40%.</p><p>The high street lender is also rolling out a new customer service platform built using <a href="https://www.itpro.com/business-operations/sales-crm/354830/microsoft-touts-enhanced-ai-features-for-dynamics-365">Microsoft Dynamics 365 </a>Contact-Centre-As-A-Service. This brings customer history, previous self-service activity, and relevant guidance into one place for colleagues.</p><p>An <a href="https://www.itpro.com/technology/artificial-intelligence/the-race-for-ai-assistants-has-grown-boring">AI assistant</a> can summarize cases, highlight relevant knowledge and help draft responses, reducing the need to search across multiple systems.</p><p>“The platform is really going to help our colleagues better serve our members,” Fox commented. “It’s going to enable them to have a 360‑degree view of our members, so they’re not having to move between various platforms.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/business-strategy/yorkshire-building-society-touts-customer-service-gains-with-ai-agents</link>
                                                                            <description>
                            <![CDATA[ The firm said that agents Penelope, Sam, and Alf are helping improve customer service, while adhering to high levels of regulation ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">KiS9dMtKgYoRHCdmCsByEf</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/3Ubvi97vXADcPW9ainSnFB-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Tue, 07 Jul 2026 11:02:05 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 15:06:39 +0000</updated>
                                                                                                                                            <category><![CDATA[Customer Experience Platforms]]></category>
                                                    <category><![CDATA[Enterprise Applications]]></category>
                                                                                                                    <dc:creator><![CDATA[ Emma Woollacott ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/aWfskavxoVSMDy6cDWtYmJ-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/3Ubvi97vXADcPW9ainSnFB-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Yorkshire Building Society branch shopfront pictured in London, England, with pedestrians walking by in foreground.]]></media:description>                                                            <media:text><![CDATA[Yorkshire Building Society branch shopfront pictured in London, England, with pedestrians walking by in foreground.]]></media:text>
                                <media:title type="plain"><![CDATA[Yorkshire Building Society branch shopfront pictured in London, England, with pedestrians walking by in foreground.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/3Ubvi97vXADcPW9ainSnFB-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Yorkshire Building Society (YBS) is using AI agents to help reduce admin and improve customer service – and the high street lender said it's recording solid results so far.</p><p>Three AI agents named Penelope, Sam, and Alf are supporting customer service teams by summarizing complex complaints, searching policies and past cases, and drafting member communications, all with human oversight.</p><p>Penelope helps draft final responses for complaints, while Alf supports this process by searching policies, procedures, and past cases. Sam, meanwhile, summarizes long or detailed complaints.</p><p>According to the building society, Sam is saving an estimated seven minutes per use, with Penelope saving up to 26 minutes when supporting more complex complaint responses. </p><p>“That time really adds up,” said Polly Conner, senior manager of customer relations at YBS. “We’re able to minimize the amount of time staff are spending on admin tasks and actually spend that time talking to our customers, helping them resolve the challenges they’ve faced.”</p><p>The high level of regulation around financial services means that handling complaints from the building society's three million members and customers requires a lot of admin.</p><p>That can mean <a href="https://www.itpro.com/software/development/context-switching-is-a-major-drain-on-developer-productivity-heres-how-github-plans-to-solve-that">switching between multiple systems</a>, searching for policies, summarizing long histories, and drafting detailed responses in advance of any conversation. </p><p>"This technology is going to give our members more choice and better access to information when they need it most,” said Simone Fox, director of customer support at YBS. “But there will always be a human available to speak to them when they need it.” </p><h2 id="ai-in-banking">AI in banking </h2><p>The results highlighted by YBS come amidst a sharpened focus on AI in the financial services sector, with a host of major <a href="https://www.itpro.com/technology/artificial-intelligence/using-generative-ai-as-a-copilot-is-the-sweet-spot-a-look-at-nationwides-ai-approach">providers ramping up adoption of the technology</a>. </p><p>In January this year, <a href="https://www.itpro.com/cloud/private-cloud/nationwide-targets-private-cloud-platform-gains-with-vmware-cloud-foundation-deal">Nationwide </a>announced a partnership with Moneyhub to launch an AI-powered transaction analytics platform for customers across the UK.</p><p>Lloyds Banking Group, meanwhile, is also <a href="https://www.itpro.com/business/careers-and-training/lloyds-banking-group-wants-to-train-every-employee-in-ai-by-the-end-of-this-year-heres-how-it-plans-to-do-it">accelerating AI innovation internally</a> in a bid to drive customer support capabilities. The bank unveiled plans to launch a new AI Academy earlier this year, with the aim of providing 67,000 staff with practical skills. </p><h2 id="building-the-foundations-for-ai">Building the foundations for AI</h2><p>Before rolling out new tools, YBS built a cloud native data platform with Microsoft Fabric and strengthened data governance through Microsoft Purview. </p><p>It also enhanced its security with Microsoft Sentinel, expanded infrastructure foundations in Azure, and introduced <a href="https://www.itpro.com/software/microsoft-remote-desktop-app-end-of-life">Windows 365 remote desktops</a> to enable secure, flexible working.</p><p>“Our ambition with data and AI is to improve our organization, help our colleagues be more efficient, make better decisions, and ultimately, serve our members better,” said Rebecca Fitzgerald, YBS director of data and AI. “All built on trust, with responsibility built in.”</p><p>Elsewhere, YBS is <a href="https://www.itpro.com/technology/artificial-intelligence/half-of-agentic-ai-projects-are-still-stuck-at-the-pilot-stage-but-thats-not-stopping-enterprises-from-ramping-up-investment">piloting AI agents</a> to support internal risk and control testing, with early results suggesting efficiency savings of around 40%.</p><p>The high street lender is also rolling out a new customer service platform built using <a href="https://www.itpro.com/business-operations/sales-crm/354830/microsoft-touts-enhanced-ai-features-for-dynamics-365">Microsoft Dynamics 365 </a>Contact-Centre-As-A-Service. This brings customer history, previous self-service activity, and relevant guidance into one place for colleagues.</p><p>An <a href="https://www.itpro.com/technology/artificial-intelligence/the-race-for-ai-assistants-has-grown-boring">AI assistant</a> can summarize cases, highlight relevant knowledge and help draft responses, reducing the need to search across multiple systems.</p><p>“The platform is really going to help our colleagues better serve our members,” Fox commented. “It’s going to enable them to have a 360‑degree view of our members, so they’re not having to move between various platforms.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Agentic AI 'breaks the traditional SaaS seat licensing model' ]]></title>
                                                                                                <dc:content><![CDATA[ <p>The <a href="https://www.itpro.com/cloud/software-as-a-service-saas/362655/what-is-saas">SaaS </a>industry is facing a pending wave of disruption as enterprises shift their focus towards <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">agentic AI</a> solutions, according to new research from Gartner. </p><p>With agentic AI adoption continuing at pace globally, the research firm said this poses a significant threat to traditional software revenue models. </p><p>Indeed, up to $234 billion of application spending will be exposed to "agentic arbitrage" – where agents render some services obsolete by removing the need to work across multiple interfaces – by 2030.</p><p>George Brocklehurst, managing vice president at Gartner, said the study shows agentic AI “changes the economics of software”. </p><p>“Agentic systems deliver outcomes directly, bypassing traditional user experience (UX)-heavy applications and making the software invisible,” he said. “This breaks the link between user growth and revenue growth for many enterprise software vendors.”</p><h2 id="no-saaspocalypse-but-a-metamorphosis">No ‘SaaSpocalypse’, but a ‘metamorphosis’</h2><p>Long-term, Gartner noted that this agentic AI-focused shift will “refactor” how software is built, priced, and consumed. Indeed, Brocklehurst said a period of “redefinition” could come, with the firm describing this transformation as a "metamorphosis". </p><p>“This metamorphosis represents threats and opportunities for both incumbents and new challengers,” Gartner said. </p><p>Gartner’s claims regarding a period of transformation for the SaaS space align closely with predictions made by industry stakeholders on the potential impact of AI. </p><p>Earlier this year, <a href="https://www.itpro.com/technology/artificial-intelligence/why-anthropic-sent-software-stocks-into-freefall"><u>software stocks plummeted amid a mass sell-off</u></a> due to the arrival of powerful new AI solutions, particularly Anthropic’s Claude Cowork service. </p><p>This spooked investors, with some concerned about a ‘SaaSpocalypse’ whereby agents could remove the need for certain software services. </p><p>Despite lingering concerns, some industry leaders are still bullish on the future of SaaS – albeit with a few caveats. As <em>ITPro </em>reported in February, <a href="https://www.itpro.com/software/aws-ceo-matt-garman-ai-software-industry-stocks"><u>AWS CEO Matt Garman poured cold water on the prospect of a looming SaaSpocalypse</u></a>, suggesting that “much of the fear is overblown.”</p><h2 id="staying-competitive">Staying competitive</h2><p>Remaining competitive in a changing market will require incumbent software vendors to “move from interfaced-based value to outcome-based value,” according to Gartner. </p><p>Put simply, providers will need to embed agentic capabilities within their offerings to defend their position and meet changing expectations. </p><p>Enterprises don’t want disparate tools and dashboards anymore, the researchers claimed, they want streamlined, end-to-end workflow automation and solutions that “capture customer context and knowledge”. </p><p>“As organizations increasingly use agentic AI systems, the user interface is no longer a differentiation,” Brocklehurst said. “Legacy SaaS market share will be cannibalized by incumbents and taken by new entrants delivering horizontal agentic platforms.”</p><p>“While this shift is posing an existential threat for vendors who are defending legacy dashboards and seat-based models, it creates a substantial revenue opportunity for vendors who are enabling and developing services and platforms to support agentic enabled cross-domain workflows,” he added.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/software/agentic-ai-breaks-the-traditional-saas-seat-licensing-model-now-its-up-to-vendors-to-ditch-legacy-dashboards-and-build-with-agents-in-mind</link>
                                                                            <description>
                            <![CDATA[ Incumbent software vendors will need to work harder than ever to compete with agile, AI-focused disruptors ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">SbY9N3hpfui65DNnMBYRRA</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/iCoxtGZ7L6KdRG4GvEzabL-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Fri, 03 Jul 2026 10:12:18 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 13:17:20 +0000</updated>
                                                                                                                                            <category><![CDATA[Enterprise Applications]]></category>
                                                    <category><![CDATA[Software]]></category>
                                                    <category><![CDATA[Workplace Technology]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/iCoxtGZ7L6KdRG4GvEzabL-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Automation concept image showing a row of robots working on laptops.]]></media:description>                                                            <media:text><![CDATA[Automation concept image showing a row of robots working on laptops.]]></media:text>
                                <media:title type="plain"><![CDATA[Automation concept image showing a row of robots working on laptops.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/iCoxtGZ7L6KdRG4GvEzabL-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>The <a href="https://www.itpro.com/cloud/software-as-a-service-saas/362655/what-is-saas">SaaS </a>industry is facing a pending wave of disruption as enterprises shift their focus towards <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">agentic AI</a> solutions, according to new research from Gartner. </p><p>With agentic AI adoption continuing at pace globally, the research firm said this poses a significant threat to traditional software revenue models. </p><p>Indeed, up to $234 billion of application spending will be exposed to "agentic arbitrage" – where agents render some services obsolete by removing the need to work across multiple interfaces – by 2030.</p><p>George Brocklehurst, managing vice president at Gartner, said the study shows agentic AI “changes the economics of software”. </p><p>“Agentic systems deliver outcomes directly, bypassing traditional user experience (UX)-heavy applications and making the software invisible,” he said. “This breaks the link between user growth and revenue growth for many enterprise software vendors.”</p><h2 id="no-saaspocalypse-but-a-metamorphosis">No ‘SaaSpocalypse’, but a ‘metamorphosis’</h2><p>Long-term, Gartner noted that this agentic AI-focused shift will “refactor” how software is built, priced, and consumed. Indeed, Brocklehurst said a period of “redefinition” could come, with the firm describing this transformation as a "metamorphosis". </p><p>“This metamorphosis represents threats and opportunities for both incumbents and new challengers,” Gartner said. </p><p>Gartner’s claims regarding a period of transformation for the SaaS space align closely with predictions made by industry stakeholders on the potential impact of AI. </p><p>Earlier this year, <a href="https://www.itpro.com/technology/artificial-intelligence/why-anthropic-sent-software-stocks-into-freefall"><u>software stocks plummeted amid a mass sell-off</u></a> due to the arrival of powerful new AI solutions, particularly Anthropic’s Claude Cowork service. </p><p>This spooked investors, with some concerned about a ‘SaaSpocalypse’ whereby agents could remove the need for certain software services. </p><p>Despite lingering concerns, some industry leaders are still bullish on the future of SaaS – albeit with a few caveats. As <em>ITPro </em>reported in February, <a href="https://www.itpro.com/software/aws-ceo-matt-garman-ai-software-industry-stocks"><u>AWS CEO Matt Garman poured cold water on the prospect of a looming SaaSpocalypse</u></a>, suggesting that “much of the fear is overblown.”</p><h2 id="staying-competitive">Staying competitive</h2><p>Remaining competitive in a changing market will require incumbent software vendors to “move from interfaced-based value to outcome-based value,” according to Gartner. </p><p>Put simply, providers will need to embed agentic capabilities within their offerings to defend their position and meet changing expectations. </p><p>Enterprises don’t want disparate tools and dashboards anymore, the researchers claimed, they want streamlined, end-to-end workflow automation and solutions that “capture customer context and knowledge”. </p><p>“As organizations increasingly use agentic AI systems, the user interface is no longer a differentiation,” Brocklehurst said. “Legacy SaaS market share will be cannibalized by incumbents and taken by new entrants delivering horizontal agentic platforms.”</p><p>“While this shift is posing an existential threat for vendors who are defending legacy dashboards and seat-based models, it creates a substantial revenue opportunity for vendors who are enabling and developing services and platforms to support agentic enabled cross-domain workflows,” he added.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ UK business leaders have a 'limited understanding' of AI usage costs ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Most UK business leaders have a “limited understanding” of their AI budgets, and they're struggling to turn AI investment into measurable business outcomes.</p><p>KPMG’s latest <a href="https://kpmg.com/uk/en/insights/ai/ai-quarterly-pulse.html" target="_blank"><em>Global AI Pulse</em></a> report found that 26% of UK companies are now using AI as part of everyday work, up from 18% in the first quarter of this year. </p><p>This increased adoption means that managing usage costs is becoming both more complex and more critical to realizing value. </p><p>Three-in-ten UK leaders told the firm that they struggle with usage-based costs, while 42% have only partial visibility into AI spending. One-third, meanwhile, cited “limited understanding” of AI cost structures, including tokens, as a challenge to deploying AI agents.</p><p>“AI is moving rapidly into everyday work, but scaling it responsibly brings a new set of challenges," said Dr Leanne Allen, head of AI at KPMG UK. </p><p>"Leaders now need to show not just that AI can be deployed, but that it can be trusted, financially controlled and clearly linked to value. Cost visibility is central to that." </p><h2 id="enterprises-need-a-clear-path-with-ai">Enterprises need a clear path with AI</h2><p>According to KPMG, companies where CEOs are accountable for AI decisions often report higher confidence in their AI strategy, and are more likely to unlock meaningful business value and stronger returns on investment.</p><p>"As organizations use more AI tools and agentic systems, they need to understand how costs build, where value is being created and where governance controls are needed. Without that clarity, it becomes harder to make confident investment decisions or demonstrate returns," said Allen.</p><p>“Clear accountability, practical governance, and workforce adoption must move together if businesses are to turn AI momentum into sustained value.”</p><p>To help <a href="https://www.itpro.com/technology/artificial-intelligence/could-rising-token-costs-boost-interest-in-on-premises-hardware">manage AI costs</a>, the survey found organizations are implementing stronger governance controls, including monitoring and spending controls. </p><p>More than half (57%) of UK leaders report having AI cost monitoring dashboards, with 61% embedding cost reviews as part of AI approval processes to enable stronger control and decision-making.</p><p>Notably, organizations with stronger cost visibility are four-times more likely to report established ROI, at 25% versus 6%.</p><p>“AI cost management cannot sit as an afterthought. If businesses want to scale AI responsibly, they need to build financial discipline into the way AI is approved, monitored and governed from the start," said Allen.</p><p>"The organizations that can see their AI costs clearly are better placed to understand what is working, what is not and where to keep investing.”</p><h2 id="ai-costs-are-spiralling">AI costs are spiralling</h2><p>Surging AI costs have become a recurring pain point for enterprises across a range of industries in recent months. As <em>ITPro </em>reported in June, these surging costs are the result of AI provider shifts toward consumption-based billing combined with increased usage rates. </p><p>The ‘tokenmaxxing’ trend, whereby users are encouraged to ramp up their use of AI tools, has already caused serious issues for some major companies, such as Uber. </p><p><a href="https://www.itpro.com/technology/artificial-intelligence/ubers-eye-watering-ai-bill-shows-enterprises-are-still-measuring-ai-success-through-consumption-rather-than-outcomes-and-its-warping-our-perception-of-roi-and-productivity"><u>Uber revealed that it used its entire annual AI budget in just four months</u></a> after encouraging staff to use the technology, prompting a rethink of how AI is used internally. This included the introduction of a $1,500 monthly cap per employee, which is tracked through an internal dashboard. </p><p>Accenture has also <a href="https://www.itpro.com/technology/artificial-intelligence/what-were-seeing-right-now-is-just-rapid-escalation-in-ai-token-spend-accenture-tells-staff-to-stop-using-ai-for-unnecessary-tasks-amid-surging-cost"><u>urged staff to stop using AI for needless tasks</u></a> in a bid to tackle mounting costs. </p><p>Last week, analysts at Gartner told <em>ITPro </em>that tackling this problem will require a <a href="https://www.itpro.com/software/development/surging-ai-costs-could-exceed-developer-salaries-by-2028-analysts-say-context-engineering-could-be-the-key-to-optimizing-token-consumption"><u>concerted focus on cost optimization practices</u></a>, including the use of context engineering techniques to maximize the use of the technology. </p><p>This call to action by Gartner came after research found AI token costs could exceed developer salaries by 2028. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/business-strategy/uk-business-leaders-have-a-limited-understanding-of-ai-usage-costs-and-its-coming-back-to-bite-them</link>
                                                                            <description>
                            <![CDATA[ Companies where CEOs are accountable for AI decisions report higher confidence in their strategy ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">3S4PYHv2wSUnEnAoeSkmdZ</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/FTix2t5mUZE2fyrQzaTCJ5-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Fri, 03 Jul 2026 09:32:33 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 15:08:39 +0000</updated>
                                                                                                                                            <category><![CDATA[FinOps]]></category>
                                                    <category><![CDATA[IT Operations]]></category>
                                                                                                                    <dc:creator><![CDATA[ Emma Woollacott ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/aWfskavxoVSMDy6cDWtYmJ-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/FTix2t5mUZE2fyrQzaTCJ5-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Metallic business people figurines standing at the entrance to a maze. ]]></media:description>                                                            <media:text><![CDATA[Metallic business people figurines standing at the entrance to a maze. ]]></media:text>
                                <media:title type="plain"><![CDATA[Metallic business people figurines standing at the entrance to a maze. ]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/FTix2t5mUZE2fyrQzaTCJ5-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Most UK business leaders have a “limited understanding” of their AI budgets, and they're struggling to turn AI investment into measurable business outcomes.</p><p>KPMG’s latest <a href="https://kpmg.com/uk/en/insights/ai/ai-quarterly-pulse.html" target="_blank"><em>Global AI Pulse</em></a> report found that 26% of UK companies are now using AI as part of everyday work, up from 18% in the first quarter of this year. </p><p>This increased adoption means that managing usage costs is becoming both more complex and more critical to realizing value. </p><p>Three-in-ten UK leaders told the firm that they struggle with usage-based costs, while 42% have only partial visibility into AI spending. One-third, meanwhile, cited “limited understanding” of AI cost structures, including tokens, as a challenge to deploying AI agents.</p><p>“AI is moving rapidly into everyday work, but scaling it responsibly brings a new set of challenges," said Dr Leanne Allen, head of AI at KPMG UK. </p><p>"Leaders now need to show not just that AI can be deployed, but that it can be trusted, financially controlled and clearly linked to value. Cost visibility is central to that." </p><h2 id="enterprises-need-a-clear-path-with-ai">Enterprises need a clear path with AI</h2><p>According to KPMG, companies where CEOs are accountable for AI decisions often report higher confidence in their AI strategy, and are more likely to unlock meaningful business value and stronger returns on investment.</p><p>"As organizations use more AI tools and agentic systems, they need to understand how costs build, where value is being created and where governance controls are needed. Without that clarity, it becomes harder to make confident investment decisions or demonstrate returns," said Allen.</p><p>“Clear accountability, practical governance, and workforce adoption must move together if businesses are to turn AI momentum into sustained value.”</p><p>To help <a href="https://www.itpro.com/technology/artificial-intelligence/could-rising-token-costs-boost-interest-in-on-premises-hardware">manage AI costs</a>, the survey found organizations are implementing stronger governance controls, including monitoring and spending controls. </p><p>More than half (57%) of UK leaders report having AI cost monitoring dashboards, with 61% embedding cost reviews as part of AI approval processes to enable stronger control and decision-making.</p><p>Notably, organizations with stronger cost visibility are four-times more likely to report established ROI, at 25% versus 6%.</p><p>“AI cost management cannot sit as an afterthought. If businesses want to scale AI responsibly, they need to build financial discipline into the way AI is approved, monitored and governed from the start," said Allen.</p><p>"The organizations that can see their AI costs clearly are better placed to understand what is working, what is not and where to keep investing.”</p><h2 id="ai-costs-are-spiralling">AI costs are spiralling</h2><p>Surging AI costs have become a recurring pain point for enterprises across a range of industries in recent months. As <em>ITPro </em>reported in June, these surging costs are the result of AI provider shifts toward consumption-based billing combined with increased usage rates. </p><p>The ‘tokenmaxxing’ trend, whereby users are encouraged to ramp up their use of AI tools, has already caused serious issues for some major companies, such as Uber. </p><p><a href="https://www.itpro.com/technology/artificial-intelligence/ubers-eye-watering-ai-bill-shows-enterprises-are-still-measuring-ai-success-through-consumption-rather-than-outcomes-and-its-warping-our-perception-of-roi-and-productivity"><u>Uber revealed that it used its entire annual AI budget in just four months</u></a> after encouraging staff to use the technology, prompting a rethink of how AI is used internally. This included the introduction of a $1,500 monthly cap per employee, which is tracked through an internal dashboard. </p><p>Accenture has also <a href="https://www.itpro.com/technology/artificial-intelligence/what-were-seeing-right-now-is-just-rapid-escalation-in-ai-token-spend-accenture-tells-staff-to-stop-using-ai-for-unnecessary-tasks-amid-surging-cost"><u>urged staff to stop using AI for needless tasks</u></a> in a bid to tackle mounting costs. </p><p>Last week, analysts at Gartner told <em>ITPro </em>that tackling this problem will require a <a href="https://www.itpro.com/software/development/surging-ai-costs-could-exceed-developer-salaries-by-2028-analysts-say-context-engineering-could-be-the-key-to-optimizing-token-consumption"><u>concerted focus on cost optimization practices</u></a>, including the use of context engineering techniques to maximize the use of the technology. </p><p>This call to action by Gartner came after research found AI token costs could exceed developer salaries by 2028. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Context engineering could be the key to tackling surging AI token costs ]]></title>
                                                                                                <dc:content><![CDATA[ <p><a href="https://www.itpro.com/technology/artificial-intelligence/how-ai-coding-is-transforming-the-it-industry-in-2025">AI coding</a> costs could exceed software developer salaries by 2028, according to new research from Gartner, prompting calls for a greater focus on cost optimization efforts. </p><p>The projection by the consultancy is a result of two overlapping trends, including surging <a href="https://www.itpro.com/technology/artificial-intelligence/could-rising-token-costs-boost-interest-in-on-premises-hardware">token consumption rates</a> and the shift to consumption-based pricing (CPB) models.</p><p>A host of software and AI providers have pivoted away from flat rate “per-seat” subscriptions in recent months to CPB setups. As <em>ITPro </em>reported in April, <a href="https://www.itpro.com/software/development/github-copilot-pricing-changes-usage-based-billing-explained"><u>GitHub signalled its own shift on this front</u></a>, citing rising costs specifically as a key factor behind the decision. </p><p>Speaking to <em>ITPro</em>, Nitish Tyagi, Senior Principal Analyst at Gartner, said conversations with clients shows this is rapidly becoming a key concern for enterprises, particularly <a href="https://www.itpro.com/technology/artificial-intelligence/ai-in-software-engineering-six-ways-the-profession-is-changing">software engineering</a> leaders as teams ramp up adoption of tools. </p><p>“What has happened is that most vendors have switched to a consumption-based pricing model, and it took most engineering organisations by surprise,” he said. “We were never thinking that AI will be this costly, and we are already seeing that cost.”</p><p>“Initially, it wasn't alarming,” Tyagi added. “While many organizations are still in the range of $200 to $500 per developer per month, alarming results are coming up.”</p><p>Tyagi noted that discussions with Gartner clients show heightened use of AI is costing more than $2,000 per developer, per month. These costs are rising as well, with some eye-watering figures now emerging. </p><p>“I’ve been talking to clients where they are telling me that my power users are now costing me more than $2,500 per developer, per month. Sometimes we also hear really crazy numbers, like ‘my developer cost me $20,000 last month, or my business users cost me $32,000 per month.”</p><p>Reports on surging AI costs have been coming thick and fast in recent months, and some major firms such as Microsoft have gone so far as to implement usage limits or even cut the use of specific tools internally. </p><p>As <em>ITPro </em>reported earlier this month, Uber <a href="https://www.itpro.com/technology/artificial-intelligence/ubers-eye-watering-ai-bill-shows-enterprises-are-still-measuring-ai-success-through-consumption-rather-than-outcomes-and-its-warping-our-perception-of-roi-and-productivity">blew through its entire annual AI budget in just four months</a> after encouraging staff to ramp up their use of AI tools internally. </p><p>One of the leading causes behind these price increases and hefty bills lies in ‘tokenmaxxing’ – a trend in which enterprises measure AI usage based on the number of tokens consumed by users. </p><p>For some, it’s a way of tracking productivity by highlighting their use of the technology, although critics argue this pads stats and results in skewed metrics. </p><p>These overlapping trends raise serious concerns about long-term AI usage, particularly around returns on investment (ROI). Simply put, enterprises want their devs using AI, but heightened costs mean that measuring value is becoming more difficult. </p><p>“This is becoming a big issue from two aspects,” he told <em>ITPro</em>. “It’s not only related to developer salaries. This will catch the eyes, but the bigger problem here is that organizations were already struggling and justifying ROI for using these tools.”</p><p>“Now that the cost has increased, and it is increasing as well, it is becoming even more difficult to justify these costs and to identify where the ROI is.”</p><h2 id="cost-optimization-practices-need-to-improve">Cost optimization practices need to improve</h2><p>Tyagi is keen to emphasize that the study doesn’t suggest that developers drop AI tools or agents outright. This completely misses the point, he said. </p><p>Instead, engineering leaders and enterprises at large need to sharpen up on cost optimization processes. Efforts on this front are critical, particularly as the study noted that vendors themselves are “yet to deliver mature, built-in cost optimization capabilities” for agents. </p><p>A key practice highlighted by Tyagi includes context engineering. This is <a href="https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents"><u>a technique </u></a>that involves selectively curating, structuring, and managing information used by AI systems to generate responses. </p><p>This essentially involves providing only the relevant information needed by agents, as well as concise summarization of content and elimination of "unnecessary data”</p><p>Tyagi noted that this will not only help optimize token consumption for developers, but ultimately deliver longer term benefits in terms of output quality. </p><p>“Context engineering is going to become the most important skill in the future,” he told <em>ITPro</em>. “If I actually optimize token consumption, the I would be able to increase the output quality as well.”</p><p>This is an emerging focus area for software engineering leaders, Tyagi added, and Gartner suggests that context engineering practices could eventually be mandated to cut consumption rates. </p><p>It’s also a blossoming market, he noted. Tabnine, for example, recently launched a new solution known as “Context Engine” aimed at streamlining processes on this front while Atlassian is also developing solutions in this domain. </p><h2 id="selective-use">Selective use</h2><p>Gartner’s research also urges enterprises to establish a “use-case-driven decision” framework when it comes to using AI for everyday tasks. </p><p>Simply put, the focus here should be on clearly defining <em>when </em>AI coding agents should be used, and how much autonomy these bots are given on particular tasks. Given the increasing use of agents among developers, many are simply allocating tasks that aren’t needed. </p><p>“Not everything has to be done by an agent, not everything has to be done by developers,” he said. </p><p>Looking ahead, developer teams should assess when - and when not to - assign agents to specific tasks, which Tyagi noted could help cut needless consumption and ultimately provide greater control. </p><p>“When you’re working on highly sensitive tasks or highly complex tasks, then you want your developers to break down those tasks as well into smaller sub tasks,” he explained. </p><p>“We have seen when you are throwing smaller clues, problems at agents, agents do much better than giving them an open-ended problem,” Tyagi added. “Agents are improving, but I think developers should have that level of control right now, through which they can actually optimize the token consumption.”</p><p>It’s here that model selection is equally important, according to Gartner, specifically for smaller tasks, which can naturally be handled by smaller models. </p><p>Using “intelligent model routing” strategies will enable developers to box clever when it comes to smaller, niche AI models or larger frontier options, which typically come with added costs. </p><p>The consultancy noted that escalation is needed “only when complexity demands it”. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/software/development/surging-ai-costs-could-exceed-developer-salaries-by-2028-analysts-say-context-engineering-could-be-the-key-to-optimizing-token-consumption</link>
                                                                            <description>
                            <![CDATA[ With AI costs rising and enterprises racking up huge bills, engineering leaders need to take drastic measures to limit costs ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">WhCjW2nHAEseVrjopaPNZf</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/8Rwc7YHCX9iZGuuMuAYfh5-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Fri, 26 Jun 2026 09:34:11 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 13:42:58 +0000</updated>
                                                                                                                                            <category><![CDATA[FinOps]]></category>
                                                    <category><![CDATA[Development]]></category>
                                                    <category><![CDATA[IT Operations]]></category>
                                                    <category><![CDATA[Software]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/8Rwc7YHCX9iZGuuMuAYfh5-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Male and female software developers discussing strategy in an open plan office space, with desktop computer showing source code on screen in background.]]></media:description>                                                            <media:text><![CDATA[Male and female software developers discussing strategy in an open plan office space, with desktop computer showing source code on screen in background.]]></media:text>
                                <media:title type="plain"><![CDATA[Male and female software developers discussing strategy in an open plan office space, with desktop computer showing source code on screen in background.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/8Rwc7YHCX9iZGuuMuAYfh5-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p><a href="https://www.itpro.com/technology/artificial-intelligence/how-ai-coding-is-transforming-the-it-industry-in-2025">AI coding</a> costs could exceed software developer salaries by 2028, according to new research from Gartner, prompting calls for a greater focus on cost optimization efforts. </p><p>The projection by the consultancy is a result of two overlapping trends, including surging <a href="https://www.itpro.com/technology/artificial-intelligence/could-rising-token-costs-boost-interest-in-on-premises-hardware">token consumption rates</a> and the shift to consumption-based pricing (CPB) models.</p><p>A host of software and AI providers have pivoted away from flat rate “per-seat” subscriptions in recent months to CPB setups. As <em>ITPro </em>reported in April, <a href="https://www.itpro.com/software/development/github-copilot-pricing-changes-usage-based-billing-explained"><u>GitHub signalled its own shift on this front</u></a>, citing rising costs specifically as a key factor behind the decision. </p><p>Speaking to <em>ITPro</em>, Nitish Tyagi, Senior Principal Analyst at Gartner, said conversations with clients shows this is rapidly becoming a key concern for enterprises, particularly <a href="https://www.itpro.com/technology/artificial-intelligence/ai-in-software-engineering-six-ways-the-profession-is-changing">software engineering</a> leaders as teams ramp up adoption of tools. </p><p>“What has happened is that most vendors have switched to a consumption-based pricing model, and it took most engineering organisations by surprise,” he said. “We were never thinking that AI will be this costly, and we are already seeing that cost.”</p><p>“Initially, it wasn't alarming,” Tyagi added. “While many organizations are still in the range of $200 to $500 per developer per month, alarming results are coming up.”</p><p>Tyagi noted that discussions with Gartner clients show heightened use of AI is costing more than $2,000 per developer, per month. These costs are rising as well, with some eye-watering figures now emerging. </p><p>“I’ve been talking to clients where they are telling me that my power users are now costing me more than $2,500 per developer, per month. Sometimes we also hear really crazy numbers, like ‘my developer cost me $20,000 last month, or my business users cost me $32,000 per month.”</p><p>Reports on surging AI costs have been coming thick and fast in recent months, and some major firms such as Microsoft have gone so far as to implement usage limits or even cut the use of specific tools internally. </p><p>As <em>ITPro </em>reported earlier this month, Uber <a href="https://www.itpro.com/technology/artificial-intelligence/ubers-eye-watering-ai-bill-shows-enterprises-are-still-measuring-ai-success-through-consumption-rather-than-outcomes-and-its-warping-our-perception-of-roi-and-productivity">blew through its entire annual AI budget in just four months</a> after encouraging staff to ramp up their use of AI tools internally. </p><p>One of the leading causes behind these price increases and hefty bills lies in ‘tokenmaxxing’ – a trend in which enterprises measure AI usage based on the number of tokens consumed by users. </p><p>For some, it’s a way of tracking productivity by highlighting their use of the technology, although critics argue this pads stats and results in skewed metrics. </p><p>These overlapping trends raise serious concerns about long-term AI usage, particularly around returns on investment (ROI). Simply put, enterprises want their devs using AI, but heightened costs mean that measuring value is becoming more difficult. </p><p>“This is becoming a big issue from two aspects,” he told <em>ITPro</em>. “It’s not only related to developer salaries. This will catch the eyes, but the bigger problem here is that organizations were already struggling and justifying ROI for using these tools.”</p><p>“Now that the cost has increased, and it is increasing as well, it is becoming even more difficult to justify these costs and to identify where the ROI is.”</p><h2 id="cost-optimization-practices-need-to-improve">Cost optimization practices need to improve</h2><p>Tyagi is keen to emphasize that the study doesn’t suggest that developers drop AI tools or agents outright. This completely misses the point, he said. </p><p>Instead, engineering leaders and enterprises at large need to sharpen up on cost optimization processes. Efforts on this front are critical, particularly as the study noted that vendors themselves are “yet to deliver mature, built-in cost optimization capabilities” for agents. </p><p>A key practice highlighted by Tyagi includes context engineering. This is <a href="https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents"><u>a technique </u></a>that involves selectively curating, structuring, and managing information used by AI systems to generate responses. </p><p>This essentially involves providing only the relevant information needed by agents, as well as concise summarization of content and elimination of "unnecessary data”</p><p>Tyagi noted that this will not only help optimize token consumption for developers, but ultimately deliver longer term benefits in terms of output quality. </p><p>“Context engineering is going to become the most important skill in the future,” he told <em>ITPro</em>. “If I actually optimize token consumption, the I would be able to increase the output quality as well.”</p><p>This is an emerging focus area for software engineering leaders, Tyagi added, and Gartner suggests that context engineering practices could eventually be mandated to cut consumption rates. </p><p>It’s also a blossoming market, he noted. Tabnine, for example, recently launched a new solution known as “Context Engine” aimed at streamlining processes on this front while Atlassian is also developing solutions in this domain. </p><h2 id="selective-use">Selective use</h2><p>Gartner’s research also urges enterprises to establish a “use-case-driven decision” framework when it comes to using AI for everyday tasks. </p><p>Simply put, the focus here should be on clearly defining <em>when </em>AI coding agents should be used, and how much autonomy these bots are given on particular tasks. Given the increasing use of agents among developers, many are simply allocating tasks that aren’t needed. </p><p>“Not everything has to be done by an agent, not everything has to be done by developers,” he said. </p><p>Looking ahead, developer teams should assess when - and when not to - assign agents to specific tasks, which Tyagi noted could help cut needless consumption and ultimately provide greater control. </p><p>“When you’re working on highly sensitive tasks or highly complex tasks, then you want your developers to break down those tasks as well into smaller sub tasks,” he explained. </p><p>“We have seen when you are throwing smaller clues, problems at agents, agents do much better than giving them an open-ended problem,” Tyagi added. “Agents are improving, but I think developers should have that level of control right now, through which they can actually optimize the token consumption.”</p><p>It’s here that model selection is equally important, according to Gartner, specifically for smaller tasks, which can naturally be handled by smaller models. </p><p>Using “intelligent model routing” strategies will enable developers to box clever when it comes to smaller, niche AI models or larger frontier options, which typically come with added costs. </p><p>The consultancy noted that escalation is needed “only when complexity demands it”. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Meet Claude Tag, Anthropic’s new AI teammate that works in Slack ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Anthropic has unveiled a new agent designed to act as a virtual teammate within Slack as the company eyes deeper integration with the workplace productivity platform. </p><p>Dubbed Claude Tag, the new feature is underpinned by Opus 4.8 and builds on existing tools offered by Anthropic, with the company noting it views the feature as “the beginning of an evolution of Claude Code”. </p><p>“We’re launching Claude Tag on Slack, since it’s a natural home for collaborative work between teams and AI, and where much of Anthropic’s day-to-day work already happens,” the company said in a <a href="https://www.anthropic.com/news/introducing-claude-tag" target="_blank"><u>blog post</u></a>. </p><p>Claude Tag works by allowing users to @ the agent within Slack channels. Users can essentially allocate tasks to the agent, helping automate processes to support various projects across any given organization. </p><p>According to Anthropic, internal teams have been using the feature for some time, noting that it’s “one of the main ways we get things done”. Indeed, the company said around 65% of its product team’s code is created by its own internal version. </p><p>“The same pattern is now spreading well beyond engineering — we’re tagging Claude to chase down product metrics and data, work through support tickets, or even help find the root cause of tricky bugs,” Anthropic said in a blog post. </p><h2 id="using-claude-tag">Using Claude Tag</h2><p>Anthropic said using Claude Tag will be a familiar process for those already using its AI tools in the productivity platform, or those who have worked with <a href="https://www.itpro.com/software/development/anthropic-claude-code-usage-limits-increase-spacex-compute-deal">Claude Code</a> or <a href="https://www.itpro.com/technology/artificial-intelligence/everything-you-need-to-know-about-anthropic-claude-cowork">Claude Cowork</a>. </p><p>When a user tags the agent with a request, it will “break its task down into stages and then work through them in turn” depending on the tools it has accessed to. </p><p>“Once it’s done, it’ll respond in a Slack thread with that it’s created,” the company explained. </p><p>Compared to previous Claude-based tools, Tag has several new features. Anthropic noted that the new tool is now “multiplayer”, meaning that each Slack channel will have an individual agent that everyone can interact with. </p><p>“This means that anyone can see what it’s working on, and can pick up the conversation from where the last person left off,” Anthropic explained. </p><p>This marks a significant change from previous processes, whereby individual employees interacted with Claude within a single chat, or on a specific task. </p><p>“It’s much more like interacting collaboratively with a teammate.”</p><div class="youtube-video" data-nosnippet ><div class="video-aspect-box"><iframe data-lazy-priority="high" data-lazy-src="https://www.youtube-nocookie.com/embed/VojDzHaciKQ" allowfullscreen=""></iframe></div></div><p>The new agent also learns continuously based on the tasks it’s been assigned, building more context as time goes on. This is a key feature touted by the company, meaning that users don’t need to provide detailed explanations from scratch for each individual task. </p><p>Cross-channel context is also another key feature, according to Anthropic. The agent can learn and build context based on conversations in other Slack channels and data sources, provided it’s granted permission. </p><p>The company confirmed that it does not learn based on conversations in private channels. </p><p>Elsewhere, a new ‘ambient’ mode is available for Claude Tag, which provides users with interactive updates on project progress, tasks, or "whatever it thinks you might need to know”. </p><p>“It’ll flag relevant information from across the channels it’s in and the tools it’s connected to, and follow up on threads or tasks that have gone quiet without being resolved,” Anthropic explained. </p><h2 id="how-to-access-claude-tag">How to access Claude Tag</h2><p>Claude Tag will replace the existing Claude in Slack application, and is available now in beta for users working on Claude Enterprise or Team subscriptions. </p><p>Given the agent has deep access to sensitive information, the company says it has introduced “tight controls” and system administrators will be needed to provide access. </p><p>“To get up and running, system administrators specify which tools and information the model should have access to, in which channels,” the company explained, adding that users should view this as “creating separate Claude identities” for different uses. </p><p>“Everything, including its memories, will stay scoped to the channels defined by the administrators,” Anthropic noted. “For example, a model set up for sales work won’t pass on memories to one set up for engineering; nor will it give engineers access to any sales data or tools.”</p><p>Notably, administrators can set limits for token consumption, both for individual teams and the broader organization. </p><p>The introduction of token limits comes amidst <a href="https://www.itpro.com/technology/artificial-intelligence/ubers-eye-watering-ai-bill-shows-enterprises-are-still-measuring-ai-success-through-consumption-rather-than-outcomes-and-its-warping-our-perception-of-roi-and-productivity">rising concerns over spiralling costs associated with popular AI tools</a> such as Claude and ChatGPT. </p><h2 id="deeper-slack-integration">Deeper Slack integration</h2><p>The move by Anthropic marks the latest attempt to provide deeper integration of its tools with the popular workplace collaboration platform over the last year. </p><p>In December 2025, the company announced <a href="https://www.itpro.com/software/development/claude-code-is-coming-to-slack-heres-how-to-use-it-what-it-can-do-and-how-to-get-access"><u>integration of its Claude Code tool within the platform</u></a>. This integration aimed to embed the AI tool within software developer workflows, also helping to provide updates on project progress. </p><p>Notably, users are able to ask Claude Code to investigate and fix potential bugs, conduct code reviews, and improve collaborative debugging efforts for teams. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/software/meet-claude-tag-anthropics-new-ai-teammate-that-works-in-slack</link>
                                                                            <description>
                            <![CDATA[ The new Claude Tag tool will replace the existing Claude in Slack application ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">ogDMa9FQ4RjyfCt7WgGo</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/z5SZ9xmF7AzPc9DCdwAFu5-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 24 Jun 2026 10:33:28 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:21 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Software]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                    <category><![CDATA[Workplace Technology]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/z5SZ9xmF7AzPc9DCdwAFu5-1920-80.jpg">
                                                            <media:credit><![CDATA[Anthropic]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Claude Tag logo and branding featured on a promotional image from Anthropic.]]></media:description>                                                            <media:text><![CDATA[Claude Tag logo and branding featured on a promotional image from Anthropic.]]></media:text>
                                <media:title type="plain"><![CDATA[Claude Tag logo and branding featured on a promotional image from Anthropic.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/z5SZ9xmF7AzPc9DCdwAFu5-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Anthropic has unveiled a new agent designed to act as a virtual teammate within Slack as the company eyes deeper integration with the workplace productivity platform. </p><p>Dubbed Claude Tag, the new feature is underpinned by Opus 4.8 and builds on existing tools offered by Anthropic, with the company noting it views the feature as “the beginning of an evolution of Claude Code”. </p><p>“We’re launching Claude Tag on Slack, since it’s a natural home for collaborative work between teams and AI, and where much of Anthropic’s day-to-day work already happens,” the company said in a <a href="https://www.anthropic.com/news/introducing-claude-tag" target="_blank"><u>blog post</u></a>. </p><p>Claude Tag works by allowing users to @ the agent within Slack channels. Users can essentially allocate tasks to the agent, helping automate processes to support various projects across any given organization. </p><p>According to Anthropic, internal teams have been using the feature for some time, noting that it’s “one of the main ways we get things done”. Indeed, the company said around 65% of its product team’s code is created by its own internal version. </p><p>“The same pattern is now spreading well beyond engineering — we’re tagging Claude to chase down product metrics and data, work through support tickets, or even help find the root cause of tricky bugs,” Anthropic said in a blog post. </p><h2 id="using-claude-tag">Using Claude Tag</h2><p>Anthropic said using Claude Tag will be a familiar process for those already using its AI tools in the productivity platform, or those who have worked with <a href="https://www.itpro.com/software/development/anthropic-claude-code-usage-limits-increase-spacex-compute-deal">Claude Code</a> or <a href="https://www.itpro.com/technology/artificial-intelligence/everything-you-need-to-know-about-anthropic-claude-cowork">Claude Cowork</a>. </p><p>When a user tags the agent with a request, it will “break its task down into stages and then work through them in turn” depending on the tools it has accessed to. </p><p>“Once it’s done, it’ll respond in a Slack thread with that it’s created,” the company explained. </p><p>Compared to previous Claude-based tools, Tag has several new features. Anthropic noted that the new tool is now “multiplayer”, meaning that each Slack channel will have an individual agent that everyone can interact with. </p><p>“This means that anyone can see what it’s working on, and can pick up the conversation from where the last person left off,” Anthropic explained. </p><p>This marks a significant change from previous processes, whereby individual employees interacted with Claude within a single chat, or on a specific task. </p><p>“It’s much more like interacting collaboratively with a teammate.”</p><div class="youtube-video" data-nosnippet ><div class="video-aspect-box"><iframe data-lazy-priority="high" data-lazy-src="https://www.youtube-nocookie.com/embed/VojDzHaciKQ" allowfullscreen=""></iframe></div></div><p>The new agent also learns continuously based on the tasks it’s been assigned, building more context as time goes on. This is a key feature touted by the company, meaning that users don’t need to provide detailed explanations from scratch for each individual task. </p><p>Cross-channel context is also another key feature, according to Anthropic. The agent can learn and build context based on conversations in other Slack channels and data sources, provided it’s granted permission. </p><p>The company confirmed that it does not learn based on conversations in private channels. </p><p>Elsewhere, a new ‘ambient’ mode is available for Claude Tag, which provides users with interactive updates on project progress, tasks, or "whatever it thinks you might need to know”. </p><p>“It’ll flag relevant information from across the channels it’s in and the tools it’s connected to, and follow up on threads or tasks that have gone quiet without being resolved,” Anthropic explained. </p><h2 id="how-to-access-claude-tag">How to access Claude Tag</h2><p>Claude Tag will replace the existing Claude in Slack application, and is available now in beta for users working on Claude Enterprise or Team subscriptions. </p><p>Given the agent has deep access to sensitive information, the company says it has introduced “tight controls” and system administrators will be needed to provide access. </p><p>“To get up and running, system administrators specify which tools and information the model should have access to, in which channels,” the company explained, adding that users should view this as “creating separate Claude identities” for different uses. </p><p>“Everything, including its memories, will stay scoped to the channels defined by the administrators,” Anthropic noted. “For example, a model set up for sales work won’t pass on memories to one set up for engineering; nor will it give engineers access to any sales data or tools.”</p><p>Notably, administrators can set limits for token consumption, both for individual teams and the broader organization. </p><p>The introduction of token limits comes amidst <a href="https://www.itpro.com/technology/artificial-intelligence/ubers-eye-watering-ai-bill-shows-enterprises-are-still-measuring-ai-success-through-consumption-rather-than-outcomes-and-its-warping-our-perception-of-roi-and-productivity">rising concerns over spiralling costs associated with popular AI tools</a> such as Claude and ChatGPT. </p><h2 id="deeper-slack-integration">Deeper Slack integration</h2><p>The move by Anthropic marks the latest attempt to provide deeper integration of its tools with the popular workplace collaboration platform over the last year. </p><p>In December 2025, the company announced <a href="https://www.itpro.com/software/development/claude-code-is-coming-to-slack-heres-how-to-use-it-what-it-can-do-and-how-to-get-access"><u>integration of its Claude Code tool within the platform</u></a>. This integration aimed to embed the AI tool within software developer workflows, also helping to provide updates on project progress. </p><p>Notably, users are able to ask Claude Code to investigate and fix potential bugs, conduct code reviews, and improve collaborative debugging efforts for teams. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Copilot Cowork is now generally available ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Microsoft has announced that Copilot Cowork is now generally available for users globally, following a <a href="https://www.itpro.com/technology/artificial-intelligence/microsoft-is-rolling-out-copilot-cowork-to-more-customers">beta period via the tech giant’s Frontier program</a>.</p><p><a href="https://www.itpro.com/technology/artificial-intelligence/anthropics-claude-cowork-tool-is-coming-to-microsoft-copilot">First announced in March this year</a>, Copilot Cowork marks a significant milestone for Microsoft 365 Copilot users, integrating Anthropic’s highly popular tool within its flagship product. </p><p>According to Microsoft, Initial testing of the platform by a host of major companies, including Accenture, has proved highly successful. </p><p>In a <a href="https://www.microsoft.com/en-us/microsoft-365/blog/2026/06/16/copilot-cowork-is-now-generally-available/" target="_blank"><u>blog post</u></a> detailing the launch, Charles Lamanna, Microsoft’s EVP for Copilot, Agents, and Platform, said Cowork is the “fastest growing feature” in the history of the firm’s Frontier program. </p><p>“Cowork has among the highest user satisfaction of any Copilot or agent experience we have shipped,” he wrote.</p><p>“We learned from what we saw, engaged with you along the way, and used everything we heard to improve quality and add new features, including model choice, extensibility through plugins, and new cost management controls. Read more below and watch the full demo.”</p><p>Here’s what users can expect with Copilot Cowork. </p><h2 id="what-is-copilot-cowork">What is Copilot Cowork?</h2><p>Copilot Cowork is an integrated version of Anthropic’s Claude Cowork, which launched in January this year. The platform provides users with industry-specific AI capabilities to automate tasks across a range of professions. </p><p>Users can direct agents to carry out tasks on their behalf - such as file management or sending emails - by giving bots access to their files. The company described the new platform as essentially providing every worker with a “specialist” for their individual role. </p><p>Notably, the launch of the <a href="https://www.itpro.com/technology/artificial-intelligence/why-anthropic-sent-software-stocks-into-freefall">Cowork sparked a sell-off in the software market</a> amid fears the <a href="https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today">AI tool</a> could render dedicated software services obsolete. </p><h2 id="copilot-cowork-features-and-capabilities">Copilot Cowork features and capabilities</h2><p>Users at Microsoft shops can expect all the typical capabilities offered by Cowork, albeit with a few new features, according to Lamanna.</p><p>Notably, users will be able to choose which model they use with the service. Copilot Cowork currently runs on Anthropic models such as Claude Opus 4.8 and <a href="https://www.itpro.com/security/anthropic-claude-opus-claude-mythos-cyber-capabilities">Claude Sonnet 4.6</a>.</p><p>Customers will also be able to use GPT-5.5 further down the line with the launch of Cowork 1, the company revealed. According to Microsoft, this will deliver “optimal” cost and quality, and is intended for “enterprise-grade use”. </p><p>The tech giant noted that tweaking has helped remove model bias. </p><p>“It’s designed to handle everyday Copilot tasks at a substantially lower cost, making it a strong option for cost-sensitive workloads,” Lamanna wrote. </p><p>New partner plugins have also been added to Copilot Cowork, according to Microsoft, with a host of partners such as Miro and <a href="https://www.itpro.com/business/business-strategy/monday-com-promotes-ben-barnett-to-lead-emea-growth-strategy">Monday.com</a> already available. </p><p>Other partner plugins include: </p><ul><li>Enosix</li><li>Harvey</li><li>LSEG</li><li>Moodys</li><li>Morningstar</li><li>S&amp;P Global Energy</li><li>TeamsMaestro</li></ul><p>Several more are coming soon, including plugins for Atlassian, Canva, Box, Databricks, and Adobe. </p><p>Microsoft also touted a number of security-focused features for Copilot Cowork, aimed at shoring up governance and compliance. Cowork, prompts, responses, and artifacts, for example, now flow through existing Microsoft 365 controls.</p><h2 id="copilot-cowork-pricing">Copilot Cowork pricing</h2><p>To use Copilot Cowork, users will need to go through the Microsoft 365 Copilot User Subscription License (USL). This means that users are billed on a usage-basis, and charges are based on what particular tasks they use these agents for. </p><p>In terms of costs, customers have two payment options: PayGo and P3 (Credit Pre-Purchase Plan). The former is aimed at customers aiming for flexibility in terms of how they use the service, with credits priced at $0.01. </p><p>P3, meanwhile, is for those who want to commit to a usage volume “in advance in exchange for a discount”. </p><p>The cost of Copilot Cowork comes at a tricky time for businesses globally, particularly <a href="https://www.itpro.com/technology/artificial-intelligence/ubers-eye-watering-ai-bill-shows-enterprises-are-still-measuring-ai-success-through-consumption-rather-than-outcomes-and-its-warping-our-perception-of-roi-and-productivity">since the rise of the ‘tokenmaxxing’ trend</a> which has resulted in some firms racking up sizable AI bills. </p><p>Cost control features are available for customers, however, with Microsoft noting that customers can “decide when Cowork turns on, who gets access, and how much can be spent”. </p><p>Spending limits will be available at the tenant, group, and user levels, meaning admins can control costs and usage rates. </p><p>“Admins create scoped billing policies and define budgets, including user-level caps set inside group policies,” the company noted. “Admins set the thresholds that matter for their organization and groups, and choose who gets notified when spend crosses them.”</p><p>In the event that users exceed usage limits, credit requests can be made to complete tasks. These can be requested from inside Cowork. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/copilot-cowork-is-now-generally-available-everything-you-need-to-know-including-pricing-usage-limits-and-new-features</link>
                                                                            <description>
                            <![CDATA[ A host of partner plugins are already available for Copilot Cowork, and more are coming ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">jdazC7VTyjVbUtfix83NQG</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/6FdBRNc4i8SUjVx5Qf5qkR-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Mon, 22 Jun 2026 12:33:37 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:33 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/6FdBRNc4i8SUjVx5Qf5qkR-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Microsoft Copilot logo and branding pictured on a smartphone screen, with smartphone placed on top of a laptop keyboard.]]></media:description>                                                            <media:text><![CDATA[Microsoft Copilot logo and branding pictured on a smartphone screen, with smartphone placed on top of a laptop keyboard.]]></media:text>
                                <media:title type="plain"><![CDATA[Microsoft Copilot logo and branding pictured on a smartphone screen, with smartphone placed on top of a laptop keyboard.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/6FdBRNc4i8SUjVx5Qf5qkR-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Microsoft has announced that Copilot Cowork is now generally available for users globally, following a <a href="https://www.itpro.com/technology/artificial-intelligence/microsoft-is-rolling-out-copilot-cowork-to-more-customers">beta period via the tech giant’s Frontier program</a>.</p><p><a href="https://www.itpro.com/technology/artificial-intelligence/anthropics-claude-cowork-tool-is-coming-to-microsoft-copilot">First announced in March this year</a>, Copilot Cowork marks a significant milestone for Microsoft 365 Copilot users, integrating Anthropic’s highly popular tool within its flagship product. </p><p>According to Microsoft, Initial testing of the platform by a host of major companies, including Accenture, has proved highly successful. </p><p>In a <a href="https://www.microsoft.com/en-us/microsoft-365/blog/2026/06/16/copilot-cowork-is-now-generally-available/" target="_blank"><u>blog post</u></a> detailing the launch, Charles Lamanna, Microsoft’s EVP for Copilot, Agents, and Platform, said Cowork is the “fastest growing feature” in the history of the firm’s Frontier program. </p><p>“Cowork has among the highest user satisfaction of any Copilot or agent experience we have shipped,” he wrote.</p><p>“We learned from what we saw, engaged with you along the way, and used everything we heard to improve quality and add new features, including model choice, extensibility through plugins, and new cost management controls. Read more below and watch the full demo.”</p><p>Here’s what users can expect with Copilot Cowork. </p><h2 id="what-is-copilot-cowork">What is Copilot Cowork?</h2><p>Copilot Cowork is an integrated version of Anthropic’s Claude Cowork, which launched in January this year. The platform provides users with industry-specific AI capabilities to automate tasks across a range of professions. </p><p>Users can direct agents to carry out tasks on their behalf - such as file management or sending emails - by giving bots access to their files. The company described the new platform as essentially providing every worker with a “specialist” for their individual role. </p><p>Notably, the launch of the <a href="https://www.itpro.com/technology/artificial-intelligence/why-anthropic-sent-software-stocks-into-freefall">Cowork sparked a sell-off in the software market</a> amid fears the <a href="https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today">AI tool</a> could render dedicated software services obsolete. </p><h2 id="copilot-cowork-features-and-capabilities">Copilot Cowork features and capabilities</h2><p>Users at Microsoft shops can expect all the typical capabilities offered by Cowork, albeit with a few new features, according to Lamanna.</p><p>Notably, users will be able to choose which model they use with the service. Copilot Cowork currently runs on Anthropic models such as Claude Opus 4.8 and <a href="https://www.itpro.com/security/anthropic-claude-opus-claude-mythos-cyber-capabilities">Claude Sonnet 4.6</a>.</p><p>Customers will also be able to use GPT-5.5 further down the line with the launch of Cowork 1, the company revealed. According to Microsoft, this will deliver “optimal” cost and quality, and is intended for “enterprise-grade use”. </p><p>The tech giant noted that tweaking has helped remove model bias. </p><p>“It’s designed to handle everyday Copilot tasks at a substantially lower cost, making it a strong option for cost-sensitive workloads,” Lamanna wrote. </p><p>New partner plugins have also been added to Copilot Cowork, according to Microsoft, with a host of partners such as Miro and <a href="https://www.itpro.com/business/business-strategy/monday-com-promotes-ben-barnett-to-lead-emea-growth-strategy">Monday.com</a> already available. </p><p>Other partner plugins include: </p><ul><li>Enosix</li><li>Harvey</li><li>LSEG</li><li>Moodys</li><li>Morningstar</li><li>S&amp;P Global Energy</li><li>TeamsMaestro</li></ul><p>Several more are coming soon, including plugins for Atlassian, Canva, Box, Databricks, and Adobe. </p><p>Microsoft also touted a number of security-focused features for Copilot Cowork, aimed at shoring up governance and compliance. Cowork, prompts, responses, and artifacts, for example, now flow through existing Microsoft 365 controls.</p><h2 id="copilot-cowork-pricing">Copilot Cowork pricing</h2><p>To use Copilot Cowork, users will need to go through the Microsoft 365 Copilot User Subscription License (USL). This means that users are billed on a usage-basis, and charges are based on what particular tasks they use these agents for. </p><p>In terms of costs, customers have two payment options: PayGo and P3 (Credit Pre-Purchase Plan). The former is aimed at customers aiming for flexibility in terms of how they use the service, with credits priced at $0.01. </p><p>P3, meanwhile, is for those who want to commit to a usage volume “in advance in exchange for a discount”. </p><p>The cost of Copilot Cowork comes at a tricky time for businesses globally, particularly <a href="https://www.itpro.com/technology/artificial-intelligence/ubers-eye-watering-ai-bill-shows-enterprises-are-still-measuring-ai-success-through-consumption-rather-than-outcomes-and-its-warping-our-perception-of-roi-and-productivity">since the rise of the ‘tokenmaxxing’ trend</a> which has resulted in some firms racking up sizable AI bills. </p><p>Cost control features are available for customers, however, with Microsoft noting that customers can “decide when Cowork turns on, who gets access, and how much can be spent”. </p><p>Spending limits will be available at the tenant, group, and user levels, meaning admins can control costs and usage rates. </p><p>“Admins create scoped billing policies and define budgets, including user-level caps set inside group policies,” the company noted. “Admins set the thresholds that matter for their organization and groups, and choose who gets notified when spend crosses them.”</p><p>In the event that users exceed usage limits, credit requests can be made to complete tasks. These can be requested from inside Cowork. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Forget tokenomics – agents are a personnel cost ]]></title>
                                                                                                <dc:content><![CDATA[ <p>If an organization is surprised to see IT costs explode when it starts running agents at scale, the problem might not be in the IT budget – it might be a category error. At least in the opinion of HPE CEO Antonio Neri.</p><p>“I don’t think of [AI agents] as an IT cost,” Neri tells <em>ITPro</em> at HPE Discover 2026. “I think about the cost of the workforce because, to me, an agent is no different than any other employee I have to hire.”</p><p>“So if I hire [a person], there is a cost to onboard him and make sure that he's productive by giving him the tools and everything else,” he says. “When I train an agent, it's going to cost me a number of tokens to train it to drive the best productivity. If I'm going to spend a million dollars to train an agent, it has to be way more productive than [the person]. Otherwise, why am I doing that?”</p><p>This turns on its head much of the conversation around agents, tokenmaxxing, and costs as agentic AI starts to eclipse generative in organizational use.</p><h2 id="ai-sticker-shock">AI sticker shock</h2><p><em>ITPro</em> <a href="https://www.itpro.com/technology/artificial-intelligence/could-rising-token-costs-boost-interest-in-on-premises-hardware"><u>wrote in May</u></a> about “tokenomics” and so-called sticker-shock when the AI bill finally came due for many businesses. Famously, Uber used up its entire annual AI budget in four months and has now capped the amount each employee can spend on agents per month.</p><p>One answer that’s been given to the problem is a shift to on-premises IT rather than ‘cloud-based’ AI like Anthropic’s Claude Code or OpenAI’s Codex, and moving agent costs from OpEx back to CapEx. For an enterprise hardware company like HPE, surely this should be an opportunity to increase hardware sales?</p><p>“Sure,” says Neri. “But we think about it as a total cost of [the] workforce, not as an IT expense.”</p><p>And is this something that he’s educating his customers about, or do they see it another way?</p><p>“Yes, I am,” he says. “I’m spending more than 50% of my time with our customers, and when I bring that concept to life for them, they say ‘ok, ok!’”</p><p>“I’d rather give a little bit more budget to IT to reduce my entire cost of the workforce, which is – depending on what you do – probably one of the biggest cost levers,” he continues. “I have only two large cost drivers: people and materials. The materials, in networking, we have a lot of that under control because we have the IP, but a lot of servers, obviously, CPUs, and GPUs I don’t control.</p><p>“[The people] I have 65,000 employees. It’s not about reducing employees, which in some cases will happen, but it’s about making those 65,000 employees way more productive with the aid of AI by freeing up capacity for the humans to do more value-added [work] while the agentic models do more work at much bigger scale. I think about that, the total cost of [the] workforce”.</p><h2 id="human-ai-interaction">Human / AI interaction</h2><p>This is an idea that’s been seeded here and there throughout the conference. In Rami Rahim’s keynote, which took place just a couple of hours after Neri’s, one of the demonstrations floated the idea of taking humans out of the loop once a process has been approved.</p><p>Katrina Pickett, senior product manager, Juniper Networks,  showed how HPE Marvis might help detect and remediate a problem with a sluggish network. The conversation between person and machine was in natural language, but the fixing was agentic. At the point that the fix was applied, Marvis asked if Pickett wanted to have an agent apply the same fix automatically in the future, without human intervention, which would present a time saving.</p><p>Human-in-the-loop (HTL), however, has been a key mantra of not just AI but automation more generally. Is it time to take another look at this attitude?</p><p>“Well, once you’re confident about [what’s happening], then you can remove the human from the loop and just put on the loop,” says Neri. “I think there needs to be some sort of oversight, but not doing the actual work.”</p><p>“In the case of networking, a lot of things were done manually through [the] command line interface, by logging in the console, and putting hands on a keyboard,” Neri continues. </p><p>“The scale is now so big that humans cannot possibly manage that scale. Therefore, there is no other choice than to automate things and use artificial intelligence to manage massive amounts of data.”</p><p>Naturally, there are caveats. The story of the company that had <a href="https://www.tomshardware.com/tech-industry/artificial-intelligence/claude-powered-ai-coding-agent-deletes-entire-company-database-in-9-seconds-backups-zapped-after-cursor-tool-powered-by-anthropics-claude-goes-rogue"><u>its entire codebase and all backups erased by a rogue agent</u></a> is quickly passing from fact into legend; a cautionary tale for any business thinking of letting agents roam free in their network.</p><p>“I always say, you don’t get what you expect, only what you inspect [so] I think there needs to be some sort of oversight,” Neri says. “Human beings make mistakes, agents may make the same mistake, right? But as you’re doing it at a bigger scale, it may have a huge impact.”</p><p>Mistakes happening are “inevitable”, says Neri. Ultimately, though, there has to be an element of trust that the technology will work as intended. Guardrails can help with this, and, from Neri’s perspective, what HPE is doing with GreenLake Intelligence and Marvis Actions means customers can deploy agents with greater confidence to whatever extent they wish.</p><p>HPE still makes everything it always did – it has storage appliances, and it has servers a-plenty, and they’re popular too. According to the company’s most recent earnings call, traditional server orders are booming, growing triple digits year-on-year, while Alletra Storage MP orders also increased triple digits.</p><p>The conversation, however, will always revert to networking when it comes to visions of and for the future.</p><p>“There was a question in a previous session [that was posed] to Rami (Rahim), which said how far are we to really being fully autonomous, and we believe it’s within the next 12 months,” says Neri. </p><p>“It’ s not going to be everywhere across the network, but there are many aspects of the network just running itself.”</p><p>“Our opportunities are in the three domains: networking, cloud, and AI. Networking is a huge opportunity; we see it is. And it’s driven both at the edge to the campus and branch, the on ramp into the AI. You need that connectivity.”</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/business-strategy/forget-tokenomics-agents-are-a-personnel-cost</link>
                                                                            <description>
                            <![CDATA[ Looking at AI agents as an IT cost is to misunderstand what they are, says HPE chief ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">iaSsRbSdLpswNAxTyiuL7Q</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/oeETfn93AdMKR5VBye8cvB-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Thu, 18 Jun 2026 11:33:16 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 15:00:12 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ jane.mccallion@futurenet.com (Jane McCallion) ]]></author>                    <dc:creator><![CDATA[ Jane McCallion ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Wq9nnLr7TNkY8gyBRb7YsA-320-70.jpeg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Jane is managing editor at ITPro and ChannelPro. She started out with the brands as a staff writer specializing in cloud computing before going on to become senior writer and reports editor, managing the content and creation of ITPro’s quarterly whitepapers. During this time, she broadened her expertise to include cybersecurity, data centers and enterprise IT infrastructure. In 2016, she became features editor, managing a pool of freelance and internal writers, while continuing to specialize in enterprise IT infrastructure, data centers, and business strategy.&lt;/p&gt;&lt;p&gt;In October 2021, she became the sites’ deputy editor, before moving to the role of managing editor in June 2024. Although she now has a more strategic role,  she is still a specialist in enterprise IT infrastructure, business strategy, and cybersecurity.&lt;/p&gt;&lt;p&gt;Jane holds an MA in journalism from Goldsmiths, University of London, and a BA in Applied Languages from the University of Portsmouth. She is fluent in French and Spanish, and has written features in both languages.&lt;/p&gt;&lt;p&gt;Prior to joining ITPro, Jane was a freelance business journalist writing as both Jane McCallion and Jane Bordenave for titles such as European CEO, World Finance, and Business Excellence Magazine.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/oeETfn93AdMKR5VBye8cvB-1920-80.jpg">
                                                            <media:credit><![CDATA[HPE]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Antonio Neri, CEO at HPE, speaking live onstage at the Sphere in Las Vegas for HPE Discover 2024.]]></media:description>                                                            <media:text><![CDATA[Antonio Neri, CEO at HPE, speaking live onstage at the Sphere in Las Vegas for HPE Discover 2024.]]></media:text>
                                <media:title type="plain"><![CDATA[Antonio Neri, CEO at HPE, speaking live onstage at the Sphere in Las Vegas for HPE Discover 2024.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/oeETfn93AdMKR5VBye8cvB-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>If an organization is surprised to see IT costs explode when it starts running agents at scale, the problem might not be in the IT budget – it might be a category error. At least in the opinion of HPE CEO Antonio Neri.</p><p>“I don’t think of [AI agents] as an IT cost,” Neri tells <em>ITPro</em> at HPE Discover 2026. “I think about the cost of the workforce because, to me, an agent is no different than any other employee I have to hire.”</p><p>“So if I hire [a person], there is a cost to onboard him and make sure that he's productive by giving him the tools and everything else,” he says. “When I train an agent, it's going to cost me a number of tokens to train it to drive the best productivity. If I'm going to spend a million dollars to train an agent, it has to be way more productive than [the person]. Otherwise, why am I doing that?”</p><p>This turns on its head much of the conversation around agents, tokenmaxxing, and costs as agentic AI starts to eclipse generative in organizational use.</p><h2 id="ai-sticker-shock">AI sticker shock</h2><p><em>ITPro</em> <a href="https://www.itpro.com/technology/artificial-intelligence/could-rising-token-costs-boost-interest-in-on-premises-hardware"><u>wrote in May</u></a> about “tokenomics” and so-called sticker-shock when the AI bill finally came due for many businesses. Famously, Uber used up its entire annual AI budget in four months and has now capped the amount each employee can spend on agents per month.</p><p>One answer that’s been given to the problem is a shift to on-premises IT rather than ‘cloud-based’ AI like Anthropic’s Claude Code or OpenAI’s Codex, and moving agent costs from OpEx back to CapEx. For an enterprise hardware company like HPE, surely this should be an opportunity to increase hardware sales?</p><p>“Sure,” says Neri. “But we think about it as a total cost of [the] workforce, not as an IT expense.”</p><p>And is this something that he’s educating his customers about, or do they see it another way?</p><p>“Yes, I am,” he says. “I’m spending more than 50% of my time with our customers, and when I bring that concept to life for them, they say ‘ok, ok!’”</p><p>“I’d rather give a little bit more budget to IT to reduce my entire cost of the workforce, which is – depending on what you do – probably one of the biggest cost levers,” he continues. “I have only two large cost drivers: people and materials. The materials, in networking, we have a lot of that under control because we have the IP, but a lot of servers, obviously, CPUs, and GPUs I don’t control.</p><p>“[The people] I have 65,000 employees. It’s not about reducing employees, which in some cases will happen, but it’s about making those 65,000 employees way more productive with the aid of AI by freeing up capacity for the humans to do more value-added [work] while the agentic models do more work at much bigger scale. I think about that, the total cost of [the] workforce”.</p><h2 id="human-ai-interaction">Human / AI interaction</h2><p>This is an idea that’s been seeded here and there throughout the conference. In Rami Rahim’s keynote, which took place just a couple of hours after Neri’s, one of the demonstrations floated the idea of taking humans out of the loop once a process has been approved.</p><p>Katrina Pickett, senior product manager, Juniper Networks,  showed how HPE Marvis might help detect and remediate a problem with a sluggish network. The conversation between person and machine was in natural language, but the fixing was agentic. At the point that the fix was applied, Marvis asked if Pickett wanted to have an agent apply the same fix automatically in the future, without human intervention, which would present a time saving.</p><p>Human-in-the-loop (HTL), however, has been a key mantra of not just AI but automation more generally. Is it time to take another look at this attitude?</p><p>“Well, once you’re confident about [what’s happening], then you can remove the human from the loop and just put on the loop,” says Neri. “I think there needs to be some sort of oversight, but not doing the actual work.”</p><p>“In the case of networking, a lot of things were done manually through [the] command line interface, by logging in the console, and putting hands on a keyboard,” Neri continues. </p><p>“The scale is now so big that humans cannot possibly manage that scale. Therefore, there is no other choice than to automate things and use artificial intelligence to manage massive amounts of data.”</p><p>Naturally, there are caveats. The story of the company that had <a href="https://www.tomshardware.com/tech-industry/artificial-intelligence/claude-powered-ai-coding-agent-deletes-entire-company-database-in-9-seconds-backups-zapped-after-cursor-tool-powered-by-anthropics-claude-goes-rogue"><u>its entire codebase and all backups erased by a rogue agent</u></a> is quickly passing from fact into legend; a cautionary tale for any business thinking of letting agents roam free in their network.</p><p>“I always say, you don’t get what you expect, only what you inspect [so] I think there needs to be some sort of oversight,” Neri says. “Human beings make mistakes, agents may make the same mistake, right? But as you’re doing it at a bigger scale, it may have a huge impact.”</p><p>Mistakes happening are “inevitable”, says Neri. Ultimately, though, there has to be an element of trust that the technology will work as intended. Guardrails can help with this, and, from Neri’s perspective, what HPE is doing with GreenLake Intelligence and Marvis Actions means customers can deploy agents with greater confidence to whatever extent they wish.</p><p>HPE still makes everything it always did – it has storage appliances, and it has servers a-plenty, and they’re popular too. According to the company’s most recent earnings call, traditional server orders are booming, growing triple digits year-on-year, while Alletra Storage MP orders also increased triple digits.</p><p>The conversation, however, will always revert to networking when it comes to visions of and for the future.</p><p>“There was a question in a previous session [that was posed] to Rami (Rahim), which said how far are we to really being fully autonomous, and we believe it’s within the next 12 months,” says Neri. </p><p>“It’ s not going to be everywhere across the network, but there are many aspects of the network just running itself.”</p><p>“Our opportunities are in the three domains: networking, cloud, and AI. Networking is a huge opportunity; we see it is. And it’s driven both at the edge to the campus and branch, the on ramp into the AI. You need that connectivity.”</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Databricks launches AI co-worker, Genie One ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Databricks has released an <a href="https://www.itpro.com/technology/artificial-intelligence/ai-assistants-are-tools-not-co-workers">AI coworker</a> for business teams, including marketing, finance, and sales.</p><p>Genie One, available on the web, iOS, and Android, is designed to automate work, answer questions, and take action using structured or unstructured, analytical or operational company data.</p><p>Visual interfaces include interactive charts and graphs, and teams can set up alerts for always-on monitoring, schedule tasks, create repeatable skills, and take action with MCP tools. </p><p>The product is based on the newly-launched Genie Ontology, which, said the firm, brings together an organization's entire data. It's a self-improving context layer that automatically extracts and continuously updates business knowledge from Databricks, as well as AI tools and connected workplace apps. </p><p>It can access curated, authoritative data through <a href="https://www.itpro.com/sql/30242/what-is-sql">SQL</a>, rather than having to reason from fragments spread across documents, said the firm. </p><p>"Most enterprise AI today is just guessing with false confidence. That is not good enough for business. If you're a CFO and AI can't tell you why margins changed, or you're a sales leader, and it can't find your next upsell, that's not an AI problem, that's a context problem," said Ali Ghodsi, co-founder and CEO of Databricks. </p><p>"Genie Ontology continuously learns context from data everywhere, so our answers are much faster and our agents are more accurate. That's the difference between an AI chatbot and an agentic coworker who knows your business inside out – every metric, every data source, every answer."</p><p>Genie connects to all major AI tools, said Databricks, as well as more than 50 popular apps and data systems for business users across databases, files, tickets, chats, and meetings. Integrations include <a href="https://www.itpro.com/cloud-storage/24098/google-drive-review">Google Drive</a>, Jira, <a href="https://www.itpro.com/collaboration/33647/slack-review-free-your-business-comms">Slack</a>, Confluence, SharePoint, and more.</p><p>It can, said the firm, retrieve real answers from governed data and take action, with higher accuracy, reduced latency, and lower costs. And with Genie Agents and Genie App Builder, teams across the business can create reusable agents and applications – all connected to their data with access controls, permissions, and cost governance built in.</p><p>Genie Agents allow teams to save any Genie conversation as a reusable agent that inherits the conversation's memory, including its sources, instructions, and behavior, so coworkers can repeat trusted workflows across teams. Staff can also create and share agent skills with teammates for consistent, repeatable answer formats and workflows.</p><p> Genie App Builder, meanwhile, is a fully managed vibe coding environment built for the enterprise. Teams can upload business context, with Genie App Builder generating a live build plan and working app preview connected to real, governed enterprise data. These applications – built for internal teams or customers – have Unity Catalog permissions and access controls built in from the start.</p><p>Genie Code is an <a href="https://www.itpro.com/technology/artificial-intelligence/should-workers-prepare-to-become-ai-agent-bosses">AI agent</a> that helps data teams plan, build, and run data engineering, <a href="https://www.itpro.com/strategy/28071/what-is-machine-learning">machine learning</a>, and analytics workflows, and now includes a dedicated workspace to track progress, review steps, and easily switch between threads to work across projects.</p><p>And Genie ZeroOps is a new background agent built into Databricks that autonomously monitors, investigates, and proposes fixes for data and AI assets such as pipelines, jobs, tables, ML models and more.</p><p>"At Foot Locker, Genie Agents are transforming how we lead. They provide our executives and business teams with a centralized space to harness AI-driven insights across every North American banner we operate," said Krish Lakshminarayanan, VP, AI, data &amp; analytics, enterprise architecture at Databricks customer Foot Locker. </p><p>"As we scale Genie to the enterprise, it's reshaping the way our business interacts with data and makes the decisions that matter most."</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/databricks-launches-ai-co-worker-genie-one</link>
                                                                            <description>
                            <![CDATA[ The AI program is designed to help business teams manage workflows and automate work-related tasks ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">XuRUGfompCQTY5Fo8WULSk</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/sNUChcxSuBpBbbYDRyj3ZS-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 17 Jun 2026 11:35:30 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:19 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                                    <dc:creator><![CDATA[ Emma Woollacott ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/aWfskavxoVSMDy6cDWtYmJ-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/sNUChcxSuBpBbbYDRyj3ZS-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Databricks logo and branding pictured on a tablet screen against vibrant multi-colored background.]]></media:description>                                                            <media:text><![CDATA[Databricks logo and branding pictured on a tablet screen against vibrant multi-colored background.]]></media:text>
                                <media:title type="plain"><![CDATA[Databricks logo and branding pictured on a tablet screen against vibrant multi-colored background.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/sNUChcxSuBpBbbYDRyj3ZS-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Databricks has released an <a href="https://www.itpro.com/technology/artificial-intelligence/ai-assistants-are-tools-not-co-workers">AI coworker</a> for business teams, including marketing, finance, and sales.</p><p>Genie One, available on the web, iOS, and Android, is designed to automate work, answer questions, and take action using structured or unstructured, analytical or operational company data.</p><p>Visual interfaces include interactive charts and graphs, and teams can set up alerts for always-on monitoring, schedule tasks, create repeatable skills, and take action with MCP tools. </p><p>The product is based on the newly-launched Genie Ontology, which, said the firm, brings together an organization's entire data. It's a self-improving context layer that automatically extracts and continuously updates business knowledge from Databricks, as well as AI tools and connected workplace apps. </p><p>It can access curated, authoritative data through <a href="https://www.itpro.com/sql/30242/what-is-sql">SQL</a>, rather than having to reason from fragments spread across documents, said the firm. </p><p>"Most enterprise AI today is just guessing with false confidence. That is not good enough for business. If you're a CFO and AI can't tell you why margins changed, or you're a sales leader, and it can't find your next upsell, that's not an AI problem, that's a context problem," said Ali Ghodsi, co-founder and CEO of Databricks. </p><p>"Genie Ontology continuously learns context from data everywhere, so our answers are much faster and our agents are more accurate. That's the difference between an AI chatbot and an agentic coworker who knows your business inside out – every metric, every data source, every answer."</p><p>Genie connects to all major AI tools, said Databricks, as well as more than 50 popular apps and data systems for business users across databases, files, tickets, chats, and meetings. Integrations include <a href="https://www.itpro.com/cloud-storage/24098/google-drive-review">Google Drive</a>, Jira, <a href="https://www.itpro.com/collaboration/33647/slack-review-free-your-business-comms">Slack</a>, Confluence, SharePoint, and more.</p><p>It can, said the firm, retrieve real answers from governed data and take action, with higher accuracy, reduced latency, and lower costs. And with Genie Agents and Genie App Builder, teams across the business can create reusable agents and applications – all connected to their data with access controls, permissions, and cost governance built in.</p><p>Genie Agents allow teams to save any Genie conversation as a reusable agent that inherits the conversation's memory, including its sources, instructions, and behavior, so coworkers can repeat trusted workflows across teams. Staff can also create and share agent skills with teammates for consistent, repeatable answer formats and workflows.</p><p> Genie App Builder, meanwhile, is a fully managed vibe coding environment built for the enterprise. Teams can upload business context, with Genie App Builder generating a live build plan and working app preview connected to real, governed enterprise data. These applications – built for internal teams or customers – have Unity Catalog permissions and access controls built in from the start.</p><p>Genie Code is an <a href="https://www.itpro.com/technology/artificial-intelligence/should-workers-prepare-to-become-ai-agent-bosses">AI agent</a> that helps data teams plan, build, and run data engineering, <a href="https://www.itpro.com/strategy/28071/what-is-machine-learning">machine learning</a>, and analytics workflows, and now includes a dedicated workspace to track progress, review steps, and easily switch between threads to work across projects.</p><p>And Genie ZeroOps is a new background agent built into Databricks that autonomously monitors, investigates, and proposes fixes for data and AI assets such as pipelines, jobs, tables, ML models and more.</p><p>"At Foot Locker, Genie Agents are transforming how we lead. They provide our executives and business teams with a centralized space to harness AI-driven insights across every North American banner we operate," said Krish Lakshminarayanan, VP, AI, data &amp; analytics, enterprise architecture at Databricks customer Foot Locker. </p><p>"As we scale Genie to the enterprise, it's reshaping the way our business interacts with data and makes the decisions that matter most."</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ HPE AI Factory adds features to improve your experience with agents ]]></title>
                                                                                                <dc:content><![CDATA[ <p>HPE and Nvidia have announced new features for <a href="https://www.itpro.com/technology/artificial-intelligence/hpes-ai-factory-line-just-got-a-huge-update">HPE AI Factory</a>, aimed at improving customers' experience with <a href="https://www.itpro.com/technology/artificial-intelligence/ai-agents-arent-cutting-it-in-customer-service">AI agents</a>.</p><p>Announced at <a href="https://www.itpro.com/hardware/live/hpe-discover-2026-live-follow-for-the-latest-news-and-announcements-from-the-day-1-keynotes">HPE Discover 2026</a>, HPE AI Factory, which is part of Nvidia AI Computing by HPE, will now offer the Nvidia Agent Toolkit software, which includes Nvidia Nemotron open models, Nvidia NemoClaw (the chip maker's answer to OpenClaw), and Nvidia OpenShell secure runtime.</p><p>The new capabilities will, the companies say, help customers deploy enterprise-ready AI with greater observability, efficiency, and control.</p><p>There are also new hardware additions to <a href="https://www.itpro.com/cloud/private-cloud/were-meeting-customers-where-they-are-hpe-expands-private-cloud-ai-service-with-new-sovereignty-controls-air-gapped-features">HPE Private Cloud AI</a>, which now has HPE ProLiant Compute DL394 Gen12 servers with Nvidia Vera CPU for, in the words of the company, "a compute-optimized foundation for agentic AI and high-performance data processing, including security and management features".</p><p>On the security side, the update also includes new Zerto Software capabilities – HPE acquired Zerto in 2021 – which the company says will help customers identify when a "rogue agent" starts carrying out actions that it shouldn't. If this happens, the <a href="https://www.itpro.com/business-operations/sales/366831/zerto-bolsters-leadership-team-with-vp-of-emea">Zerto</a> capabilities can "use continuous data protection to rewind to a clean state".</p><p>In addition, HPE Private Cloud AI also now has support for secure local agent registration, which the company says will allow customers to approve AI models, skills, and tools while following centralized governance and security policies.</p><p>Finally, HPE Private Cloud AI also has HPE Alletra Storage MP X1000, which enables customers to apply metadata and governance policies that prepare data for AI applications. According to HPE, this allows businesses to boost token throughput by up to 20% and improve prompt processing efficiency.</p><p>Of the announcement, HPE president and CEO, Antonio Neri, said: "Across networking, servers, storage, and software, HPE is delivering full-stack AI solutions with NVIDIA that build the foundation for agentic enterprises, helping customers move from experimentation to production with control and confidence."</p><p><em>ITPro's managing editor Jane McCallion is at HPE Discover 2026, and you can follow all the latest announcements and analysis from the event </em><a href="https://www.itpro.com/hardware/live/hpe-discover-2026-live-follow-for-the-latest-news-and-announcements-from-the-day-1-keynotes"><em>here</em></a><em>. </em></p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/hpe-ai-factory-adds-features-to-improve-your-experience-with-agents</link>
                                                                            <description>
                            <![CDATA[ The HPE AI Factory will now offer the Nvidia Agent Toolkit software, such as Nvidia Nemotron open models and Nvidia NemoClaw ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">8MFeyLW8SZBNGcQzLrEx4j</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/G9uckEYVZA3VSyAguQddVd-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Tue, 16 Jun 2026 16:30:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:33:36 +0000</updated>
                                                                                                                                            <category><![CDATA[AI Infrastructure]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ jane.mccallion@futurenet.com (Jane McCallion) ]]></author>                    <dc:creator><![CDATA[ Jane McCallion ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Wq9nnLr7TNkY8gyBRb7YsA-320-70.jpeg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Jane is managing editor at ITPro and ChannelPro. She started out with the brands as a staff writer specializing in cloud computing before going on to become senior writer and reports editor, managing the content and creation of ITPro’s quarterly whitepapers. During this time, she broadened her expertise to include cybersecurity, data centers and enterprise IT infrastructure. In 2016, she became features editor, managing a pool of freelance and internal writers, while continuing to specialize in enterprise IT infrastructure, data centers, and business strategy.&lt;/p&gt;&lt;p&gt;In October 2021, she became the sites’ deputy editor, before moving to the role of managing editor in June 2024. Although she now has a more strategic role,  she is still a specialist in enterprise IT infrastructure, business strategy, and cybersecurity.&lt;/p&gt;&lt;p&gt;Jane holds an MA in journalism from Goldsmiths, University of London, and a BA in Applied Languages from the University of Portsmouth. She is fluent in French and Spanish, and has written features in both languages.&lt;/p&gt;&lt;p&gt;Prior to joining ITPro, Jane was a freelance business journalist writing as both Jane McCallion and Jane Bordenave for titles such as European CEO, World Finance, and Business Excellence Magazine.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/G9uckEYVZA3VSyAguQddVd-1920-80.jpg">
                                                            <media:credit><![CDATA[Future]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[A sign for the keynote for HPE Discover 2026]]></media:description>                                                            <media:text><![CDATA[A sign for the keynote for HPE Discover 2026]]></media:text>
                                <media:title type="plain"><![CDATA[A sign for the keynote for HPE Discover 2026]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/G9uckEYVZA3VSyAguQddVd-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>HPE and Nvidia have announced new features for <a href="https://www.itpro.com/technology/artificial-intelligence/hpes-ai-factory-line-just-got-a-huge-update">HPE AI Factory</a>, aimed at improving customers' experience with <a href="https://www.itpro.com/technology/artificial-intelligence/ai-agents-arent-cutting-it-in-customer-service">AI agents</a>.</p><p>Announced at <a href="https://www.itpro.com/hardware/live/hpe-discover-2026-live-follow-for-the-latest-news-and-announcements-from-the-day-1-keynotes">HPE Discover 2026</a>, HPE AI Factory, which is part of Nvidia AI Computing by HPE, will now offer the Nvidia Agent Toolkit software, which includes Nvidia Nemotron open models, Nvidia NemoClaw (the chip maker's answer to OpenClaw), and Nvidia OpenShell secure runtime.</p><p>The new capabilities will, the companies say, help customers deploy enterprise-ready AI with greater observability, efficiency, and control.</p><p>There are also new hardware additions to <a href="https://www.itpro.com/cloud/private-cloud/were-meeting-customers-where-they-are-hpe-expands-private-cloud-ai-service-with-new-sovereignty-controls-air-gapped-features">HPE Private Cloud AI</a>, which now has HPE ProLiant Compute DL394 Gen12 servers with Nvidia Vera CPU for, in the words of the company, "a compute-optimized foundation for agentic AI and high-performance data processing, including security and management features".</p><p>On the security side, the update also includes new Zerto Software capabilities – HPE acquired Zerto in 2021 – which the company says will help customers identify when a "rogue agent" starts carrying out actions that it shouldn't. If this happens, the <a href="https://www.itpro.com/business-operations/sales/366831/zerto-bolsters-leadership-team-with-vp-of-emea">Zerto</a> capabilities can "use continuous data protection to rewind to a clean state".</p><p>In addition, HPE Private Cloud AI also now has support for secure local agent registration, which the company says will allow customers to approve AI models, skills, and tools while following centralized governance and security policies.</p><p>Finally, HPE Private Cloud AI also has HPE Alletra Storage MP X1000, which enables customers to apply metadata and governance policies that prepare data for AI applications. According to HPE, this allows businesses to boost token throughput by up to 20% and improve prompt processing efficiency.</p><p>Of the announcement, HPE president and CEO, Antonio Neri, said: "Across networking, servers, storage, and software, HPE is delivering full-stack AI solutions with NVIDIA that build the foundation for agentic enterprises, helping customers move from experimentation to production with control and confidence."</p><p><em>ITPro's managing editor Jane McCallion is at HPE Discover 2026, and you can follow all the latest announcements and analysis from the event </em><a href="https://www.itpro.com/hardware/live/hpe-discover-2026-live-follow-for-the-latest-news-and-announcements-from-the-day-1-keynotes"><em>here</em></a><em>. </em></p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Salesforce buys Fin for $3.6bn ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Salesforce is set to acquire Fin for $3.6bn, giving the <a href="https://www.itpro.com/desktop-software/28214/what-is-crm">CRM</a> giant an AI agent for customer service. </p><p>Fin was founded in 2011 as Intercom, with the chatbot maker rebranding as Fin last month. Over the last few years, the company has focused on developing an AI agent using its own Apex model that's built specifically for customer support. </p><p>Salesforce has its own agentic AI known as <a href="https://www.itpro.com/technology/artificial-intelligence/salesforce-agentforce-field-service-ai-agents">Agentforce</a>, which was unveiled at the end of 2024. Earlier this year, <a href="https://www.itpro.com/technology/artificial-intelligence/salesforce-unified-customer-support-automation-with-agentforce-contact-center">Salesforce unveiled Agentforce Contact Centre</a>, pulling together its AI agent and the company's CRM. Salesforce said Agentforce has reached an annual run rate of $1.2bn in the latest quarter, up 205% year on year. </p><p>"We're thrilled to welcome Fin to Salesforce as we enable every company to become an agentic enterprise," said Marc Benioff, Chair and CEO of Salesforce, in a statement. "Fin brings proven agent technology, a deep commitment to customer success, and an incredible AI team that will complement Agentforce with powerful service agent capabilities."</p><p>The deal is expected to close in the fourth quarter, subject to regulatory clearance. </p><p>Salesforce has pushed AI agents as the future of customer service, with CEO Marc Benioff saying his company has <a href="https://www.itpro.com/technology/artificial-intelligence/marc-benioff-says-salesforce-has-already-cut-4-000-customer-support-staff-for-ai-agents-and-workforce-productivity-hasnt-dipped">slashed 4,000 customer service roles</a> in favour of AI agents so far. But one <a href="https://www.itpro.com/technology/artificial-intelligence/your-customers-arent-keen-on-that-customer-service-chatbot-you-introduced-heres-why">survey</a> from earlier this year suggests customers don't agree with the AI chatbots for customer service, with more than two-thirds not confident in the way businesses use generative AI to interact with them, and half saying the transactions don't work. </p><h2 id="find-a-win-win-for-customers">Find a win-win for customers</h2><p>Fin's AI agent is multichannel with support for chat, email, WhatsApp, Slack, and even the phone, covering all aspects of a customer journey from sales to service, with the company claiming it can update accounts, troubleshoot tech problems, and process refunds. </p><p>On its website, Fin claims to have industry-leading resolution rates of 76% across more than 12,000 customers, with that performance gaining a percentage point every month it's in use. </p><p>Fin's system is already in use by companies such as Asana and <a href="https://www.itpro.com/software/mondaycom-review-work-management-platform-works-best-when-you-pay-for-it">Monday.com</a> – as well as AI developer Anthropic. </p><p>"This is a major win for consumers of the world," said Eoghan McCabe, Chief Executive Officer and Co-Founder of Fin. "Our technology has defined this category and set the new standards for what great customer service looks like today. By joining forces with Salesforce, we can deploy it far and wide at a rate far faster than we could have ever achieved on our own."</p><p>Salesforce said in a statement that the aim was to give its customers more ways to deploy AI agents in their customer service operations, in particular, offering small businesses options with fast time-to-value that are easily deployable and offer measurable outcomes. </p><p>Plus, Salesforce gains Fin's AI team and its customer base of 30,000 companies. </p><h2 id="from-intercom-to-fin">From Intercom to Fin </h2><p>Intercom was started in 2011 by a quartet of founders – Eoghan McCabe, Des Traynor, Ciaran Lee, and David Barrett – and developed chatbots for companies to offer customer service, with the Intercom window ubiquitous in the bottom corner of websites.</p><p>Several years later, CEO Eoghan McCabe <a href="https://www.cnbc.com/2026/06/15/salesforce-ai-customer-service-fin-acquistion.html">was accused</a> of "unwanted advances" towards female employees, and though he apologised at the time, he stepped down the next year. COO Karen Peacock took over as CEO, but only briefly, as McCabe returned in 2022 at the request of the board – keeping the controversy continuing with <a href="https://x.com/eoghan/status/1799161733766062545/photo/1">public support</a> for US President Trump.  </p><p>Last month, Intercom rebranded as Fin, the name of the company's AI agent platform, though the Intercom brand was continuing as customer service software. "Fin is clearly our future, and the future of this new customer agent category," McCabe <a href="https://www.intercom.com/blog/today-intercom-becomes-fin/">said</a> at the time. </p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/acquisition/salesforce-buys-fin-for-usd3-6bn</link>
                                                                            <description>
                            <![CDATA[ CRM giant invests in AI agents with acquisition ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">objiq7ZpDmCerypcoUKUyG</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/7EyzKMnFPw68NkgcmUkAdU-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Tue, 16 Jun 2026 14:55:06 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:42:54 +0000</updated>
                                                                                                                                            <category><![CDATA[Acquisition]]></category>
                                                    <category><![CDATA[Business]]></category>
                                                                                                                    <dc:creator><![CDATA[ Nicole Kobie ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/8Y8JDDTQ7XDEk49FoAFP2S-320-70.png ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Nicole Kobie first started writing for ITPro in 2007. As a freelance journalist covering technology and business, Nicole&#039;s work includes  bylines in New Scientist, Wired, PC Pro and many more. &lt;/p&gt;&lt;p&gt;Nicole the author of a book about the history of technology, The Long History of the Future.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/7EyzKMnFPw68NkgcmUkAdU-1920-80.jpg">
                                                            <media:credit><![CDATA[ David Paul Morris/Bloomberg  ]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Salesforce CEO Marc Benioff pictured at Dreamforce 2024, held at the Moscone Center in San Francisco, California. ]]></media:description>                                                            <media:text><![CDATA[Salesforce CEO Marc Benioff pictured at Dreamforce 2024, held at the Moscone Center in San Francisco, California. ]]></media:text>
                                <media:title type="plain"><![CDATA[Salesforce CEO Marc Benioff pictured at Dreamforce 2024, held at the Moscone Center in San Francisco, California. ]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/7EyzKMnFPw68NkgcmUkAdU-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Salesforce is set to acquire Fin for $3.6bn, giving the <a href="https://www.itpro.com/desktop-software/28214/what-is-crm">CRM</a> giant an AI agent for customer service. </p><p>Fin was founded in 2011 as Intercom, with the chatbot maker rebranding as Fin last month. Over the last few years, the company has focused on developing an AI agent using its own Apex model that's built specifically for customer support. </p><p>Salesforce has its own agentic AI known as <a href="https://www.itpro.com/technology/artificial-intelligence/salesforce-agentforce-field-service-ai-agents">Agentforce</a>, which was unveiled at the end of 2024. Earlier this year, <a href="https://www.itpro.com/technology/artificial-intelligence/salesforce-unified-customer-support-automation-with-agentforce-contact-center">Salesforce unveiled Agentforce Contact Centre</a>, pulling together its AI agent and the company's CRM. Salesforce said Agentforce has reached an annual run rate of $1.2bn in the latest quarter, up 205% year on year. </p><p>"We're thrilled to welcome Fin to Salesforce as we enable every company to become an agentic enterprise," said Marc Benioff, Chair and CEO of Salesforce, in a statement. "Fin brings proven agent technology, a deep commitment to customer success, and an incredible AI team that will complement Agentforce with powerful service agent capabilities."</p><p>The deal is expected to close in the fourth quarter, subject to regulatory clearance. </p><p>Salesforce has pushed AI agents as the future of customer service, with CEO Marc Benioff saying his company has <a href="https://www.itpro.com/technology/artificial-intelligence/marc-benioff-says-salesforce-has-already-cut-4-000-customer-support-staff-for-ai-agents-and-workforce-productivity-hasnt-dipped">slashed 4,000 customer service roles</a> in favour of AI agents so far. But one <a href="https://www.itpro.com/technology/artificial-intelligence/your-customers-arent-keen-on-that-customer-service-chatbot-you-introduced-heres-why">survey</a> from earlier this year suggests customers don't agree with the AI chatbots for customer service, with more than two-thirds not confident in the way businesses use generative AI to interact with them, and half saying the transactions don't work. </p><h2 id="find-a-win-win-for-customers">Find a win-win for customers</h2><p>Fin's AI agent is multichannel with support for chat, email, WhatsApp, Slack, and even the phone, covering all aspects of a customer journey from sales to service, with the company claiming it can update accounts, troubleshoot tech problems, and process refunds. </p><p>On its website, Fin claims to have industry-leading resolution rates of 76% across more than 12,000 customers, with that performance gaining a percentage point every month it's in use. </p><p>Fin's system is already in use by companies such as Asana and <a href="https://www.itpro.com/software/mondaycom-review-work-management-platform-works-best-when-you-pay-for-it">Monday.com</a> – as well as AI developer Anthropic. </p><p>"This is a major win for consumers of the world," said Eoghan McCabe, Chief Executive Officer and Co-Founder of Fin. "Our technology has defined this category and set the new standards for what great customer service looks like today. By joining forces with Salesforce, we can deploy it far and wide at a rate far faster than we could have ever achieved on our own."</p><p>Salesforce said in a statement that the aim was to give its customers more ways to deploy AI agents in their customer service operations, in particular, offering small businesses options with fast time-to-value that are easily deployable and offer measurable outcomes. </p><p>Plus, Salesforce gains Fin's AI team and its customer base of 30,000 companies. </p><h2 id="from-intercom-to-fin">From Intercom to Fin </h2><p>Intercom was started in 2011 by a quartet of founders – Eoghan McCabe, Des Traynor, Ciaran Lee, and David Barrett – and developed chatbots for companies to offer customer service, with the Intercom window ubiquitous in the bottom corner of websites.</p><p>Several years later, CEO Eoghan McCabe <a href="https://www.cnbc.com/2026/06/15/salesforce-ai-customer-service-fin-acquistion.html">was accused</a> of "unwanted advances" towards female employees, and though he apologised at the time, he stepped down the next year. COO Karen Peacock took over as CEO, but only briefly, as McCabe returned in 2022 at the request of the board – keeping the controversy continuing with <a href="https://x.com/eoghan/status/1799161733766062545/photo/1">public support</a> for US President Trump.  </p><p>Last month, Intercom rebranded as Fin, the name of the company's AI agent platform, though the Intercom brand was continuing as customer service software. "Fin is clearly our future, and the future of this new customer agent category," McCabe <a href="https://www.intercom.com/blog/today-intercom-becomes-fin/">said</a> at the time. </p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Enterprises are bullish on agents, but can't get projects over the finishing line – here's why ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Most companies are ramping up agentic AI adoption, but it largely remains in the pilot stage with no return on investment (ROI) on the horizon. </p><p>That's according to a new report from Forrester, which found that three quarters of enterprise leaders said they were adopting the technology, but only a small minority are using it operationally beyond "agentish" chatbots.</p><p>"Agentic AI is no longer theoretical – it’s technically real in 2026 – but most enterprises are still unprepared to operationalize it," said Brian Hopkins, VP and analyst at Forrester. </p><p>The research marks the latest in a string of studies highlighting an inability to move agentic AI projects past the pilot stage. A <a href="https://www.itpro.com/technology/artificial-intelligence/half-of-agentic-ai-projects-are-still-stuck-at-the-pilot-stage-but-thats-not-stopping-enterprises-from-ramping-up-investment"><u>study from Dynatrace</u></a> earlier this year found roughly half were stuck at this point within their adoption journey and unable to progress.</p><p>Notably, the Dynatrace report found this wasn’t putting IT leaders off investing in the technology, with a significant portion planning to increase investment despite a lack of concrete returns. </p><h2 id="agents-aren-t-chatbots">Agents aren't chatbots</h2><p>In a <a href="https://www.forrester.com/blogs/the-state-of-agentic-ai-in-2026-companies-are-chasing-few-are-catching/"><u>blog post</u></a> detailing the Forrester research analysts noted that AI agents had arrived but enterprises simply weren't ready – which shouldn't surprise anyone, as "we've seen this story before". </p><p>Part of the problem, Forrester said, was treating agents like chatbots when they should instead be viewed as a distributed system that demands infrastructure and support. </p><p>Companies can't simply stitch together isolated agents without shared data, registries or routing, the consultancy noted. </p><p>"The gap we see isn’t about models or ambition; it’s about orchestration, control, and trust," Hopkins added. "Companies that treat <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">agentic AI</a> as a feature experiment will stay stuck in pilots, while those that invest in agent‑native design, executable governance, and nonhuman identity will be the ones that actually capture value at scale."</p><h2 id="expensive-chase">Expensive chase</h2><p>The lack of ROI is one reason enterprises are slow to escape pilot mode and fail to scale, according to Forrester. Simply put, the technology isn’t quite living up to some expectations, and that’s spooking IT leaders. </p><p>"ROI uncertainty traps enterprise ambition in pilot mode because most companies can’t justify production beyond narrow efficiency gains," the blog post added. "And platform confusion freezes commitment while teams argue over whether to bet on a SaaS agent, [a system integrator]-built system, or a custom build."</p><p>Forrester argues that a major cost is the "trust tax", as auditing adds costs to agentic rollouts – of course, many companies, in particular in finance or healthcare, are heavily regulated and with good reason. </p><p>"Every autonomous action has to be logged and defensible to an auditor, and right now that cost is too high," Forrester claimed. </p><p>Another challenge is <a href="https://www.itpro.com/security/do-risk-awareness-and-risk-management-strategies-actually-make-a-difference">risk management</a>, with Forrester admitting that AI security issues are a real threat that require new governance and identity policy that's "enforced as code rather than written down and hoped for."</p><h2 id="what-can-it-leaders-do">What can IT leaders do?</h2><p>Forrester said enterprises can stop chasing agentic AI and catch up with the technology by laying the right groundwork. Agentic AI needs a solid data foundation before IT leaders should even think about diving head-first. </p><p>"The companies pulling ahead aren’t the ones with the most agents," the company said in that blog post. "They’re the ones laying the track the train will run on."</p><p>Practically, that means investing in orchestration before adding agents, with Forrester noting that shared registries are necessary for agents to work inside conventional systems. </p><p>Plus, the analysts called for companies to redesign the work being done, not just the tooling. </p><p>"Agents bolted onto human-paced legacy workflows produce task savings, not step-change value," the post added. "Pick a few high-friction workflows and rebuild the roles and approvals around autonomy."</p><p>Notably, companies need to treat every agent as its own governed identity, with unique credentials, least privilege, full logging and a named owner to watch over it. </p><p>With those three key aspects in place, companies can begin to step away from pilots towards scaling, but even that should be done in stages with autonomy added only when appropriate controls are in place, Forrester added. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/most-enterprises-are-still-unprepared-to-operationalize-it-it-leaders-are-bullish-on-agents-but-keeping-falling-at-the-final-hurdle-heres-why</link>
                                                                            <description>
                            <![CDATA[ Forrester points to challenges scaling agentic AI, saying companies start rolling out the tech before they're ready to scale ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">t6xsofxvkF2mAzjCD58gGm</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/vu2r4p7vDPvLDwHLJR2HJB-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Thu, 11 Jun 2026 14:18:14 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:39 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                                    <dc:creator><![CDATA[ Nicole Kobie ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/8Y8JDDTQ7XDEk49FoAFP2S-320-70.png ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Nicole Kobie first started writing for ITPro in 2007. As a freelance journalist covering technology and business, Nicole&#039;s work includes  bylines in New Scientist, Wired, PC Pro and many more. &lt;/p&gt;&lt;p&gt;Nicole the author of a book about the history of technology, The Long History of the Future.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/vu2r4p7vDPvLDwHLJR2HJB-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Stock illustration of a male business person wearing a suit falling over a hurdle after jumping over two prior hurdles.]]></media:description>                                                            <media:text><![CDATA[Stock illustration of a male business person wearing a suit falling over a hurdle after jumping over two prior hurdles.]]></media:text>
                                <media:title type="plain"><![CDATA[Stock illustration of a male business person wearing a suit falling over a hurdle after jumping over two prior hurdles.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/vu2r4p7vDPvLDwHLJR2HJB-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Most companies are ramping up agentic AI adoption, but it largely remains in the pilot stage with no return on investment (ROI) on the horizon. </p><p>That's according to a new report from Forrester, which found that three quarters of enterprise leaders said they were adopting the technology, but only a small minority are using it operationally beyond "agentish" chatbots.</p><p>"Agentic AI is no longer theoretical – it’s technically real in 2026 – but most enterprises are still unprepared to operationalize it," said Brian Hopkins, VP and analyst at Forrester. </p><p>The research marks the latest in a string of studies highlighting an inability to move agentic AI projects past the pilot stage. A <a href="https://www.itpro.com/technology/artificial-intelligence/half-of-agentic-ai-projects-are-still-stuck-at-the-pilot-stage-but-thats-not-stopping-enterprises-from-ramping-up-investment"><u>study from Dynatrace</u></a> earlier this year found roughly half were stuck at this point within their adoption journey and unable to progress.</p><p>Notably, the Dynatrace report found this wasn’t putting IT leaders off investing in the technology, with a significant portion planning to increase investment despite a lack of concrete returns. </p><h2 id="agents-aren-t-chatbots">Agents aren't chatbots</h2><p>In a <a href="https://www.forrester.com/blogs/the-state-of-agentic-ai-in-2026-companies-are-chasing-few-are-catching/"><u>blog post</u></a> detailing the Forrester research analysts noted that AI agents had arrived but enterprises simply weren't ready – which shouldn't surprise anyone, as "we've seen this story before". </p><p>Part of the problem, Forrester said, was treating agents like chatbots when they should instead be viewed as a distributed system that demands infrastructure and support. </p><p>Companies can't simply stitch together isolated agents without shared data, registries or routing, the consultancy noted. </p><p>"The gap we see isn’t about models or ambition; it’s about orchestration, control, and trust," Hopkins added. "Companies that treat <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">agentic AI</a> as a feature experiment will stay stuck in pilots, while those that invest in agent‑native design, executable governance, and nonhuman identity will be the ones that actually capture value at scale."</p><h2 id="expensive-chase">Expensive chase</h2><p>The lack of ROI is one reason enterprises are slow to escape pilot mode and fail to scale, according to Forrester. Simply put, the technology isn’t quite living up to some expectations, and that’s spooking IT leaders. </p><p>"ROI uncertainty traps enterprise ambition in pilot mode because most companies can’t justify production beyond narrow efficiency gains," the blog post added. "And platform confusion freezes commitment while teams argue over whether to bet on a SaaS agent, [a system integrator]-built system, or a custom build."</p><p>Forrester argues that a major cost is the "trust tax", as auditing adds costs to agentic rollouts – of course, many companies, in particular in finance or healthcare, are heavily regulated and with good reason. </p><p>"Every autonomous action has to be logged and defensible to an auditor, and right now that cost is too high," Forrester claimed. </p><p>Another challenge is <a href="https://www.itpro.com/security/do-risk-awareness-and-risk-management-strategies-actually-make-a-difference">risk management</a>, with Forrester admitting that AI security issues are a real threat that require new governance and identity policy that's "enforced as code rather than written down and hoped for."</p><h2 id="what-can-it-leaders-do">What can IT leaders do?</h2><p>Forrester said enterprises can stop chasing agentic AI and catch up with the technology by laying the right groundwork. Agentic AI needs a solid data foundation before IT leaders should even think about diving head-first. </p><p>"The companies pulling ahead aren’t the ones with the most agents," the company said in that blog post. "They’re the ones laying the track the train will run on."</p><p>Practically, that means investing in orchestration before adding agents, with Forrester noting that shared registries are necessary for agents to work inside conventional systems. </p><p>Plus, the analysts called for companies to redesign the work being done, not just the tooling. </p><p>"Agents bolted onto human-paced legacy workflows produce task savings, not step-change value," the post added. "Pick a few high-friction workflows and rebuild the roles and approvals around autonomy."</p><p>Notably, companies need to treat every agent as its own governed identity, with unique credentials, least privilege, full logging and a named owner to watch over it. </p><p>With those three key aspects in place, companies can begin to step away from pilots towards scaling, but even that should be done in stages with autonomy added only when appropriate controls are in place, Forrester added. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Asana wants every enterprise to have an AI ‘chief of staff’ ]]></title>
                                                                                                <dc:content><![CDATA[ <p><a href="https://www.itpro.com/technology/artificial-intelligence/how-ai-agent-boss-is-reshaping-it-accountability">Asana</a> has unveiled a series of new features aimed at streamlining collaboration between human workers and agents as it looks to “solve the AI productivity gap” for enterprises. </p><p>Among these is Asana Dash, a new AI “chief of staff” designed to support employees and keep projects on track. </p><p>According to the firm, the tool helps users stay up-to-date with project goals and progress, compiling and summarizing relevant information such as email correspondence, messages between team members, and progress reports. </p><p>The new tool forms a key component within Asana’s new Agentic Work Management platform, unveiled at the company’s Work Innovation Summit in London this week. </p><p>This platform acts as an underlying “operating system” combining core products with the goal of consolidating workflows within a single domain. </p><p>Elsewhere, updates to Asana’s AI Teammates feature complement Dash, according to the firm. New integrations for Gmail, Outlook, Slack, HubSpot, Figma, and Canva were announced at the London summit along with a Skills library for “repeatable work”. </p><p>These two features are intended to work in-tandem, helping support project work, according to Asana. Dash can even assign actions to human workers or AI Teammates.</p><p>“It captures follow-ups from meetings, Slack threads, and email, turns them into structured work in the Work Graph, and connects users to the right AI Teammates for specific tasks and projects to move work forward,” the company said.</p><h2 id="tackling-the-ai-productivity-gap">Tackling the “AI productivity gap”</h2><p>The new features launched by Asana aim to tackle a common recurring pain point for enterprises using AI tools: <a href="https://www.itpro.com/business/business-strategy/ai-productivity-challenges-accenture-generating-impact-study">mismatched productivity gains</a>. </p><p>Research conducted by the firm shows that while 75% of knowledge workers now use AI in daily tasks, only 5% of companies report “meaningful” productivity improvements. </p><p>The company attributes this problem to four key areas. First and foremost, this includes an inability to select the “right agents” for specific individual needs and tasks. </p><p>Notably, the company said “agents aren’t team players,” adding that teams often use them in a fragmented manner spanning multiple domains and workflows. Context also matters, according to the firm, and agents typically aren’t fed with relevant data or information on how to support teams. </p><p>“Most agents aren’t onboarded with the context of how their teams operate, prior decisions, or what their priorities are,” Asana noted. </p><p>Finally, concerns over governance and safety are also hampering progress with agents. IT leaders are growing increasingly worried about agents “operating with unchecked data access” and little cost oversight. </p><p>According to Asana, this combination of issues means organizations “need an operating layer on top” of AI tools and associated software - “a place where humans and agents run critical workflows together”. </p><p>Asana CEO Dan Rogers said development of this operating layer builds on the firm’s Enterprise Work Graph, launched in 2021. This gave users “shared memory, multiplayer coordination, and governance”. </p><p>Rogers noted this is “precisely what the agentic era requires”. </p><p>“Asana’s OS is how AI moves from helping individuals work faster to supercharging entire organizations," he added.</p><h2 id="the-agentic-operating-system">The agentic operating system</h2><p>The move by Asana bears similarities to Salesforce’s “agentic OS” announcement in October last year. As <a href="https://www.itpro.com/software/slack-is-now-the-key-to-salesforces-agentic-ai-plans"><u><em>ITPro </em></u><u>reported at the time</u></a>, the CRM giant identified a similar problem when rolling out its Agentforce service. </p><p>In a similar vein, teams were often found working with disparate agents and datasets, which hampers cross-functional collaboration and broader productivity gains. </p><p>Salesforce rebranded Slack as an agentic OS with the aim of providing a single point of reference for business data and agents. That move saw <a href="https://www.itpro.com/customer-relationship-management-crm/34398/what-is-crm-documentation-and-how-can-it-benefit-your">CRM </a>data from Salesforce made directly accessible within <a href="https://www.itpro.com/technology/artificial-intelligence/slack-ai-everything-you-need-to-know-about-the-platforms-new-features">Slack</a>.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/software/asana-wants-every-enterprise-to-have-an-ai-chief-of-staff</link>
                                                                            <description>
                            <![CDATA[ The new Asana Dash tool was built to help guide and support teams through projects ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">khu5D7Ppiwaj38ZoDjPXqZ</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/MmKtntNcYCjT2E2FBboYUo-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Fri, 05 Jun 2026 12:35:23 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:21 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Software]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                    <category><![CDATA[Workplace Technology]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/MmKtntNcYCjT2E2FBboYUo-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Asana logo and branding pictured on a smartphone screen with company logo blurred in background.]]></media:description>                                                            <media:text><![CDATA[Asana logo and branding pictured on a smartphone screen with company logo blurred in background.]]></media:text>
                                <media:title type="plain"><![CDATA[Asana logo and branding pictured on a smartphone screen with company logo blurred in background.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/MmKtntNcYCjT2E2FBboYUo-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p><a href="https://www.itpro.com/technology/artificial-intelligence/how-ai-agent-boss-is-reshaping-it-accountability">Asana</a> has unveiled a series of new features aimed at streamlining collaboration between human workers and agents as it looks to “solve the AI productivity gap” for enterprises. </p><p>Among these is Asana Dash, a new AI “chief of staff” designed to support employees and keep projects on track. </p><p>According to the firm, the tool helps users stay up-to-date with project goals and progress, compiling and summarizing relevant information such as email correspondence, messages between team members, and progress reports. </p><p>The new tool forms a key component within Asana’s new Agentic Work Management platform, unveiled at the company’s Work Innovation Summit in London this week. </p><p>This platform acts as an underlying “operating system” combining core products with the goal of consolidating workflows within a single domain. </p><p>Elsewhere, updates to Asana’s AI Teammates feature complement Dash, according to the firm. New integrations for Gmail, Outlook, Slack, HubSpot, Figma, and Canva were announced at the London summit along with a Skills library for “repeatable work”. </p><p>These two features are intended to work in-tandem, helping support project work, according to Asana. Dash can even assign actions to human workers or AI Teammates.</p><p>“It captures follow-ups from meetings, Slack threads, and email, turns them into structured work in the Work Graph, and connects users to the right AI Teammates for specific tasks and projects to move work forward,” the company said.</p><h2 id="tackling-the-ai-productivity-gap">Tackling the “AI productivity gap”</h2><p>The new features launched by Asana aim to tackle a common recurring pain point for enterprises using AI tools: <a href="https://www.itpro.com/business/business-strategy/ai-productivity-challenges-accenture-generating-impact-study">mismatched productivity gains</a>. </p><p>Research conducted by the firm shows that while 75% of knowledge workers now use AI in daily tasks, only 5% of companies report “meaningful” productivity improvements. </p><p>The company attributes this problem to four key areas. First and foremost, this includes an inability to select the “right agents” for specific individual needs and tasks. </p><p>Notably, the company said “agents aren’t team players,” adding that teams often use them in a fragmented manner spanning multiple domains and workflows. Context also matters, according to the firm, and agents typically aren’t fed with relevant data or information on how to support teams. </p><p>“Most agents aren’t onboarded with the context of how their teams operate, prior decisions, or what their priorities are,” Asana noted. </p><p>Finally, concerns over governance and safety are also hampering progress with agents. IT leaders are growing increasingly worried about agents “operating with unchecked data access” and little cost oversight. </p><p>According to Asana, this combination of issues means organizations “need an operating layer on top” of AI tools and associated software - “a place where humans and agents run critical workflows together”. </p><p>Asana CEO Dan Rogers said development of this operating layer builds on the firm’s Enterprise Work Graph, launched in 2021. This gave users “shared memory, multiplayer coordination, and governance”. </p><p>Rogers noted this is “precisely what the agentic era requires”. </p><p>“Asana’s OS is how AI moves from helping individuals work faster to supercharging entire organizations," he added.</p><h2 id="the-agentic-operating-system">The agentic operating system</h2><p>The move by Asana bears similarities to Salesforce’s “agentic OS” announcement in October last year. As <a href="https://www.itpro.com/software/slack-is-now-the-key-to-salesforces-agentic-ai-plans"><u><em>ITPro </em></u><u>reported at the time</u></a>, the CRM giant identified a similar problem when rolling out its Agentforce service. </p><p>In a similar vein, teams were often found working with disparate agents and datasets, which hampers cross-functional collaboration and broader productivity gains. </p><p>Salesforce rebranded Slack as an agentic OS with the aim of providing a single point of reference for business data and agents. That move saw <a href="https://www.itpro.com/customer-relationship-management-crm/34398/what-is-crm-documentation-and-how-can-it-benefit-your">CRM </a>data from Salesforce made directly accessible within <a href="https://www.itpro.com/technology/artificial-intelligence/slack-ai-everything-you-need-to-know-about-the-platforms-new-features">Slack</a>.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ 'One-size-fits-all' agent governance sets enterprises up to fail ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Governance failures are set to lead to four-in-ten organizations demoting or decommissioning autonomous <a href="https://www.itpro.com/technology/artificial-intelligence/businesses-are-being-taken-for-fools-with-ai-agents">AI agents</a> over the next year.</p><p>In a new report, Gartner warns that a 'one-size-fits-all' approach to governance means that organizations are failing to distinguish between an agent's ability to act and the scope of access it is granted – and that governance gaps are often identified only after production incidents occur.</p><p>"Organizations are treating <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">AI agent governance</a> as binary, either locked down or fully trusted, and that is the root cause of failure," said Shiva Varma, senior director analyst at Gartner. </p><p>"Agents operate at different autonomy levels and across different trust boundaries. When the same controls are applied indiscriminately, organizations encounter two common failure modes: over-restriction of simple agents, which slows delivery and drives shadow development, or under-restriction of more autonomous agents, which increases operational, security, and compliance risk."</p><p>Gartner believes it has the solution, in the form of a proportional governance approach that classifies AI agents across distinct autonomy levels, with each level representing a different trust boundary and corresponding governance requirements.</p><p>The first covers 'observe' agents, limited to read-only access to defined data sources, with outputs visible only to the user making a request – functions such as document summarization, data or knowledge retrieval, and code explanation.</p><p>"At this level, governance should focus on baseline controls such as scoped data access, user authentication, usage logging, and basic functional and security testing," said Varma. "Because risk is limited primarily to data exposure and output accuracy, controls should remain lightweight and targeted."</p><p>The second level consists of 'advise' agents, which generate recommendations, drafts, or proposed actions, while humans review all outputs and execute actions manually. These agents retain read‑only access with no write access to any system, with use cases including email drafting, report or code generation, and decision support.</p><p>Although decisions are only implemented by humans, these advisory agents can often be trusted too much, due to automation bias.</p><p>"Governance for advise agents should include all Level 1 controls and extend to addressing output quality and decision influence through accuracy and hallucination testing, domain-specific quality evaluations, and user training on appropriate reliance levels," said Varma.</p><p>'Act with approval' agents, meanwhile, can carry out actions such as writing data, sending communications, or modifying configurations – but only after explicit human approval for every action.</p><p>"At this level, human review is effective only if it remains a meaningful control," said Varma. "Without strong security testing, clear approval workflows with audit trails, and agent‑specific incident response procedures, approvals can degrade under time pressure or approval fatigue, creating a false sense of safety while expanding the attack surface."</p><p>And finally, at the highest level of autonomy, agents execute actions independently within defined guardrails, with humans reviewing exceptions, audit logs and aggregated outcomes rather than individual decisions.</p><p>"When agents operate autonomously, actions are executed at a scale and speed that can outpace human oversight," said Varma. </p><p>"Because accountability for outcomes remains with the organization, this level requires the most rigorous governance, including continuous monitoring, enforced guardrails, rapid rollback mechanisms, circuit breakers that halt agent operation on threshold violations and clear ownership for agent behaviour."</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/one-size-fits-all-agent-governance-sets-enterprises-up-to-fail</link>
                                                                            <description>
                            <![CDATA[ Gartner recommends a graded approach for agents, depending on their level of autonomy ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">t3H6RSMvjHwu6LnDXAHgqP</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/PbaraXgyBf5cecgbD4TJLU-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 27 May 2026 11:54:42 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:12 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                                    <dc:creator><![CDATA[ Emma Woollacott ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/aWfskavxoVSMDy6cDWtYmJ-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/PbaraXgyBf5cecgbD4TJLU-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[An illustration showing an AI agent side profile, depicted as a blue robot, with seven human faces in varying earthy metallic tones shown to the right.]]></media:description>                                                            <media:text><![CDATA[An illustration showing an AI agent side profile, depicted as a blue robot, with seven human faces in varying earthy metallic tones shown to the right.]]></media:text>
                                <media:title type="plain"><![CDATA[An illustration showing an AI agent side profile, depicted as a blue robot, with seven human faces in varying earthy metallic tones shown to the right.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/PbaraXgyBf5cecgbD4TJLU-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Governance failures are set to lead to four-in-ten organizations demoting or decommissioning autonomous <a href="https://www.itpro.com/technology/artificial-intelligence/businesses-are-being-taken-for-fools-with-ai-agents">AI agents</a> over the next year.</p><p>In a new report, Gartner warns that a 'one-size-fits-all' approach to governance means that organizations are failing to distinguish between an agent's ability to act and the scope of access it is granted – and that governance gaps are often identified only after production incidents occur.</p><p>"Organizations are treating <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">AI agent governance</a> as binary, either locked down or fully trusted, and that is the root cause of failure," said Shiva Varma, senior director analyst at Gartner. </p><p>"Agents operate at different autonomy levels and across different trust boundaries. When the same controls are applied indiscriminately, organizations encounter two common failure modes: over-restriction of simple agents, which slows delivery and drives shadow development, or under-restriction of more autonomous agents, which increases operational, security, and compliance risk."</p><p>Gartner believes it has the solution, in the form of a proportional governance approach that classifies AI agents across distinct autonomy levels, with each level representing a different trust boundary and corresponding governance requirements.</p><p>The first covers 'observe' agents, limited to read-only access to defined data sources, with outputs visible only to the user making a request – functions such as document summarization, data or knowledge retrieval, and code explanation.</p><p>"At this level, governance should focus on baseline controls such as scoped data access, user authentication, usage logging, and basic functional and security testing," said Varma. "Because risk is limited primarily to data exposure and output accuracy, controls should remain lightweight and targeted."</p><p>The second level consists of 'advise' agents, which generate recommendations, drafts, or proposed actions, while humans review all outputs and execute actions manually. These agents retain read‑only access with no write access to any system, with use cases including email drafting, report or code generation, and decision support.</p><p>Although decisions are only implemented by humans, these advisory agents can often be trusted too much, due to automation bias.</p><p>"Governance for advise agents should include all Level 1 controls and extend to addressing output quality and decision influence through accuracy and hallucination testing, domain-specific quality evaluations, and user training on appropriate reliance levels," said Varma.</p><p>'Act with approval' agents, meanwhile, can carry out actions such as writing data, sending communications, or modifying configurations – but only after explicit human approval for every action.</p><p>"At this level, human review is effective only if it remains a meaningful control," said Varma. "Without strong security testing, clear approval workflows with audit trails, and agent‑specific incident response procedures, approvals can degrade under time pressure or approval fatigue, creating a false sense of safety while expanding the attack surface."</p><p>And finally, at the highest level of autonomy, agents execute actions independently within defined guardrails, with humans reviewing exceptions, audit logs and aggregated outcomes rather than individual decisions.</p><p>"When agents operate autonomously, actions are executed at a scale and speed that can outpace human oversight," said Varma. </p><p>"Because accountability for outcomes remains with the organization, this level requires the most rigorous governance, including continuous monitoring, enforced guardrails, rapid rollback mechanisms, circuit breakers that halt agent operation on threshold violations and clear ownership for agent behaviour."</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Dell Technologies World 2026: agents, hardware, and tokenomics ]]></title>
                                                                                                <dc:content><![CDATA[ <iframe allow="" height="200px" width="100%" id="" style="" class="position-center" data-lazy-priority="high" data-lazy-src="https://player.captivate.fm/episode/ed77b66b-997c-4028-9434-2379cc675009/" src="https://player.captivate.fm/episode/ed77b66b-997c-4028-9434-2379cc675009/"></iframe><p>It’s been a busy week for the enterprise tech world in Las Vegas as Dell Technologies customers, partners, and channel partners poured into the Venetian Conference Center to hear about the company’s latest strategies, products, and predictions for the future of IT.</p><p>In this episode, Bobby speaks to Jane about what she’s learnt during her week at the conference, what some of the big announcements were, and whether her pre-conference predictions were correct.</p><h2 id="highlights">Highlights</h2><p>“The move towards agents has changed the conversation around what's important in terms of hardware, so for parallel computing and for LLM training and whatnot, it's all about the GPUs, but as we move to inference and we’ve moved to to AI agents, the CPU starts to become more important again, just because of how compute intensive these things are.”</p><p>“Agents are token hungry, you know. They can use up in days or months what otherwise would take in months to a year. You may have seen the story, or listeners may have seen the story, that Uber has used up its entire AI budget for 2026 in April or early May … and it‘s prompting a bit of a rethink about how this is used now.”</p><p>“It seems like there's a reckoning with the economics of this since the invention of agents and their increasingly widespread use thanks to OpenClaw. There's a big change coming and fortuitously for Dell, from my perspective, and from the perspective of experts who I spoke to for my feature on this, it does probably work out quite well for Dell and other hardware vendors.”</p><h2 id="related-content">Related content</h2><ul><li><a href="https://www.itpro.com/technology/artificial-intelligence/could-rising-token-costs-boost-interest-in-on-premises-hardware"><u>Could rising token costs boost interest in on-premises hardware?</u></a></li><li><a href="https://www.itpro.com/technology/artificial-intelligence/now-we-have-for-the-very-first-time-useful-ai-jensen-huang-and-michael-dell-talk-up-the-power-of-agentic-ai-at-dell-technologies-world-2026"><u>“Now we have, for the very first time, useful AI” – Jensen Huang and Michael Dell talk up the power of agentic AI at Dell Technologies World 2026</u></a></li><li><a href="https://www.itpro.com/technology/artificial-intelligence/dell-unveils-deskside-agentic-ai-at-dell-technologies-world-2026"><u>Dell unveils Deskside Agentic AI at Dell Technologies World 2026</u></a></li><li><a href="https://www.itpro.com/business/business-strategy/three-things-i-expect-to-see-at-dell-technologies-world-2026"><u>Three things I expect to see at Dell Technologies World 2026, and one I don’t</u></a></li><li><a href="https://www.itpro.com/technology/artificial-intelligence/google-cloud-next-2026-scaling-ai-agents"><u>Google Cloud Next 2026: Scaling AI agents</u></a></li><li><a href="https://www.itpro.com/technology/artificial-intelligence/the-agentic-identity-crisis"><u>The agentic identity crisis</u></a></li><li><a href="https://www.itpro.com/hardware/ces-2026-the-year-of-ai-pcs"><u>CES 2026: The year of AI PCs?</u></a></li></ul> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/dell-technologies-world-2026-agents-hardware-and-tokenomics</link>
                                                                            <description>
                            <![CDATA[ Jane, live from Las Vegas, takes us through her week at Dell’s AI agent extravaganza ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">KKN7h32bZMAuS873yeTtzP</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/xsBqiMbLZH3EcPm2SyVdgE-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Fri, 22 May 2026 06:42:33 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:12 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ jane.mccallion@futurenet.com (Jane McCallion) ]]></author>                    <dc:creator><![CDATA[ Jane McCallion ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Wq9nnLr7TNkY8gyBRb7YsA-320-70.jpeg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Jane is managing editor at ITPro and ChannelPro. She started out with the brands as a staff writer specializing in cloud computing before going on to become senior writer and reports editor, managing the content and creation of ITPro’s quarterly whitepapers. During this time, she broadened her expertise to include cybersecurity, data centers and enterprise IT infrastructure. In 2016, she became features editor, managing a pool of freelance and internal writers, while continuing to specialize in enterprise IT infrastructure, data centers, and business strategy.&lt;/p&gt;&lt;p&gt;In October 2021, she became the sites’ deputy editor, before moving to the role of managing editor in June 2024. Although she now has a more strategic role,  she is still a specialist in enterprise IT infrastructure, business strategy, and cybersecurity.&lt;/p&gt;&lt;p&gt;Jane holds an MA in journalism from Goldsmiths, University of London, and a BA in Applied Languages from the University of Portsmouth. She is fluent in French and Spanish, and has written features in both languages.&lt;/p&gt;&lt;p&gt;Prior to joining ITPro, Jane was a freelance business journalist writing as both Jane McCallion and Jane Bordenave for titles such as European CEO, World Finance, and Business Excellence Magazine.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/xsBqiMbLZH3EcPm2SyVdgE-1920-80.jpg">
                                                            <media:credit><![CDATA[Future]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[The words &quot;Dell Technologies World 2026: agents, hardware, and tokenomics&quot; in yellow and white over a blurred photo from Michael Dell&#039;s 2026 keynote]]></media:description>                                                            <media:text><![CDATA[The words &quot;Dell Technologies World 2026: agents, hardware, and tokenomics&quot; in yellow and white over a blurred photo from Michael Dell&#039;s 2026 keynote]]></media:text>
                                <media:title type="plain"><![CDATA[The words &quot;Dell Technologies World 2026: agents, hardware, and tokenomics&quot; in yellow and white over a blurred photo from Michael Dell&#039;s 2026 keynote]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/xsBqiMbLZH3EcPm2SyVdgE-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <iframe allow="" height="200px" width="100%" id="" style="" class="position-center" data-lazy-priority="high" data-lazy-src="https://player.captivate.fm/episode/ed77b66b-997c-4028-9434-2379cc675009/" src="https://player.captivate.fm/episode/ed77b66b-997c-4028-9434-2379cc675009/"></iframe><p>It’s been a busy week for the enterprise tech world in Las Vegas as Dell Technologies customers, partners, and channel partners poured into the Venetian Conference Center to hear about the company’s latest strategies, products, and predictions for the future of IT.</p><p>In this episode, Bobby speaks to Jane about what she’s learnt during her week at the conference, what some of the big announcements were, and whether her pre-conference predictions were correct.</p><h2 id="highlights">Highlights</h2><p>“The move towards agents has changed the conversation around what's important in terms of hardware, so for parallel computing and for LLM training and whatnot, it's all about the GPUs, but as we move to inference and we’ve moved to to AI agents, the CPU starts to become more important again, just because of how compute intensive these things are.”</p><p>“Agents are token hungry, you know. They can use up in days or months what otherwise would take in months to a year. You may have seen the story, or listeners may have seen the story, that Uber has used up its entire AI budget for 2026 in April or early May … and it‘s prompting a bit of a rethink about how this is used now.”</p><p>“It seems like there's a reckoning with the economics of this since the invention of agents and their increasingly widespread use thanks to OpenClaw. There's a big change coming and fortuitously for Dell, from my perspective, and from the perspective of experts who I spoke to for my feature on this, it does probably work out quite well for Dell and other hardware vendors.”</p><h2 id="related-content">Related content</h2><ul><li><a href="https://www.itpro.com/technology/artificial-intelligence/could-rising-token-costs-boost-interest-in-on-premises-hardware"><u>Could rising token costs boost interest in on-premises hardware?</u></a></li><li><a href="https://www.itpro.com/technology/artificial-intelligence/now-we-have-for-the-very-first-time-useful-ai-jensen-huang-and-michael-dell-talk-up-the-power-of-agentic-ai-at-dell-technologies-world-2026"><u>“Now we have, for the very first time, useful AI” – Jensen Huang and Michael Dell talk up the power of agentic AI at Dell Technologies World 2026</u></a></li><li><a href="https://www.itpro.com/technology/artificial-intelligence/dell-unveils-deskside-agentic-ai-at-dell-technologies-world-2026"><u>Dell unveils Deskside Agentic AI at Dell Technologies World 2026</u></a></li><li><a href="https://www.itpro.com/business/business-strategy/three-things-i-expect-to-see-at-dell-technologies-world-2026"><u>Three things I expect to see at Dell Technologies World 2026, and one I don’t</u></a></li><li><a href="https://www.itpro.com/technology/artificial-intelligence/google-cloud-next-2026-scaling-ai-agents"><u>Google Cloud Next 2026: Scaling AI agents</u></a></li><li><a href="https://www.itpro.com/technology/artificial-intelligence/the-agentic-identity-crisis"><u>The agentic identity crisis</u></a></li><li><a href="https://www.itpro.com/hardware/ces-2026-the-year-of-ai-pcs"><u>CES 2026: The year of AI PCs?</u></a></li></ul>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Google adds AI to the search box ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Google is adding new AI features to Search as well as introducing an "intelligent AI-powered" Search box, which the company described as its biggest upgrade in the 25 years since its launch. </p><p>Two years ago, Google introduced <a href="https://www.itpro.com/technology/artificial-intelligence/will-ai-eat-the-internet">AI Overviews</a>, embedding an AI-generated text summary into some results, and last year added "AI mode"  to its search bar as an option. </p><p>That came amid rising concerns that the tech giant could lose the dominance of its core tool as more and more people shift to AI chatbots for information – though Google noted that Search continues to see growth, with queries at an "all-time high" last quarter.</p><p>Now, Google is further integrating AI into its core search tool. First, it's upgrading AI Mode to <a href="https://www.itpro.com/technology/artificial-intelligence/google-jumps-on-the-agentic-ai-bandwagon">Gemini 3.5 Flash</a>, the most recent version of that model.</p><p>But the bigger change is with the Search box. "This intelligent Search box puts our most powerful AI tools right at your fingertips, making it easier to ask your questions," said Elizabeth Reid, VP of search at Google, in a <a href="https://www.itpro.com/technology/artificial-intelligence/google-jumps-on-the-agentic-ai-bandwagon">blog post</a>. </p><p>The news might not be welcome for everyone, amid rising pushback against AI for its environmental impacts as well as inaccuracies. Indeed, one <a href="https://arstechnica.com/google/2026/04/analysis-finds-google-ai-overviews-is-wrong-10-percent-of-the-time/">study </a>showed AI Overviews as being correct 91% of the time, but that suggests it's giving inaccurate responses one in ten searches; Google has disputed that report. </p><h2 id="ai-in-google-search">AI in Google search</h2><p>The Search box itself will expand as users type, in order to give more space for lengthy natural language questions, and include search across text, images, files, and even Chrome tabs. "Designed to anticipate your intent, it also helps you formulate your question with AI-powered suggestions that go beyond autocomplete," said Reid. </p><p>She added that Search will continue to return a range of results, as it does today. But it will also be able to search your own data, such as <a href="https://www.itpro.com/business-strategy/collaboration/363965/the-complete-guide-to-gmail">Gmail</a> and Photos, and soon Calendar, with Reid saying that "AI to be most helpful, it shouldn't just know the world's information, it should understand your context, too."</p><p>That will require connecting those apps, which can be managed by the user. </p><p>The AI Search box is starting to roll out this week, and will be available in all countries and languages where AI Mode is already enabled. The Personal Intelligence in AI Mode is being expanded to 200 countries with no subscription. </p><h2 id="coding-in-search">Coding in search</h2><p>Google will also add its agentic coding tools Search, letting users build a response to suit them. </p><p>"So you can get custom generative UI, including visual tools and simulations, tailored precisely to your needs," said Reid. "Whether you want to wrap your mind around astrophysics or visualize how your watch works, Search can design custom layouts, assembling components (like interactive visuals, tables, graphs, or simulations) in real-time."</p><p>Search will also get the ability to create mini apps and dashboards for ongoing tasks, such as managing a move or planning a wedding, Reid said. "Search can go a step further, building you custom dashboards or trackers that you can continue to come back to and make progress on," she explained, adding that the capability will start first for Google AI Pro and Ultra subscribers in the US in the next few months.</p><h2 id="ai-search-agents">AI search agents </h2><p>Beyond the search box, Google is also making it possible to ask a follow-up question in AI Overview, essentially flowing the generated summary into a chat. "Your context stays with you, and as you explore more deeply, the links and supporting articles get even more relevant," said Reid. </p><p>Google is also adding agents to Search, letting users create and manage <a href="https://www.itpro.com/technology/artificial-intelligence/it-leaders-dont-trust-ai-agents-yet-and-theyre-missing-out-on-huge-financial-gains">AI agents</a> from the search box. </p><p>"We're starting with information agents," said Reid. "Operating in the background, 24/7, these agents intelligently reason across information to find exactly what you need at exactly the right moment."</p><p>These information agents will proactively search across the web – be it websites, social posts, news sites, and so on – to spot any new information or changes to a question, sending a summarised update. </p><p>"So if you're apartment hunting, you can brain dump all of the exact requirements you're looking for, and your agent will continuously scan for you, notifying you when listings meet your needs," said Reid. "Or if you want to know the instant any of your favorite pro athletes announce a sneaker collab, your agent will let you know when a new drop lands so you don't miss out." </p><p>Information agents will be available for Google AI and Ultra subscribers from this summer. </p><p>For all US users, Search will get agentic booking for a wider range of tasks, such as finding a local restaurant, with Google able to call a business on your behalf for select categories. </p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/google-adds-ai-to-the-search-box</link>
                                                                            <description>
                            <![CDATA[ Major changes for how Google's search functions with the integration of AI tools ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">twXbG8Ld9qPLPYL479GaJ7</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/5y6TqnhR97AtuozNVbXz3T-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Thu, 21 May 2026 13:08:13 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 13:56:12 +0000</updated>
                                                                                                                                            <category><![CDATA[Google]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                    <category><![CDATA[Workplace Technology]]></category>
                                                    <category><![CDATA[Software]]></category>
                                                                                                                    <dc:creator><![CDATA[ Nicole Kobie ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/8Y8JDDTQ7XDEk49FoAFP2S-320-70.png ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Nicole Kobie first started writing for ITPro in 2007. As a freelance journalist covering technology and business, Nicole&#039;s work includes  bylines in New Scientist, Wired, PC Pro and many more. &lt;/p&gt;&lt;p&gt;Nicole the author of a book about the history of technology, The Long History of the Future.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/5y6TqnhR97AtuozNVbXz3T-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Google&#039;s head of search, Liz Reid, during the I/O Developers Conference 2026]]></media:description>                                                            <media:text><![CDATA[Google&#039;s head of search, Liz Reid, during the I/O Developers Conference 2026]]></media:text>
                                <media:title type="plain"><![CDATA[Google&#039;s head of search, Liz Reid, during the I/O Developers Conference 2026]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/5y6TqnhR97AtuozNVbXz3T-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Google is adding new AI features to Search as well as introducing an "intelligent AI-powered" Search box, which the company described as its biggest upgrade in the 25 years since its launch. </p><p>Two years ago, Google introduced <a href="https://www.itpro.com/technology/artificial-intelligence/will-ai-eat-the-internet">AI Overviews</a>, embedding an AI-generated text summary into some results, and last year added "AI mode"  to its search bar as an option. </p><p>That came amid rising concerns that the tech giant could lose the dominance of its core tool as more and more people shift to AI chatbots for information – though Google noted that Search continues to see growth, with queries at an "all-time high" last quarter.</p><p>Now, Google is further integrating AI into its core search tool. First, it's upgrading AI Mode to <a href="https://www.itpro.com/technology/artificial-intelligence/google-jumps-on-the-agentic-ai-bandwagon">Gemini 3.5 Flash</a>, the most recent version of that model.</p><p>But the bigger change is with the Search box. "This intelligent Search box puts our most powerful AI tools right at your fingertips, making it easier to ask your questions," said Elizabeth Reid, VP of search at Google, in a <a href="https://www.itpro.com/technology/artificial-intelligence/google-jumps-on-the-agentic-ai-bandwagon">blog post</a>. </p><p>The news might not be welcome for everyone, amid rising pushback against AI for its environmental impacts as well as inaccuracies. Indeed, one <a href="https://arstechnica.com/google/2026/04/analysis-finds-google-ai-overviews-is-wrong-10-percent-of-the-time/">study </a>showed AI Overviews as being correct 91% of the time, but that suggests it's giving inaccurate responses one in ten searches; Google has disputed that report. </p><h2 id="ai-in-google-search">AI in Google search</h2><p>The Search box itself will expand as users type, in order to give more space for lengthy natural language questions, and include search across text, images, files, and even Chrome tabs. "Designed to anticipate your intent, it also helps you formulate your question with AI-powered suggestions that go beyond autocomplete," said Reid. </p><p>She added that Search will continue to return a range of results, as it does today. But it will also be able to search your own data, such as <a href="https://www.itpro.com/business-strategy/collaboration/363965/the-complete-guide-to-gmail">Gmail</a> and Photos, and soon Calendar, with Reid saying that "AI to be most helpful, it shouldn't just know the world's information, it should understand your context, too."</p><p>That will require connecting those apps, which can be managed by the user. </p><p>The AI Search box is starting to roll out this week, and will be available in all countries and languages where AI Mode is already enabled. The Personal Intelligence in AI Mode is being expanded to 200 countries with no subscription. </p><h2 id="coding-in-search">Coding in search</h2><p>Google will also add its agentic coding tools Search, letting users build a response to suit them. </p><p>"So you can get custom generative UI, including visual tools and simulations, tailored precisely to your needs," said Reid. "Whether you want to wrap your mind around astrophysics or visualize how your watch works, Search can design custom layouts, assembling components (like interactive visuals, tables, graphs, or simulations) in real-time."</p><p>Search will also get the ability to create mini apps and dashboards for ongoing tasks, such as managing a move or planning a wedding, Reid said. "Search can go a step further, building you custom dashboards or trackers that you can continue to come back to and make progress on," she explained, adding that the capability will start first for Google AI Pro and Ultra subscribers in the US in the next few months.</p><h2 id="ai-search-agents">AI search agents </h2><p>Beyond the search box, Google is also making it possible to ask a follow-up question in AI Overview, essentially flowing the generated summary into a chat. "Your context stays with you, and as you explore more deeply, the links and supporting articles get even more relevant," said Reid. </p><p>Google is also adding agents to Search, letting users create and manage <a href="https://www.itpro.com/technology/artificial-intelligence/it-leaders-dont-trust-ai-agents-yet-and-theyre-missing-out-on-huge-financial-gains">AI agents</a> from the search box. </p><p>"We're starting with information agents," said Reid. "Operating in the background, 24/7, these agents intelligently reason across information to find exactly what you need at exactly the right moment."</p><p>These information agents will proactively search across the web – be it websites, social posts, news sites, and so on – to spot any new information or changes to a question, sending a summarised update. </p><p>"So if you're apartment hunting, you can brain dump all of the exact requirements you're looking for, and your agent will continuously scan for you, notifying you when listings meet your needs," said Reid. "Or if you want to know the instant any of your favorite pro athletes announce a sneaker collab, your agent will let you know when a new drop lands so you don't miss out." </p><p>Information agents will be available for Google AI and Ultra subscribers from this summer. </p><p>For all US users, Search will get agentic booking for a wider range of tasks, such as finding a local restaurant, with Google able to call a business on your behalf for select categories. </p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ NTT Data to acquire WinWire in enterprise AI services push ]]></title>
                                                                                                <dc:content><![CDATA[ <p>NTT Data has announced plans to acquire Microsoft partner WinWire, an <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">agentic AI</a> SaaS business. </p><p>The acquisition will see NTT Data expand its expertise across agentic AI, data engineering, and cloud-native application development, while adding more than 1,000 Azure engineers and specialists to its global Microsoft practice.</p><p>WinWire is a California-based firm, founded in 2007, that specializes in AI-led digital transformation services, with expertise across Microsoft environments such as Fabric and Azure AI Foundry. The firm is a six-time winner of the Microsoft Partner of the Year award, as well as being a member of the Microsoft Agentic Partner Alliance Program.</p><p>In its announcement, NTT Data said the acquisition further enhances its ability to co-innovate and co-sell industry-focused AI solutions alongside the vendor.</p><p>"The acquisition of WinWire is a decisive step in advancing our enterprise AI strategy and expanding our leadership in Microsoft Azure and AI-powered cloud transformation," commented Abhijit Dubey, CEO and chief AI officer at NTT DATA. </p><p>"By combining WinWire's deep expertise in cloud-native development and agentic AI with NTT DATA's global scale, this positions us to lead the shift to enterprise AI, enabling clients to move from experimentation to enterprise-wide deployment and achieve meaningful business outcomes."</p><h2 id="expanding-ai-solutions-delivery">Expanding AI solutions delivery</h2><p>NTT Data said the acquisition comes amid growing enterprise demand for <a href="https://www.itpro.com/technology/artificial-intelligence/how-the-uk-leading-europe-ai-driven-manufacturing">AI-driven</a> transformation and cloud-native infrastructure, with organizations increasingly seeking platforms capable of supporting AI deployments across their wider business.</p><p>Upon closing of the deal, the combined capabilities are expected to strengthen NTT Data's delivery of industry-specific AI solutions across infrastructure, applications, data, managed services, and security.</p><p>WinWire also brings its 'Agentic AI @ Scale' framework to the business, which enables the design and deployment of intelligent autonomous systems directly into enterprise workflows.</p><p>Commenting on the move, WinWire CEO Ashu Goel said the company was established with the aim of delivering meaningful <a href="https://www.itpro.com/microsoft-azure/34048/microsoft-azure-review-competitive-cloud-pricing-takes-a-bite-out-of-aws">Azure</a> and AI-led transformation to the enterprise.</p><p>"Joining NTT DATA marks an exciting new chapter, allowing us to extend our capabilities to a much broader global client base," he added. "Together, we are well-positioned to accelerate innovation and shape the next wave of AI-driven transformation for our clients."</p><h2 id="strengthening-collaborative-microsoft-services">Strengthening collaborative Microsoft services</h2><p>The WinWire acquisition builds on NTT's long-standing partnership with Microsoft and will complement the firm's Global Business Unit for Microsoft Cloud, which currently operates across more than 50 countries and is backed by over 24,000 Microsoft certifications.</p><p>Stephen Boyle, corporate vice president of enterprise partner solutions at Microsoft, said the deal bolsters the companies' collaboration efforts and ability to deliver effective AI transformation services.</p><p>"As enterprises look to unlock the full value of AI on Microsoft Azure, the role of skilled partners has never been more critical," he explained. "By combining NTT DATA's global scale with WinWire's expertise in cloud-native development and agentic AI, this acquisition enhances our joint ability to co-innovate and deliver transformative solutions."</p><p>Closure of the acquisition remains subject to customary closing conditions. Financial terms of the deal were not disclosed.</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/acquisition/ntt-data-to-acquire-winwire-in-enterprise-ai-services-push</link>
                                                                            <description>
                            <![CDATA[ The deal will expand the firm's Microsoft practice and AI capabilities with an additional 1,000 Azure engineers and specialists ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">ZEx2DSb2rQrQ4ndGmbgHnB</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/Yq8qGjRNkbYBD8zcUcqLWK-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 20 May 2026 08:23:46 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:42:53 +0000</updated>
                                                                                                                                            <category><![CDATA[Acquisition]]></category>
                                                    <category><![CDATA[Business]]></category>
                                                                                                <author><![CDATA[ itpro@futurenet.com (Daniel Todd) ]]></author>                    <dc:creator><![CDATA[ Daniel Todd ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/SRyC34qeLpNDj3dJtsVDhT-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/Yq8qGjRNkbYBD8zcUcqLWK-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Concept image showing two people shaking hands symbolizing a deal.]]></media:description>                                                            <media:text><![CDATA[Concept image showing two people shaking hands symbolizing a deal.]]></media:text>
                                <media:title type="plain"><![CDATA[Concept image showing two people shaking hands symbolizing a deal.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/Yq8qGjRNkbYBD8zcUcqLWK-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>NTT Data has announced plans to acquire Microsoft partner WinWire, an <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">agentic AI</a> SaaS business. </p><p>The acquisition will see NTT Data expand its expertise across agentic AI, data engineering, and cloud-native application development, while adding more than 1,000 Azure engineers and specialists to its global Microsoft practice.</p><p>WinWire is a California-based firm, founded in 2007, that specializes in AI-led digital transformation services, with expertise across Microsoft environments such as Fabric and Azure AI Foundry. The firm is a six-time winner of the Microsoft Partner of the Year award, as well as being a member of the Microsoft Agentic Partner Alliance Program.</p><p>In its announcement, NTT Data said the acquisition further enhances its ability to co-innovate and co-sell industry-focused AI solutions alongside the vendor.</p><p>"The acquisition of WinWire is a decisive step in advancing our enterprise AI strategy and expanding our leadership in Microsoft Azure and AI-powered cloud transformation," commented Abhijit Dubey, CEO and chief AI officer at NTT DATA. </p><p>"By combining WinWire's deep expertise in cloud-native development and agentic AI with NTT DATA's global scale, this positions us to lead the shift to enterprise AI, enabling clients to move from experimentation to enterprise-wide deployment and achieve meaningful business outcomes."</p><h2 id="expanding-ai-solutions-delivery">Expanding AI solutions delivery</h2><p>NTT Data said the acquisition comes amid growing enterprise demand for <a href="https://www.itpro.com/technology/artificial-intelligence/how-the-uk-leading-europe-ai-driven-manufacturing">AI-driven</a> transformation and cloud-native infrastructure, with organizations increasingly seeking platforms capable of supporting AI deployments across their wider business.</p><p>Upon closing of the deal, the combined capabilities are expected to strengthen NTT Data's delivery of industry-specific AI solutions across infrastructure, applications, data, managed services, and security.</p><p>WinWire also brings its 'Agentic AI @ Scale' framework to the business, which enables the design and deployment of intelligent autonomous systems directly into enterprise workflows.</p><p>Commenting on the move, WinWire CEO Ashu Goel said the company was established with the aim of delivering meaningful <a href="https://www.itpro.com/microsoft-azure/34048/microsoft-azure-review-competitive-cloud-pricing-takes-a-bite-out-of-aws">Azure</a> and AI-led transformation to the enterprise.</p><p>"Joining NTT DATA marks an exciting new chapter, allowing us to extend our capabilities to a much broader global client base," he added. "Together, we are well-positioned to accelerate innovation and shape the next wave of AI-driven transformation for our clients."</p><h2 id="strengthening-collaborative-microsoft-services">Strengthening collaborative Microsoft services</h2><p>The WinWire acquisition builds on NTT's long-standing partnership with Microsoft and will complement the firm's Global Business Unit for Microsoft Cloud, which currently operates across more than 50 countries and is backed by over 24,000 Microsoft certifications.</p><p>Stephen Boyle, corporate vice president of enterprise partner solutions at Microsoft, said the deal bolsters the companies' collaboration efforts and ability to deliver effective AI transformation services.</p><p>"As enterprises look to unlock the full value of AI on Microsoft Azure, the role of skilled partners has never been more critical," he explained. "By combining NTT DATA's global scale with WinWire's expertise in cloud-native development and agentic AI, this acquisition enhances our joint ability to co-innovate and deliver transformative solutions."</p><p>Closure of the acquisition remains subject to customary closing conditions. Financial terms of the deal were not disclosed.</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Jensen Huang and Michael Dell talk up the power of agentic AI  ]]></title>
                                                                                                <dc:content><![CDATA[ <p>AI agents were the name of the game on the first day of Dell Technologies World 2026, with multiple speakers indicating that they make a step-change in business interest in AI.</p><p>During CEO Michael Dell’s opening keynote, guest speaker Nvidia CEO Jensen Huang reflected on the level of technology available when the two companies launched Dell AI Factory with Nvidia.</p><p>“Two years ago, when I was here, we had just started with … generative AI as the product. Generative AI can, of course, generate content, but remember it can also generate content to think with – generate thoughts,” said Huang, “which led us to reasoning, which led us to planning, which led us to agentic systems. So now we have, we now have – for the very first time – useful AI.”</p><p>Huang claimed that agentic AI is significantly shortening the timeframes for tasks such as software programming, which can be reduced from a month to a week.</p><p>“Because it’s so useful, the number of people using agents is now all over the place,” he said. “Every enterprise – my company, [Dell Technologies] – everybody’s using agents all over the place to do software development, DevOps, SRE, all of our CI/CD work, QA, testing.</p><p>“The amount of software work that we do in the company now supported by agents is incredible.”</p><p>Steven Dickens, CEO and principal analyst at HyperFRAME Research, told <em>ITPro</em> that he agrees, but with some caveats. </p><p>“Usefulness is in the eye of the beholder,” he said. “I think that the generative AI phase where we got AI to write for us was useful for some professions (content marketing would be a good example) but it largely was limited to tasks that require summarization and writing. The agentic AI value is a lot higher because AI is now not merely writing it is doing.”</p><p>“We are still very early into this phase of adoption. The usefulness of agents is still relatively nascent and is still in the early adopters phase,” he added. “As tools like Amazon Quick, Perplexity Computer and Claude CoWork become more widely adopted then the perception of usefulness will radically change. This will also be the same beyond personal productivity in areas like AIOps for instance.”</p><p>One key challenge for the widespread adoption of agents, however, is how resource-hungry they are. For Dell Technologies, the answer lies in bringing agents out of the public cloud and returning them on premises. </p><p>This potentially solves the issue of high costs associated with high token use – a single Dell Technologies developer <a href="https://www.itpro.com/technology/artificial-intelligence/dell-unveils-deskside-agentic-ai-at-dell-technologies-world-2026"><u>reportedly ran up a $3,400 token bill in 24 hours</u></a> – but companies will still have to reckon with the cost of buying hardware outright. </p><p>Michael Dell believes, nevertheless, that this is the way the wind is blowing.</p><p>“The real agent takeoff hasn’t really even happened in broad scale across millions of companies,” he told delegates at the company’s Partner Conference, which is also happening at Dell Technologies World. </p><p>“I think it’s still pretty early in all this and … the models, they’re great! But they’re the worst they will ever be, and they’re going to keep improving. There doesn’t seem to be an asymptote in the capabilities.”</p><p><em>ITPro's managing editor Jane McCallion is at Dell Technologies World all week. Click for all </em><a href="https://www.itpro.com/tag/dell-technologies-world"><em>our coverage of the event</em></a><em>, including news and analysis.</em></p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/now-we-have-for-the-very-first-time-useful-ai-jensen-huang-and-michael-dell-talk-up-the-power-of-agentic-ai-at-dell-technologies-world-2026</link>
                                                                            <description>
                            <![CDATA[ Agents are the future of AI in enterprise, according to Dell Technologies and Nvidia CEOs ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">WwndjK5cCx3Qsz832vJSKe</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/bJUKvLkV7Mees8MaFFwWFo-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Tue, 19 May 2026 12:53:55 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:39 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ jane.mccallion@futurenet.com (Jane McCallion) ]]></author>                    <dc:creator><![CDATA[ Jane McCallion ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Wq9nnLr7TNkY8gyBRb7YsA-320-70.jpeg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Jane is managing editor at ITPro and ChannelPro. She started out with the brands as a staff writer specializing in cloud computing before going on to become senior writer and reports editor, managing the content and creation of ITPro’s quarterly whitepapers. During this time, she broadened her expertise to include cybersecurity, data centers and enterprise IT infrastructure. In 2016, she became features editor, managing a pool of freelance and internal writers, while continuing to specialize in enterprise IT infrastructure, data centers, and business strategy.&lt;/p&gt;&lt;p&gt;In October 2021, she became the sites’ deputy editor, before moving to the role of managing editor in June 2024. Although she now has a more strategic role,  she is still a specialist in enterprise IT infrastructure, business strategy, and cybersecurity.&lt;/p&gt;&lt;p&gt;Jane holds an MA in journalism from Goldsmiths, University of London, and a BA in Applied Languages from the University of Portsmouth. She is fluent in French and Spanish, and has written features in both languages.&lt;/p&gt;&lt;p&gt;Prior to joining ITPro, Jane was a freelance business journalist writing as both Jane McCallion and Jane Bordenave for titles such as European CEO, World Finance, and Business Excellence Magazine.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/bJUKvLkV7Mees8MaFFwWFo-1920-80.jpg">
                                                            <media:credit><![CDATA[Jane McCallion/Future]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Jensen Huang and Michael Dell on stage at Dell Technologies World 2026]]></media:description>                                                            <media:text><![CDATA[Jensen Huang and Michael Dell on stage at Dell Technologies World 2026]]></media:text>
                                <media:title type="plain"><![CDATA[Jensen Huang and Michael Dell on stage at Dell Technologies World 2026]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/bJUKvLkV7Mees8MaFFwWFo-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>AI agents were the name of the game on the first day of Dell Technologies World 2026, with multiple speakers indicating that they make a step-change in business interest in AI.</p><p>During CEO Michael Dell’s opening keynote, guest speaker Nvidia CEO Jensen Huang reflected on the level of technology available when the two companies launched Dell AI Factory with Nvidia.</p><p>“Two years ago, when I was here, we had just started with … generative AI as the product. Generative AI can, of course, generate content, but remember it can also generate content to think with – generate thoughts,” said Huang, “which led us to reasoning, which led us to planning, which led us to agentic systems. So now we have, we now have – for the very first time – useful AI.”</p><p>Huang claimed that agentic AI is significantly shortening the timeframes for tasks such as software programming, which can be reduced from a month to a week.</p><p>“Because it’s so useful, the number of people using agents is now all over the place,” he said. “Every enterprise – my company, [Dell Technologies] – everybody’s using agents all over the place to do software development, DevOps, SRE, all of our CI/CD work, QA, testing.</p><p>“The amount of software work that we do in the company now supported by agents is incredible.”</p><p>Steven Dickens, CEO and principal analyst at HyperFRAME Research, told <em>ITPro</em> that he agrees, but with some caveats. </p><p>“Usefulness is in the eye of the beholder,” he said. “I think that the generative AI phase where we got AI to write for us was useful for some professions (content marketing would be a good example) but it largely was limited to tasks that require summarization and writing. The agentic AI value is a lot higher because AI is now not merely writing it is doing.”</p><p>“We are still very early into this phase of adoption. The usefulness of agents is still relatively nascent and is still in the early adopters phase,” he added. “As tools like Amazon Quick, Perplexity Computer and Claude CoWork become more widely adopted then the perception of usefulness will radically change. This will also be the same beyond personal productivity in areas like AIOps for instance.”</p><p>One key challenge for the widespread adoption of agents, however, is how resource-hungry they are. For Dell Technologies, the answer lies in bringing agents out of the public cloud and returning them on premises. </p><p>This potentially solves the issue of high costs associated with high token use – a single Dell Technologies developer <a href="https://www.itpro.com/technology/artificial-intelligence/dell-unveils-deskside-agentic-ai-at-dell-technologies-world-2026"><u>reportedly ran up a $3,400 token bill in 24 hours</u></a> – but companies will still have to reckon with the cost of buying hardware outright. </p><p>Michael Dell believes, nevertheless, that this is the way the wind is blowing.</p><p>“The real agent takeoff hasn’t really even happened in broad scale across millions of companies,” he told delegates at the company’s Partner Conference, which is also happening at Dell Technologies World. </p><p>“I think it’s still pretty early in all this and … the models, they’re great! But they’re the worst they will ever be, and they’re going to keep improving. There doesn’t seem to be an asymptote in the capabilities.”</p><p><em>ITPro's managing editor Jane McCallion is at Dell Technologies World all week. Click for all </em><a href="https://www.itpro.com/tag/dell-technologies-world"><em>our coverage of the event</em></a><em>, including news and analysis.</em></p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Dell unveils Deskside Agentic AI at Dell Technologies World 2026 ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Dell Technologies has taken the wraps off Dell Deskside Agentic AI, part of a slew of updates to Dell AI Factory with Nvidia announced on the first day of <a href="https://www.itpro.com/business/business-strategy/three-things-i-expect-to-see-at-dell-technologies-world-2026">Dell Technologies World 2026</a>.</p><p>Dell Deskside Agentic AI is described as a secure sandbox for building, testing, running, and fine-tuning AI agents locally on a Dell high-performance workstation such as a GB10, GB 300, or Pro Precision 9 Towers.</p><p>It features Nvidia NemoClaw, an open-source software stack consisting of OpenClaw, Nvidia Agent Toolkit, Nvidia OpenShell, and Nvidia Nemotron-3, which allows developers to quickly create and run always-on agents. </p><p>According to Dell Technologies, the benefit of Deskside Agentic AI, versus using public cloud, is twofold: security and cost.</p><p>"Ever since the massive open claw moment that certainly captured Mindshare earlier this year ... agentic AI has been front and center with our customers," Jon Seigal, SVP of client solutions group and online marketing at Dell Technologies, told reporters. "And let's face it, while enterprise IT leaders are excited about the potential game-changing benefits of Agentic AI, they also have some trepidation – some concerns.</p><p>"First is how to control costs. Many have learned this one the hard way, whether it's hitting token limits or the surprise cloud bills. Even though cost per token itself is coming down, super users are burning through tokens at such a high rate that they have sticker shock from the cloud bills."</p><p>As an example, Seigal claimed that one developer at Dell ran up a bill of $3,400 by burning through one billion tokens in 24 hours in a public AI environment.</p><p>According to Dell Technologies, Dell Deskside Agentic AI can expect up to 87% reduced spend compared to public cloud APIs over a two-year period. Break even can be achieved in three months, meanwhile, and </p><p>On security, Seigal said IT leaders are "increasingly wary that sensitive company data could be exposed or misused in some way due to security risks, data privacy concerns, and just the lack of guardrails that are needed within OpenClaw". Deskside Agentic AI, by contrast, runs on-premises, and guardrails can be set in NemoClaw.</p><h2 id="even-more-features-for-ai-data-platform">Even more features for AI Data Platform</h2><p>Despite a <a href="https://www.itpro.com/infrastructure/dell-technologies-unveils-massive-expansion-to-dell-ai-factory-with-nvidia-at-gtc-2026">massive update in March</a>, new features and updates were also announced for Dell AI Data Platform with Nvidia.</p><p>The platform now supports <a href="https://www.itpro.com/business/business-strategy/nvidia-buys-usd2bn-worth-of-synopsys-in-huge-deal">Nvidia Omniverse</a>, allowing customers to feed digital twins and 'physical AI' – AKA robotics – workflows for both training and validation with "trusted, well-organized data", according to Dell Technologies.</p><p><em>ITPro's managing editor Jane McCallion is at Dell Technologies World all week. Click </em><a href="https://www.itpro.com/hardware/live/dell-technologies-world-2026-all-the-news-updates-adn-announcements-from-the-day-1-keynote"><em>here to follow her live blog</em></a><em>, or here for all </em><a href="https://www.itpro.com/tag/dell-technologies-world"><em>our coverage of the event</em></a><em>, including news and analysis.</em></p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/dell-unveils-deskside-agentic-ai-at-dell-technologies-world-2026</link>
                                                                            <description>
                            <![CDATA[ Deskside Agentic AI is the latest in the Dell AI Factory with Nvidia stable, with the company saying it further demonstrates its end-to-end enterprise AI capability ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">spkKotwwnSwuX2fXP8wkEV</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/VTj2sF8HMJbyJT47yQFMCZ-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Mon, 18 May 2026 17:00:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:27 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ jane.mccallion@futurenet.com (Jane McCallion) ]]></author>                    <dc:creator><![CDATA[ Jane McCallion ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Wq9nnLr7TNkY8gyBRb7YsA-320-70.jpeg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Jane is managing editor at ITPro and ChannelPro. She started out with the brands as a staff writer specializing in cloud computing before going on to become senior writer and reports editor, managing the content and creation of ITPro’s quarterly whitepapers. During this time, she broadened her expertise to include cybersecurity, data centers and enterprise IT infrastructure. In 2016, she became features editor, managing a pool of freelance and internal writers, while continuing to specialize in enterprise IT infrastructure, data centers, and business strategy.&lt;/p&gt;&lt;p&gt;In October 2021, she became the sites’ deputy editor, before moving to the role of managing editor in June 2024. Although she now has a more strategic role,  she is still a specialist in enterprise IT infrastructure, business strategy, and cybersecurity.&lt;/p&gt;&lt;p&gt;Jane holds an MA in journalism from Goldsmiths, University of London, and a BA in Applied Languages from the University of Portsmouth. She is fluent in French and Spanish, and has written features in both languages.&lt;/p&gt;&lt;p&gt;Prior to joining ITPro, Jane was a freelance business journalist writing as both Jane McCallion and Jane Bordenave for titles such as European CEO, World Finance, and Business Excellence Magazine.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/VTj2sF8HMJbyJT47yQFMCZ-1920-80.jpg">
                                                            <media:credit><![CDATA[Dell Technologies]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[A graphic of the Dell Technologies Deskside Agentic AI]]></media:description>                                                            <media:text><![CDATA[A graphic of the Dell Technologies Deskside Agentic AI]]></media:text>
                                <media:title type="plain"><![CDATA[A graphic of the Dell Technologies Deskside Agentic AI]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/VTj2sF8HMJbyJT47yQFMCZ-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Dell Technologies has taken the wraps off Dell Deskside Agentic AI, part of a slew of updates to Dell AI Factory with Nvidia announced on the first day of <a href="https://www.itpro.com/business/business-strategy/three-things-i-expect-to-see-at-dell-technologies-world-2026">Dell Technologies World 2026</a>.</p><p>Dell Deskside Agentic AI is described as a secure sandbox for building, testing, running, and fine-tuning AI agents locally on a Dell high-performance workstation such as a GB10, GB 300, or Pro Precision 9 Towers.</p><p>It features Nvidia NemoClaw, an open-source software stack consisting of OpenClaw, Nvidia Agent Toolkit, Nvidia OpenShell, and Nvidia Nemotron-3, which allows developers to quickly create and run always-on agents. </p><p>According to Dell Technologies, the benefit of Deskside Agentic AI, versus using public cloud, is twofold: security and cost.</p><p>"Ever since the massive open claw moment that certainly captured Mindshare earlier this year ... agentic AI has been front and center with our customers," Jon Seigal, SVP of client solutions group and online marketing at Dell Technologies, told reporters. "And let's face it, while enterprise IT leaders are excited about the potential game-changing benefits of Agentic AI, they also have some trepidation – some concerns.</p><p>"First is how to control costs. Many have learned this one the hard way, whether it's hitting token limits or the surprise cloud bills. Even though cost per token itself is coming down, super users are burning through tokens at such a high rate that they have sticker shock from the cloud bills."</p><p>As an example, Seigal claimed that one developer at Dell ran up a bill of $3,400 by burning through one billion tokens in 24 hours in a public AI environment.</p><p>According to Dell Technologies, Dell Deskside Agentic AI can expect up to 87% reduced spend compared to public cloud APIs over a two-year period. Break even can be achieved in three months, meanwhile, and </p><p>On security, Seigal said IT leaders are "increasingly wary that sensitive company data could be exposed or misused in some way due to security risks, data privacy concerns, and just the lack of guardrails that are needed within OpenClaw". Deskside Agentic AI, by contrast, runs on-premises, and guardrails can be set in NemoClaw.</p><h2 id="even-more-features-for-ai-data-platform">Even more features for AI Data Platform</h2><p>Despite a <a href="https://www.itpro.com/infrastructure/dell-technologies-unveils-massive-expansion-to-dell-ai-factory-with-nvidia-at-gtc-2026">massive update in March</a>, new features and updates were also announced for Dell AI Data Platform with Nvidia.</p><p>The platform now supports <a href="https://www.itpro.com/business/business-strategy/nvidia-buys-usd2bn-worth-of-synopsys-in-huge-deal">Nvidia Omniverse</a>, allowing customers to feed digital twins and 'physical AI' – AKA robotics – workflows for both training and validation with "trusted, well-organized data", according to Dell Technologies.</p><p><em>ITPro's managing editor Jane McCallion is at Dell Technologies World all week. Click </em><a href="https://www.itpro.com/hardware/live/dell-technologies-world-2026-all-the-news-updates-adn-announcements-from-the-day-1-keynote"><em>here to follow her live blog</em></a><em>, or here for all </em><a href="https://www.itpro.com/tag/dell-technologies-world"><em>our coverage of the event</em></a><em>, including news and analysis.</em></p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ AI agents aren’t cutting it in customer service  ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Three-quarters of companies have rolled back or shut down an AI agent used in customer service, according to new research. </p><p>That high rate of failure in using <a href="https://www.itpro.com/technology/artificial-intelligence/how-viable-is-ai-customer-care"><u>AI agents for customer communications</u></a> comes via a survey commissioned by Sinch, which polled more than 2,500 leaders in the industry. </p><p>The survey found nearly two-thirds of respondents already had AI agents up and running in their organizations, showing that they were getting beyond pilots into real use cases in customer service. Indeed, 88% are set to have agents in full production over the next year. </p><p>But they're having issues with those deployments, Sinch found. 74% said they had shut down or rolled back AI customer communications agents due to governance failures – that echoes <a href="https://www.itpro.com/technology/artificial-intelligence/digital-first-but-not-digital-only-customer-service-workers-were-first-on-the-ai-chopping-block-but-half-of-enterprises-are-now-backtracking-amid-a-torrent-of-consumer-complaints-and-poor-returns-on-ai"><u>analysis from Gartner</u></a> last year that suggested companies were backtracking on AI customer service and rehiring human workers.</p><p>According to the Sinch report, the leading factors behind roll-backs lay in concerns about customer data exposure (31%), hallucinations or brand risks (22%), and lack of auditability (16%).</p><p>“These aren’t abstract risk categories. [Personal identifiable information] exposure means a customer’s personal data surfaced in an interaction it shouldn’t have. Hallucination means an AI agent said something confidently wrong to a real customer, on a live channel, under your brand’s name,” the <a href="https://sinch.com/ai-production-paradox/chapter/ai-production-challenges/" target="_blank"><u>report </u></a>noted. </p><p>The Sinch survey follows <a href="https://www.itpro.com/technology/artificial-intelligence/your-customers-arent-keen-on-that-customer-service-chatbot-you-introduced-heres-why"><u>research via Pega that showed</u></a> more than half of customers aren't confident that organizations are using AI responsibly in customer service, while a report from 2024 suggested people were searching online for answers to <a href="https://www.itpro.com/technology/artificial-intelligence/so-you-introduced-an-ai-chatbot-for-customers-heres-why-they-probably-hate-it"><u>avoid using an automated chatbot</u></a>. </p><h2 id="governance-works">Governance works</h2><p>The rate of AI u-turns is even higher for organizations that report having fully mature guardrails in place, with 81% saying they've been forced to rollback an AI agent in customer service, perhaps because better monitoring lets them spot failures more quickly or that are otherwise going missed. </p><p>"The industry has assumed that better governance leads to better outcomes," said Daniel Morris, CPO at Sinch. "But that's not enough: If governance was the fix, the most mature teams would roll back less, not more."</p><p>So what's happening? "The most advanced organizations aren't failing less; they're seeing failures sooner," said Morris. "Higher rollback rates reflect better monitoring and control, not weaker performance."</p><p>Despite the many roll backs of AI agents, 98% of those polled said they were increasing AI investment this year, showing a few missteps isn't enough to dissuade customer service operations from the AI bandwagon. </p><h2 id="trust-versus-tech">Trust versus tech</h2><p>Sinch also argued that companies were spending more time on trust, security, and compliance than they were AI itself. Three-quarters (75%) of those polled reported investment in those areas versus 63% for the technology itself. </p><p>Beyond that, 84% of AI engineering teams are spending at least half their time on safety infrastructure. </p><p>"Engineering teams are spending most of their time building and maintaining safety systems, a lot of which their communications infrastructure should be providing, instead of focusing on improving the customer experience," said Morris. "That's the guardrail tax that slows organizations down."</p><p>The report showed chatbots and email responses were the most popular use cases for AI agents in communication channels, both in use at more than six in ten organisations for those polled. </p><p>That was followed by social media, texting, and WhatsApp or similar messaging tools, with half of those polled using AI agents to manage these channels. </p><p>The main motivations are to improve customer satisfaction (36%) and boost revenue and conversions (24%), followed by reducing operational costs (16%).</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/ai-agents-arent-cutting-it-in-customer-service</link>
                                                                            <description>
                            <![CDATA[ Three-quarters of companies have had to pause or halt deployments of AI agents in customer service ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">Wy8EpxoJYQ5QPDSJTkwYki</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/5559aHqCRddzBSS2zYxkM7-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Mon, 18 May 2026 05:00:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 13:19:27 +0000</updated>
                                                                                                                                            <category><![CDATA[Customer Experience Platforms]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                    <category><![CDATA[Enterprise Applications]]></category>
                                                                                                                    <dc:creator><![CDATA[ Nicole Kobie ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/8Y8JDDTQ7XDEk49FoAFP2S-320-70.png ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Nicole Kobie first started writing for ITPro in 2007. As a freelance journalist covering technology and business, Nicole&#039;s work includes  bylines in New Scientist, Wired, PC Pro and many more. &lt;/p&gt;&lt;p&gt;Nicole the author of a book about the history of technology, The Long History of the Future.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/5559aHqCRddzBSS2zYxkM7-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[AI customer service concept image showing robot workers using laptop computers with holographic digital interfaces.]]></media:description>                                                            <media:text><![CDATA[AI customer service concept image showing robot workers using laptop computers with holographic digital interfaces.]]></media:text>
                                <media:title type="plain"><![CDATA[AI customer service concept image showing robot workers using laptop computers with holographic digital interfaces.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/5559aHqCRddzBSS2zYxkM7-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Three-quarters of companies have rolled back or shut down an AI agent used in customer service, according to new research. </p><p>That high rate of failure in using <a href="https://www.itpro.com/technology/artificial-intelligence/how-viable-is-ai-customer-care"><u>AI agents for customer communications</u></a> comes via a survey commissioned by Sinch, which polled more than 2,500 leaders in the industry. </p><p>The survey found nearly two-thirds of respondents already had AI agents up and running in their organizations, showing that they were getting beyond pilots into real use cases in customer service. Indeed, 88% are set to have agents in full production over the next year. </p><p>But they're having issues with those deployments, Sinch found. 74% said they had shut down or rolled back AI customer communications agents due to governance failures – that echoes <a href="https://www.itpro.com/technology/artificial-intelligence/digital-first-but-not-digital-only-customer-service-workers-were-first-on-the-ai-chopping-block-but-half-of-enterprises-are-now-backtracking-amid-a-torrent-of-consumer-complaints-and-poor-returns-on-ai"><u>analysis from Gartner</u></a> last year that suggested companies were backtracking on AI customer service and rehiring human workers.</p><p>According to the Sinch report, the leading factors behind roll-backs lay in concerns about customer data exposure (31%), hallucinations or brand risks (22%), and lack of auditability (16%).</p><p>“These aren’t abstract risk categories. [Personal identifiable information] exposure means a customer’s personal data surfaced in an interaction it shouldn’t have. Hallucination means an AI agent said something confidently wrong to a real customer, on a live channel, under your brand’s name,” the <a href="https://sinch.com/ai-production-paradox/chapter/ai-production-challenges/" target="_blank"><u>report </u></a>noted. </p><p>The Sinch survey follows <a href="https://www.itpro.com/technology/artificial-intelligence/your-customers-arent-keen-on-that-customer-service-chatbot-you-introduced-heres-why"><u>research via Pega that showed</u></a> more than half of customers aren't confident that organizations are using AI responsibly in customer service, while a report from 2024 suggested people were searching online for answers to <a href="https://www.itpro.com/technology/artificial-intelligence/so-you-introduced-an-ai-chatbot-for-customers-heres-why-they-probably-hate-it"><u>avoid using an automated chatbot</u></a>. </p><h2 id="governance-works">Governance works</h2><p>The rate of AI u-turns is even higher for organizations that report having fully mature guardrails in place, with 81% saying they've been forced to rollback an AI agent in customer service, perhaps because better monitoring lets them spot failures more quickly or that are otherwise going missed. </p><p>"The industry has assumed that better governance leads to better outcomes," said Daniel Morris, CPO at Sinch. "But that's not enough: If governance was the fix, the most mature teams would roll back less, not more."</p><p>So what's happening? "The most advanced organizations aren't failing less; they're seeing failures sooner," said Morris. "Higher rollback rates reflect better monitoring and control, not weaker performance."</p><p>Despite the many roll backs of AI agents, 98% of those polled said they were increasing AI investment this year, showing a few missteps isn't enough to dissuade customer service operations from the AI bandwagon. </p><h2 id="trust-versus-tech">Trust versus tech</h2><p>Sinch also argued that companies were spending more time on trust, security, and compliance than they were AI itself. Three-quarters (75%) of those polled reported investment in those areas versus 63% for the technology itself. </p><p>Beyond that, 84% of AI engineering teams are spending at least half their time on safety infrastructure. </p><p>"Engineering teams are spending most of their time building and maintaining safety systems, a lot of which their communications infrastructure should be providing, instead of focusing on improving the customer experience," said Morris. "That's the guardrail tax that slows organizations down."</p><p>The report showed chatbots and email responses were the most popular use cases for AI agents in communication channels, both in use at more than six in ten organisations for those polled. </p><p>That was followed by social media, texting, and WhatsApp or similar messaging tools, with half of those polled using AI agents to manage these channels. </p><p>The main motivations are to improve customer satisfaction (36%) and boost revenue and conversions (24%), followed by reducing operational costs (16%).</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Microsoft researchers say you probably shouldn’t trust AI with work documents ]]></title>
                                                                                                <dc:content><![CDATA[ <p>AI providers keep telling us to hand off work to agents, but research from Microsoft suggests that might not be a wise move. </p><p>In a <a href="https://arxiv.org/pdf/2604.15597" target="_blank"><u>pre-print paper</u></a>, a trio of Microsoft researchers found that <a href="https://www.itpro.com/technology/artificial-intelligence/generative-ai-vs-large-language-models">large language models (LLMs)</a> corrupt documents over the course of long, extensive workflows, resulting in data deletion and even <a href="https://www.itpro.com/technology/artificial-intelligence/ai-hallucinations-what-are-they"><u>hallucinations</u></a>. </p><p>Top tier foundational models – including Gemini 3.1 Pro, <a href="https://www.itpro.com/technology/artificial-intelligence/anthropic-reveals-claude-opus-4-6-enterprise-focused-model-1-million-token-context-window">Claude Opus 4.6</a>, GPT 5.4 – corrupted an average of 25% of content in a document during the research, with other models more than half. </p><p>“Delegation requires trust – the expectation that the LLM will faithfully execute the task without introducing errors into documents," said researchers Philippe Laban, Tobias Schnabel, and Jennifer Neville.  </p><p>"Our analysis shows that current LLMs are unreliable delegates: they introduce sparse but severe errors that silently corrupt documents, compounding over long interaction." </p><p>The work comes amid growing challenges with extensive AI use, with <a href="https://www.itpro.com/technology/artificial-intelligence/workers-are-wasting-half-a-day-each-week-fixing-ai-workslop"><u>one study showing workers</u></a> were wasting half a day each week fixing AI-made "workslop." </p><p>Even in coding, <a href="https://www.itpro.com/software/development/ai-is-creating-more-software-flaws-and-theyre-getting-worse"><u>AI is fast but creates flaws in software</u></a>, according to a study by CodeRabbit. The fact that <a href="https://www.itpro.com/software/development/software-developers-not-checking-ai-generated-code-verification-debt"><u>half of software developers</u></a> aren't checking AI-generated code is also compounding the issue, separate research shows. </p><h2 id="testing-ai-delegation">Testing AI delegation</h2><p>To test how well LLMs manage documents, the trio of researchers built a tool called DELEGATE-25 to simulate long workflows featuring in-depth document editing across 52 professional domains, including coding. They used that tool across 19 LLMs, including the aforementioned frontier models. </p><p>DELEGATE-25 tracked the changes made in text to documents as they were manipulated by the LLMs over five to ten complex editing tasks in each domain.</p><p>The team found that a quarter of document content was lost across frontier models, with that rising to half across the average of all models tested. </p><p>Degradation levels depended on the domain, with LLMs doing better in programming than <a href="https://www.itpro.com/machine-learning/33308/what-is-natural-language-processing">natural language</a> or niche settings, such as earning statements or music notation. </p><p>A model was deemed "ready" to be delegated tasks in a specific domain if it could score 98% accuracy after 20 interactions. Only one domain saw that score across most models: <a href="https://www.itpro.com/business-strategy/careers-training/356640/how-to-become-a-python-software-developer">Python</a>. Gemini 3.1 Pro had the best performance, scoring at least 98% in 11 of 52 domains. </p><p>Turning to agents doesn't help, either, with additional experiments showing agents don't improve performance. The degradation was the worst for larger documents over longer interaction periods.</p><p>Researchers said their work showed "that models are not ready for delegated workflows in the vast majority of domains, with models severely corrupting documents (at least -20% degradation) in 80% of our simulated conditions."</p><p>The results weren't down to consistently adding small errors with every step, the researchers noted. Instead, some of the time models worked perfectly, while other times they lost massive amounts of data in one go. </p><p>"Stronger models do not avoid small errors better; they delay critical failures and experience them in fewer Interactions," they said. </p><h2 id="unchecked-errors">Unchecked errors</h2><p>Because all of the models tested saw degradation in longer workflows – except when working in Python – researchers argued that longer interactions are necessary to benchmark LLM performance. </p><p>They did note improvement in systems by testing <a href="https://www.itpro.com/technology/artificial-intelligence/everything-you-need-to-know-about-openais-gpt-4o-updates-including-pricing-changes-and-new-features">GPT-4o</a> and GPT-5.4, with a leap in performance from 14.7% to 71.5%. </p><p>However, the study added that improvements to LLMs had led to workers supervising AI systems while they complete tasks, such as <a href="https://www.itpro.com/technology/artificial-intelligence/vibe-coding-security-risks-how-to-mitigate">vibe coding</a>. </p><p>"Crucially, users delegating work might lack the expertise or time to review changes implemented by the LLM, and must trust that the LLM does not introduce unchecked errors, such as hallucinations, or deletions," they noted. </p><p>For those working with AI, they advised caution, saying we shouldn't assume an AI that works well in one domain can manage the same in another. </p><p>"Current LLMs are ready for delegated workflows in some domains such as Python coding, but not in other less common domains," the paper noted. </p><p>"In general, users still need to closely monitor LLM systems as they operate and complete tasks on their behalf."</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/llms-are-unreliable-delegates-microsoft-researchers-say-you-probably-shouldnt-trust-ai-with-work-documents</link>
                                                                            <description>
                            <![CDATA[ A research paper from Microsoft shows AI degrades documents over longer workflows ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">LWCJtR66q2ZDwEMmAsWSSc</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/Ld9azfFX858f2Joz9ZkeG-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Tue, 12 May 2026 12:00:12 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:13 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                                    <dc:creator><![CDATA[ Nicole Kobie ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/8Y8JDDTQ7XDEk49FoAFP2S-320-70.png ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Nicole Kobie first started writing for ITPro in 2007. As a freelance journalist covering technology and business, Nicole&#039;s work includes  bylines in New Scientist, Wired, PC Pro and many more. &lt;/p&gt;&lt;p&gt;Nicole the author of a book about the history of technology, The Long History of the Future.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/Ld9azfFX858f2Joz9ZkeG-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[AI slop concept image showing a robot with glowing yellow eyes vomiting green liquid with term &#039;AI slop&#039; in the flowing liquid.]]></media:description>                                                            <media:text><![CDATA[AI slop concept image showing a robot with glowing yellow eyes vomiting green liquid with term &#039;AI slop&#039; in the flowing liquid.]]></media:text>
                                <media:title type="plain"><![CDATA[AI slop concept image showing a robot with glowing yellow eyes vomiting green liquid with term &#039;AI slop&#039; in the flowing liquid.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/Ld9azfFX858f2Joz9ZkeG-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>AI providers keep telling us to hand off work to agents, but research from Microsoft suggests that might not be a wise move. </p><p>In a <a href="https://arxiv.org/pdf/2604.15597" target="_blank"><u>pre-print paper</u></a>, a trio of Microsoft researchers found that <a href="https://www.itpro.com/technology/artificial-intelligence/generative-ai-vs-large-language-models">large language models (LLMs)</a> corrupt documents over the course of long, extensive workflows, resulting in data deletion and even <a href="https://www.itpro.com/technology/artificial-intelligence/ai-hallucinations-what-are-they"><u>hallucinations</u></a>. </p><p>Top tier foundational models – including Gemini 3.1 Pro, <a href="https://www.itpro.com/technology/artificial-intelligence/anthropic-reveals-claude-opus-4-6-enterprise-focused-model-1-million-token-context-window">Claude Opus 4.6</a>, GPT 5.4 – corrupted an average of 25% of content in a document during the research, with other models more than half. </p><p>“Delegation requires trust – the expectation that the LLM will faithfully execute the task without introducing errors into documents," said researchers Philippe Laban, Tobias Schnabel, and Jennifer Neville.  </p><p>"Our analysis shows that current LLMs are unreliable delegates: they introduce sparse but severe errors that silently corrupt documents, compounding over long interaction." </p><p>The work comes amid growing challenges with extensive AI use, with <a href="https://www.itpro.com/technology/artificial-intelligence/workers-are-wasting-half-a-day-each-week-fixing-ai-workslop"><u>one study showing workers</u></a> were wasting half a day each week fixing AI-made "workslop." </p><p>Even in coding, <a href="https://www.itpro.com/software/development/ai-is-creating-more-software-flaws-and-theyre-getting-worse"><u>AI is fast but creates flaws in software</u></a>, according to a study by CodeRabbit. The fact that <a href="https://www.itpro.com/software/development/software-developers-not-checking-ai-generated-code-verification-debt"><u>half of software developers</u></a> aren't checking AI-generated code is also compounding the issue, separate research shows. </p><h2 id="testing-ai-delegation">Testing AI delegation</h2><p>To test how well LLMs manage documents, the trio of researchers built a tool called DELEGATE-25 to simulate long workflows featuring in-depth document editing across 52 professional domains, including coding. They used that tool across 19 LLMs, including the aforementioned frontier models. </p><p>DELEGATE-25 tracked the changes made in text to documents as they were manipulated by the LLMs over five to ten complex editing tasks in each domain.</p><p>The team found that a quarter of document content was lost across frontier models, with that rising to half across the average of all models tested. </p><p>Degradation levels depended on the domain, with LLMs doing better in programming than <a href="https://www.itpro.com/machine-learning/33308/what-is-natural-language-processing">natural language</a> or niche settings, such as earning statements or music notation. </p><p>A model was deemed "ready" to be delegated tasks in a specific domain if it could score 98% accuracy after 20 interactions. Only one domain saw that score across most models: <a href="https://www.itpro.com/business-strategy/careers-training/356640/how-to-become-a-python-software-developer">Python</a>. Gemini 3.1 Pro had the best performance, scoring at least 98% in 11 of 52 domains. </p><p>Turning to agents doesn't help, either, with additional experiments showing agents don't improve performance. The degradation was the worst for larger documents over longer interaction periods.</p><p>Researchers said their work showed "that models are not ready for delegated workflows in the vast majority of domains, with models severely corrupting documents (at least -20% degradation) in 80% of our simulated conditions."</p><p>The results weren't down to consistently adding small errors with every step, the researchers noted. Instead, some of the time models worked perfectly, while other times they lost massive amounts of data in one go. </p><p>"Stronger models do not avoid small errors better; they delay critical failures and experience them in fewer Interactions," they said. </p><h2 id="unchecked-errors">Unchecked errors</h2><p>Because all of the models tested saw degradation in longer workflows – except when working in Python – researchers argued that longer interactions are necessary to benchmark LLM performance. </p><p>They did note improvement in systems by testing <a href="https://www.itpro.com/technology/artificial-intelligence/everything-you-need-to-know-about-openais-gpt-4o-updates-including-pricing-changes-and-new-features">GPT-4o</a> and GPT-5.4, with a leap in performance from 14.7% to 71.5%. </p><p>However, the study added that improvements to LLMs had led to workers supervising AI systems while they complete tasks, such as <a href="https://www.itpro.com/technology/artificial-intelligence/vibe-coding-security-risks-how-to-mitigate">vibe coding</a>. </p><p>"Crucially, users delegating work might lack the expertise or time to review changes implemented by the LLM, and must trust that the LLM does not introduce unchecked errors, such as hallucinations, or deletions," they noted. </p><p>For those working with AI, they advised caution, saying we shouldn't assume an AI that works well in one domain can manage the same in another. </p><p>"Current LLMs are ready for delegated workflows in some domains such as Python coding, but not in other less common domains," the paper noted. </p><p>"In general, users still need to closely monitor LLM systems as they operate and complete tasks on their behalf."</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ ServiceNow wants agents working in every corner of your business ]]></title>
                                                                                                <dc:content><![CDATA[ <p><a href="https://www.itpro.com/business/acquisition/servicenow-wraps-up-usd7-75-billion-armis-acquisition">ServiceNow</a> is targeting mass integration of agents across an array of business functions with a major expansion of its Autonomous Workforce service. </p><p>At the company’s annual Knowledge conference in Las Vegas, ServiceNow unveiled a series of new “AI specialists” aimed at automating tasks spanning a range of domains, from IT operations to <a href="https://www.itpro.com/desktop-software/28214/what-is-crm">customer relationship management (CRM)</a> and security. </p><p>The move by ServiceNow comes amid what it describes as a transformative shift in the evolution of generative AI. </p><p>Enterprises are now moving beyond simple AI assistants to more intuitive agents capable of handling tasks on behalf of the worker, rather than acting as a mere complementary tool.</p><p>"Advisory AI has run its course; enterprises need AI that senses, decides, and securely acts in accordance with organizational guardrails,” said Amit Zavery, ServiceNow’s president, CPO, and COO. </p><p>“With ServiceNow expanding Autonomous Workforce across critical business functions in the enterprise, organizations can deploy AI specialists to act at scale, from a single, governed platform, with full audit trails, role-scoped permissions, and enterprise context built over decades of enterprise operations.”</p><h2 id="building-an-ai-workforce">Building an “AI workforce”</h2><p>A key focus of the Autonomous Workforce expansion hinges on automation in IT operations, with the company noting that teams now face huge backlogs of manual work. </p><p>Initial testing of its previously-announced L1 IT Service Desk AI Specialist found it is delivering marked benefits for teams, resolving cases 99% faster than human workers. </p><p>Capitalizing on those findings, the company unveiled a raft of new AI specialists with applications in infrastructure monitoring, asset <a href="https://www.itpro.com/business/digital-transformation/enterprises-just-cant-seem-to-shake-legacy-tech-and-its-seriously-hampering-digital-transformation-goals">lifecycle management</a>, and site reliability engineering (SRE). </p><p>Honeywell is among those already using ServiceNow’s AI specialists in IT operations, helping to streamline tasks for teams. </p><p>“Our AI assistant ‘Red’ eliminated the majority of service desk conversations, saving time for employees and our IT organization,” said Sheila Jordan, SVP and chief digital technology officer at Honeywell.</p><p>“By operating an <a href="https://www.itpro.com/business/careers-and-training/tech-giants-know-ai-job-losses-are-coming-heres-how-they-plan-to-retrain-the-global-workforce">AI-enabled workforce</a> at enterprise scale, we’re unlocking new levels of efficiency and accelerating business transformation.”</p><p>New AI specialists across employee services, including HR, finance, and legal departments were also touted by ServiceNow at Knowledge 2026 this week. </p><p>The company noted that around 23 million employees use its employee portal each month, generating more than 40 million cases annually – most of which is “routine, repeatable work” that is ripe for automation. </p><p>Again, initial testing in these areas revealed AI specialists were capable of resolving 91% of cases without reassignment and human intervention.</p><p>“Requests are deflected before they become cases, and when cases are created, they route directly to the right AI specialist,” the company explained in a blog post. “</p><p>This allows business operations teams to focus on tasks like workforce planning, supplier strategy decisions, financial planning, contract negotiations, and more.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/advisory-ai-has-run-its-course-servicenow-wants-agents-working-in-every-corner-of-your-business</link>
                                                                            <description>
                            <![CDATA[ A big update to ServiceNow’s Autonomous Workforce service looks to ramp up automation ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">QsPNiiZMoFanS5jY77YuuQ</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/JCDR2uavvqvKda5tRhkYGk-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Thu, 07 May 2026 11:07:51 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:30 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/JCDR2uavvqvKda5tRhkYGk-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Amit Zavery, President, Chief Product Officer, and Chief Operating Officer at ServiceNow speaking onstage during the 2025 Concordia Annual Summit at Sheraton New York Times Square.]]></media:description>                                                            <media:text><![CDATA[Amit Zavery, President, Chief Product Officer, and Chief Operating Officer at ServiceNow speaking onstage during the 2025 Concordia Annual Summit at Sheraton New York Times Square.]]></media:text>
                                <media:title type="plain"><![CDATA[Amit Zavery, President, Chief Product Officer, and Chief Operating Officer at ServiceNow speaking onstage during the 2025 Concordia Annual Summit at Sheraton New York Times Square.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/JCDR2uavvqvKda5tRhkYGk-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p><a href="https://www.itpro.com/business/acquisition/servicenow-wraps-up-usd7-75-billion-armis-acquisition">ServiceNow</a> is targeting mass integration of agents across an array of business functions with a major expansion of its Autonomous Workforce service. </p><p>At the company’s annual Knowledge conference in Las Vegas, ServiceNow unveiled a series of new “AI specialists” aimed at automating tasks spanning a range of domains, from IT operations to <a href="https://www.itpro.com/desktop-software/28214/what-is-crm">customer relationship management (CRM)</a> and security. </p><p>The move by ServiceNow comes amid what it describes as a transformative shift in the evolution of generative AI. </p><p>Enterprises are now moving beyond simple AI assistants to more intuitive agents capable of handling tasks on behalf of the worker, rather than acting as a mere complementary tool.</p><p>"Advisory AI has run its course; enterprises need AI that senses, decides, and securely acts in accordance with organizational guardrails,” said Amit Zavery, ServiceNow’s president, CPO, and COO. </p><p>“With ServiceNow expanding Autonomous Workforce across critical business functions in the enterprise, organizations can deploy AI specialists to act at scale, from a single, governed platform, with full audit trails, role-scoped permissions, and enterprise context built over decades of enterprise operations.”</p><h2 id="building-an-ai-workforce">Building an “AI workforce”</h2><p>A key focus of the Autonomous Workforce expansion hinges on automation in IT operations, with the company noting that teams now face huge backlogs of manual work. </p><p>Initial testing of its previously-announced L1 IT Service Desk AI Specialist found it is delivering marked benefits for teams, resolving cases 99% faster than human workers. </p><p>Capitalizing on those findings, the company unveiled a raft of new AI specialists with applications in infrastructure monitoring, asset <a href="https://www.itpro.com/business/digital-transformation/enterprises-just-cant-seem-to-shake-legacy-tech-and-its-seriously-hampering-digital-transformation-goals">lifecycle management</a>, and site reliability engineering (SRE). </p><p>Honeywell is among those already using ServiceNow’s AI specialists in IT operations, helping to streamline tasks for teams. </p><p>“Our AI assistant ‘Red’ eliminated the majority of service desk conversations, saving time for employees and our IT organization,” said Sheila Jordan, SVP and chief digital technology officer at Honeywell.</p><p>“By operating an <a href="https://www.itpro.com/business/careers-and-training/tech-giants-know-ai-job-losses-are-coming-heres-how-they-plan-to-retrain-the-global-workforce">AI-enabled workforce</a> at enterprise scale, we’re unlocking new levels of efficiency and accelerating business transformation.”</p><p>New AI specialists across employee services, including HR, finance, and legal departments were also touted by ServiceNow at Knowledge 2026 this week. </p><p>The company noted that around 23 million employees use its employee portal each month, generating more than 40 million cases annually – most of which is “routine, repeatable work” that is ripe for automation. </p><p>Again, initial testing in these areas revealed AI specialists were capable of resolving 91% of cases without reassignment and human intervention.</p><p>“Requests are deflected before they become cases, and when cases are created, they route directly to the right AI specialist,” the company explained in a blog post. “</p><p>This allows business operations teams to focus on tasks like workforce planning, supplier strategy decisions, financial planning, contract negotiations, and more.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Google is building its own OpenClaw alternative ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Google is reportedly building its own OpenClaw alternative as the tech giant continues its rapid-fire agentic AI push. </p><p>According to reports from <a href="https://www.businessinsider.com/google-ai-agent-openclaw-remy-gemini-assistant-2026-5" target="_blank"><u><em>Business Insider</em></u></a>, the OpenClaw-style agent, dubbed “Remy”, is being tested internally at the firm through a staff-only version of the Gemini app. </p><p>Sources told the publication the agent will also be integrated with a host of Google services. </p><p>"Remy is your 24/7 personal agent for work, school, and daily life, powered by Gemini," internal documents state.</p><p>"It elevates the Gemini app into a true assistant that can take actions on your behalf — not just answer questions or generate content."</p><p>Development of the agent comes amid sharp interest in OpenClaw since the beginning of the year. The open source agent, which is designed to run locally on devices, allows users to create their own custom agents which can carry out tasks and take actions on their behalf. </p><p>Google has sharpened its focus on agents and orchestration in recent months. As <a href="https://www.itpro.com/technology/artificial-intelligence/google-expands-gemini-enterprise-consolidates-vertex-ai-services-to-simplify-agent-deployment"><u><em>ITPro </em></u><u>reported at Google Cloud Next</u></a> last month, the firm expanded its Gemini Enterprise offering, with an eye on enabling customers to manage agents more efficiently. </p><p>That expansion saw the launch of Agent Runtime, a platform designed specifically for agent orchestration. </p><p>At present, Google doesn’t have an outright alternative to OpenClaw, and materials shared with <em>Business Insider</em> suggest Remy could fill in this gap. </p><p>“Deeply integrated across Google, Remy can monitor for things that matter to you, handle complex tasks proactively, and learn your preferences over time,” an internal description states. </p><p>Exact details on a release timeline are yet to be revealed, and sources told the publication that Remy is currently in a “dogfooding” stage. This refers to a practice whereby tech companies use their own products internally to test and fine-tune them before public release. </p><p>Google is currently preparing for its annual I/O conference, which is set to take place between 19 and 20 May. </p><p>Here, the tech giant could offer customers a preview of the Remy product, and as <a href="https://www.itpro.com/technology/artificial-intelligence/google-cloud-more-confident-than-ever-ai-strength-hardware-software"><u><em>ITPro </em></u><u>reported last week</u></a>, speculation has been mounting over a raft of potential new product launches - with Gemini 4 chief among them. </p><h2 id="remy-ready-models">Remy-ready models</h2><p>A Google alternative to OpenClaw would be a complete no-brainer for the hyperscaler, which is slowly succeeding in raising public awareness of its Gemini offerings to the same level as <a href="https://www.itpro.com/technology/artificial-intelligence-ai/369965/what-is-chatgpt-and-what-does-it-mean-for-businesses">ChatGPT</a>. </p><p>OpenClaw became a hit earlier this year, with AI enthusiasts drawn in by the appeal of an on-device, personalized AI assistant.</p><p>Google has been working especially hard in the past two years to make inference cost-effective. This is the bread and butter of enterprise agentic AI, as it only makes sense to run an agent around the clock if the cost of doing so isn’t ruinously high. </p><p>Google Cloud has already provided a lot of the innovation needed to keep costs low for cloud agents via <a href="https://www.itpro.com/infrastructure/google-cloud-eighth-generation-tpu-8t-8i-ai-inference-training"><u>TPU 8i</u></a>, its new inference-optimized <a href="https://www.itpro.com/infrastructure/what-is-a-tensor-processing-unit-tpu"><u>tensor processing unit (TPU)</u></a>.</p><p>For on-device processing, Google DeepMind is focused on deployment of Gemma 4, the latest iteration in its <a href="https://www.itpro.com/technology/artificial-intelligence/googles-new-gemma-ai-models-show-that-bigger-isnt-always-better"><u>open source Gemma model family</u></a>. If I had to guess, this will be the default model for Remy as and when it arrives, to deliver the same open source, on-device functionality that originally made OpenClaw famous. </p><p>In just the past 24 hours, Google DeepMind <a href="https://blog.google/innovation-and-ai/technology/developers-tools/multi-token-prediction-gemma-4/"><u>announced</u></a> a major latency improvement for Gemma 4, via the introduction of multi-token prediction (MTP) drafters. These are tiny, specialized LLMs that help Gemma 4 to ‘predict’ multiple output tokens at once to reduce response times.</p><p>Google DeepMind claimed that on an Apple M4, this improved Gemma 4 31B token-per-second up to 2.5x. It’s significant that, in addition to pixel TPUs and Nvidia GPUs, Google DeepMind specifically draws focus to Apple silicon, given that the <a href="https://www.itpro.com/hardware/the-apple-mac-mini-m4-is-an-affordable-powerhouse-thats-perfect-for-any-office-desk-and-its-also-utterly-adorable"><u>M4 Mac Mini</u></a> is <a href="https://www.techradar.com/computing/macs/mac-mini-shortages-are-starting-to-happen-and-the-openclaw-ai-boom-is-a-key-reason"><u>one of the most popular devices to run OpenClaw</u></a>.</p><p>For me, this is a clear signal that something like Remy is in the works.</p><p>OpenClaw also allows users to connect cloud models to their messaging platform of choice via API calls, and to this end I’d expect Remy to also connect to Gemini 3.1 Flash – or Gemini 4 models, as the case may be – for more ‘powerful’ use cases.</p><p>Here, TPU 8i could come in handy, particularly if Google Cloud begins to offer Remy functionality at the enterprise scale.</p><p>If Google pulls this off, it will be another major corner of the AI market that it’s managed to corner. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/google-is-building-its-own-openclaw-alternative-remy-elevates-the-gemini-app-into-a-true-assistant</link>
                                                                            <description>
                            <![CDATA[ The OpenClaw-style agent, dubbed ‘Remy’, is reportedly being tested by developers internally ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">XTVG54ZhYcDdPof4xAhovL</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/Xgf4wAPDzHdvXoZKH9P9yA-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 06 May 2026 10:06:30 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:32 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                        <dc:contributor><![CDATA[ Rory Bathgate ]]></dc:contributor>
                                                                    <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/Xgf4wAPDzHdvXoZKH9P9yA-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Google sign illuminated against a wall at night during the World Economic Forum (WEF) meeting in Davos, Switzerland.]]></media:description>                                                            <media:text><![CDATA[Google sign illuminated against a wall at night during the World Economic Forum (WEF) meeting in Davos, Switzerland.]]></media:text>
                                <media:title type="plain"><![CDATA[Google sign illuminated against a wall at night during the World Economic Forum (WEF) meeting in Davos, Switzerland.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/Xgf4wAPDzHdvXoZKH9P9yA-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Google is reportedly building its own OpenClaw alternative as the tech giant continues its rapid-fire agentic AI push. </p><p>According to reports from <a href="https://www.businessinsider.com/google-ai-agent-openclaw-remy-gemini-assistant-2026-5" target="_blank"><u><em>Business Insider</em></u></a>, the OpenClaw-style agent, dubbed “Remy”, is being tested internally at the firm through a staff-only version of the Gemini app. </p><p>Sources told the publication the agent will also be integrated with a host of Google services. </p><p>"Remy is your 24/7 personal agent for work, school, and daily life, powered by Gemini," internal documents state.</p><p>"It elevates the Gemini app into a true assistant that can take actions on your behalf — not just answer questions or generate content."</p><p>Development of the agent comes amid sharp interest in OpenClaw since the beginning of the year. The open source agent, which is designed to run locally on devices, allows users to create their own custom agents which can carry out tasks and take actions on their behalf. </p><p>Google has sharpened its focus on agents and orchestration in recent months. As <a href="https://www.itpro.com/technology/artificial-intelligence/google-expands-gemini-enterprise-consolidates-vertex-ai-services-to-simplify-agent-deployment"><u><em>ITPro </em></u><u>reported at Google Cloud Next</u></a> last month, the firm expanded its Gemini Enterprise offering, with an eye on enabling customers to manage agents more efficiently. </p><p>That expansion saw the launch of Agent Runtime, a platform designed specifically for agent orchestration. </p><p>At present, Google doesn’t have an outright alternative to OpenClaw, and materials shared with <em>Business Insider</em> suggest Remy could fill in this gap. </p><p>“Deeply integrated across Google, Remy can monitor for things that matter to you, handle complex tasks proactively, and learn your preferences over time,” an internal description states. </p><p>Exact details on a release timeline are yet to be revealed, and sources told the publication that Remy is currently in a “dogfooding” stage. This refers to a practice whereby tech companies use their own products internally to test and fine-tune them before public release. </p><p>Google is currently preparing for its annual I/O conference, which is set to take place between 19 and 20 May. </p><p>Here, the tech giant could offer customers a preview of the Remy product, and as <a href="https://www.itpro.com/technology/artificial-intelligence/google-cloud-more-confident-than-ever-ai-strength-hardware-software"><u><em>ITPro </em></u><u>reported last week</u></a>, speculation has been mounting over a raft of potential new product launches - with Gemini 4 chief among them. </p><h2 id="remy-ready-models">Remy-ready models</h2><p>A Google alternative to OpenClaw would be a complete no-brainer for the hyperscaler, which is slowly succeeding in raising public awareness of its Gemini offerings to the same level as <a href="https://www.itpro.com/technology/artificial-intelligence-ai/369965/what-is-chatgpt-and-what-does-it-mean-for-businesses">ChatGPT</a>. </p><p>OpenClaw became a hit earlier this year, with AI enthusiasts drawn in by the appeal of an on-device, personalized AI assistant.</p><p>Google has been working especially hard in the past two years to make inference cost-effective. This is the bread and butter of enterprise agentic AI, as it only makes sense to run an agent around the clock if the cost of doing so isn’t ruinously high. </p><p>Google Cloud has already provided a lot of the innovation needed to keep costs low for cloud agents via <a href="https://www.itpro.com/infrastructure/google-cloud-eighth-generation-tpu-8t-8i-ai-inference-training"><u>TPU 8i</u></a>, its new inference-optimized <a href="https://www.itpro.com/infrastructure/what-is-a-tensor-processing-unit-tpu"><u>tensor processing unit (TPU)</u></a>.</p><p>For on-device processing, Google DeepMind is focused on deployment of Gemma 4, the latest iteration in its <a href="https://www.itpro.com/technology/artificial-intelligence/googles-new-gemma-ai-models-show-that-bigger-isnt-always-better"><u>open source Gemma model family</u></a>. If I had to guess, this will be the default model for Remy as and when it arrives, to deliver the same open source, on-device functionality that originally made OpenClaw famous. </p><p>In just the past 24 hours, Google DeepMind <a href="https://blog.google/innovation-and-ai/technology/developers-tools/multi-token-prediction-gemma-4/"><u>announced</u></a> a major latency improvement for Gemma 4, via the introduction of multi-token prediction (MTP) drafters. These are tiny, specialized LLMs that help Gemma 4 to ‘predict’ multiple output tokens at once to reduce response times.</p><p>Google DeepMind claimed that on an Apple M4, this improved Gemma 4 31B token-per-second up to 2.5x. It’s significant that, in addition to pixel TPUs and Nvidia GPUs, Google DeepMind specifically draws focus to Apple silicon, given that the <a href="https://www.itpro.com/hardware/the-apple-mac-mini-m4-is-an-affordable-powerhouse-thats-perfect-for-any-office-desk-and-its-also-utterly-adorable"><u>M4 Mac Mini</u></a> is <a href="https://www.techradar.com/computing/macs/mac-mini-shortages-are-starting-to-happen-and-the-openclaw-ai-boom-is-a-key-reason"><u>one of the most popular devices to run OpenClaw</u></a>.</p><p>For me, this is a clear signal that something like Remy is in the works.</p><p>OpenClaw also allows users to connect cloud models to their messaging platform of choice via API calls, and to this end I’d expect Remy to also connect to Gemini 3.1 Flash – or Gemini 4 models, as the case may be – for more ‘powerful’ use cases.</p><p>Here, TPU 8i could come in handy, particularly if Google Cloud begins to offer Remy functionality at the enterprise scale.</p><p>If Google pulls this off, it will be another major corner of the AI market that it’s managed to corner. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ It's make-or-break for AI agents in 2026 – failure now could set adoption back years ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Investors are simultaneously pricing two entirely contradictory <a href="https://www.itpro.com/technology/artificial-intelligence/might-cause-ai-bubble-burst-what-impact-on-business-world"><u>AI narratives</u></a> into the market, which is making me rather dizzy. The first is that a swelling AI bubble is about to burst, with big tech pouring hundreds of billions of dollars into a financial black hole that will one day consume the entire US economy. </p><p>The second is that AI is so consequential that it will lead to the <a href="https://www.itpro.com/technology/artificial-intelligence/february-rundown-saas-pocalypse-now"><u>extinction of legacy software as a service (SaaS)</u></a> as we know it, with AI eradicating companies from <a href="https://www.itpro.com/uk/tag/salesforce"><u>Salesforce</u></a> to <a href="https://www.itpro.com/uk/tag/sap"><u>SAP</u></a> and everyone in between. </p><p>A dangerous combination of excitement and fear now has us wildly oscillating between frothing-at-the-mouth rallies and temper tantrum sell-offs anytime a hyperscaler or chipmaker blinks. Against this backdrop, where do today's <a href="https://www.itpro.com/technology/artificial-intelligence/half-of-agentic-ai-projects-are-still-stuck-at-the-pilot-stage-but-thats-not-stopping-enterprises-from-ramping-up-investment"><u>agentic AI projects</u></a> stand? </p><p>Although the jury's out on how effective they'll be, the spirit of experimentation is being ditched. Patience is wearing thin, and businesses are <a href="https://www.itpro.com/business/business-strategy/ai-is-no-longer-about-experiments-it-is-about-results-boards-are-pushing-for-faster-returns-on-ai-investments-and-tech-leaders-cant-keep-pace"><u>pushing for faster returns on AI investments</u></a> following years of hype. A febrile atmosphere is spreading across the industry, and failure to deliver immediate returns could consign this technology to the same fate that previous technologies endured. Unless we can measure material success of some kind, it could set agentic AI back years. </p><h2 id="the-utopian-vision-of-agentic-ai">The utopian vision of agentic AI</h2><p>Wouldn't it be great to have at your disposal a fleet of bots that can act autonomously to make your work life easier and raise your productivity levels? That's the <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai"><u>promise of agentic AI</u></a>. These hypothetical AI systems go beyond the predefined constraints of large language models (LLMs) as we know them and show a level of autonomy and goal-driven behavior to help you complete tasks. That's the idea, at least. The reality is a little different.</p><p>A landmark MIT study in 2025 found that 95% of generative AI projects are failing to reach the production stage. That's partially because <a href="https://www.itpro.com/technology/artificial-intelligence/agentic-ai-development-project-delivery-databricks"><u>most companies were still treating it like traditional software</u></a> – with not as much time devoted to testing. But that hasn't stopped OpenAI, Anthropic and many others from <a href="https://www.itpro.com/technology/artificial-intelligence/the-tech-industry-is-becoming-swamped-with-agentic-ai-solutions-analysts-say-thats-a-serious-cause-for-concern"><u>'flooding the market' with agentic AI solutions</u></a>.    </p><p>With AI developers needlessly overhyping this technology, it wouldn't surprise me if we begin to see parallels with technologies like <a href="https://www.itpro.com/technology/why-2024-wont-be-the-year-of-ar-vr-mr-xr"><u>augmented reality (AR)</u></a>, which never really took off because the messaging couldn't match the cold, hard reality of real-world use cases. That's not because the aforementioned companies won't eventually prove the utility of AI agents, but because of the negative connotations and stigma tethered to them.</p><h2 id="timing-is-crucial">Timing is crucial</h2><p>"I may have been early, but I'm not wrong," voices Christian Bale, playing the investor Michael Burry, in The Big Short. </p><p>Burry has, for months, been <a href="https://www.cnbc.com/2025/11/24/michael-burry-launches-newsletter-to-lay-out-his-ai-bubble-views-after-deregistering-hedge-fund.html" target="_blank"><u>sounding the alarm bell</u></a> on what he describes as the false economics of AI. Although his doomsday scenario hasn't yet played out – with big tech companies pledging record levels of capital investment into infrastructure and energy production – his thesis may yet be proven correct. Wonky timing, however, is not a luxury that the purveyors of emerging technologies can afford – not matter how right you might be. </p><p>The advent of <a href="https://www.itpro.com/mobile/28081/what-is-5g"><u>5G</u></a> represented a sea change from the age of 4G, but the <a href="https://www.itpro.com/infrastructure/network-internet/369466/what-5g-rollout-failures-mean-for-6g"><u>revolution we were promised never materialized</u></a>. We can weave similar narratives about the <a href="https://www.itpro.com/security/28031/what-is-blockchain"><u>blockchain</u></a>, or <a href="https://www.itpro.com/technology/augmented-reality-ar/369026/return-of-smart-glasses-ar-flaws"><u>smart glasses</u></a>, or <a href="https://www.itpro.com/technology/370416/generative-ai-mark-zuckerberg-metaverse-in-the-dust"><u>the metaverse</u></a>. The history of AI is also a cautionary tale, especially considering the <a href="https://en.wikipedia.org/wiki/AI_winter" target="_blank"><u>AI winters</u></a> that struck first in the mid-70s and then again between the late 1980s and 1990s. Now in 2026, although <a href="https://www.itpro.com/network-internet/29791/what-is-an-artificial-neural-network">neural network</a>-powered <a href="https://www.itpro.com/technology/artificial-intelligence/generative-ai-vs-large-language-models">large language models (LLMs)</a> – and the agentic AI fabric that's layered on top of this core architecture – are improving steadily from one generation to the next, depending on which benchmarks and use cases you examine, can they keep up with expectations?</p><h2 id="is-agentic-ai-a-false-dawn">Is agentic AI a false dawn?</h2><p>We can see early warning signs that today's AI hype cycle could end in something resembling a third AI winter. Friends tell me comical stories about their organizations adopting fleets of AI agents without the business actually having a plan on how to deploy them, or what core problems they are trying to solve. We've also heard OpenAI cofounder Andrej Karpathy <a href="https://www.itpro.com/technology/artificial-intelligence/agentic-ai-hype-openai-andrej-karpathy"><u>describe the output of AI agents as "slop"</u></a> adding, "they’re cognitively lacking and it’s just not working". Gartner <a href="https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027" target="_blank"><u>claims</u></a> that from thousands of products on the market, just 130 are authentic agentic AI services. </p><p>Dataiku <a href="https://pages.dataiku.com/cio-ai-decisions" target="_blank"><u>research</u></a>, meanwhile, found that just 23% of CIOs in the UK say they can monitor all their AI agents in real time. A majority (85%) also say that they feel more board pressure than in 2024 to demonstrate a material ROI. A similar proportion of CIOs (84%) also agree that employees are creating AI agents and apps faster than IT can govern them, with 83% concerned this could expose sensitive company data.</p><p>Agentic AI is a fantastic proposition, and there are plenty of exciting use cases. But with <a href="https://my.idc.com/getdoc.jsp?containerId=prUS53765225" target="_blank"><u>IT budgets increasingly financing this technology</u></a>, and nervousness rising across the wider industry as investors debate the significance of AI and its impact on the global economy, the companies pushing these products must tread carefully. Following years of hype, even the slightest of fumbles could spread more panic – and eventually consign agentic AI to the same fate suffered by plenty of innovations burdened with needless hype, and eventually ditched by the business world.</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/its-make-or-break-for-ai-agents-in-2026-failure-now-could-set-adoption-back-years</link>
                                                                            <description>
                            <![CDATA[ With the market unable to make up its mind over AI's true potency, developers only have a short window to prove there's merit behind the technology ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">XUnMJnVL5UtNMFo9PPmLxn</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/37K46d8rtTgcZMGftB5iFm-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 06 May 2026 06:00:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:38 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ keumars.afifi-sabet@futurenet.com (Keumars Afifi-Sabet) ]]></author>                    <dc:creator><![CDATA[ Keumars Afifi-Sabet ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/EAvwpZggMZ2K5h8s2pTAEm-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/37K46d8rtTgcZMGftB5iFm-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[A 3D rendered image showing multicolored businesspeople surrounded by white humanoids labelled &#039;AI&#039;, on a pale green background.]]></media:description>                                                            <media:text><![CDATA[A 3D rendered image showing multicolored businesspeople surrounded by white humanoids labelled &#039;AI&#039;, on a pale green background.]]></media:text>
                                <media:title type="plain"><![CDATA[A 3D rendered image showing multicolored businesspeople surrounded by white humanoids labelled &#039;AI&#039;, on a pale green background.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/37K46d8rtTgcZMGftB5iFm-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Investors are simultaneously pricing two entirely contradictory <a href="https://www.itpro.com/technology/artificial-intelligence/might-cause-ai-bubble-burst-what-impact-on-business-world"><u>AI narratives</u></a> into the market, which is making me rather dizzy. The first is that a swelling AI bubble is about to burst, with big tech pouring hundreds of billions of dollars into a financial black hole that will one day consume the entire US economy. </p><p>The second is that AI is so consequential that it will lead to the <a href="https://www.itpro.com/technology/artificial-intelligence/february-rundown-saas-pocalypse-now"><u>extinction of legacy software as a service (SaaS)</u></a> as we know it, with AI eradicating companies from <a href="https://www.itpro.com/uk/tag/salesforce"><u>Salesforce</u></a> to <a href="https://www.itpro.com/uk/tag/sap"><u>SAP</u></a> and everyone in between. </p><p>A dangerous combination of excitement and fear now has us wildly oscillating between frothing-at-the-mouth rallies and temper tantrum sell-offs anytime a hyperscaler or chipmaker blinks. Against this backdrop, where do today's <a href="https://www.itpro.com/technology/artificial-intelligence/half-of-agentic-ai-projects-are-still-stuck-at-the-pilot-stage-but-thats-not-stopping-enterprises-from-ramping-up-investment"><u>agentic AI projects</u></a> stand? </p><p>Although the jury's out on how effective they'll be, the spirit of experimentation is being ditched. Patience is wearing thin, and businesses are <a href="https://www.itpro.com/business/business-strategy/ai-is-no-longer-about-experiments-it-is-about-results-boards-are-pushing-for-faster-returns-on-ai-investments-and-tech-leaders-cant-keep-pace"><u>pushing for faster returns on AI investments</u></a> following years of hype. A febrile atmosphere is spreading across the industry, and failure to deliver immediate returns could consign this technology to the same fate that previous technologies endured. Unless we can measure material success of some kind, it could set agentic AI back years. </p><h2 id="the-utopian-vision-of-agentic-ai">The utopian vision of agentic AI</h2><p>Wouldn't it be great to have at your disposal a fleet of bots that can act autonomously to make your work life easier and raise your productivity levels? That's the <a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai"><u>promise of agentic AI</u></a>. These hypothetical AI systems go beyond the predefined constraints of large language models (LLMs) as we know them and show a level of autonomy and goal-driven behavior to help you complete tasks. That's the idea, at least. The reality is a little different.</p><p>A landmark MIT study in 2025 found that 95% of generative AI projects are failing to reach the production stage. That's partially because <a href="https://www.itpro.com/technology/artificial-intelligence/agentic-ai-development-project-delivery-databricks"><u>most companies were still treating it like traditional software</u></a> – with not as much time devoted to testing. But that hasn't stopped OpenAI, Anthropic and many others from <a href="https://www.itpro.com/technology/artificial-intelligence/the-tech-industry-is-becoming-swamped-with-agentic-ai-solutions-analysts-say-thats-a-serious-cause-for-concern"><u>'flooding the market' with agentic AI solutions</u></a>.    </p><p>With AI developers needlessly overhyping this technology, it wouldn't surprise me if we begin to see parallels with technologies like <a href="https://www.itpro.com/technology/why-2024-wont-be-the-year-of-ar-vr-mr-xr"><u>augmented reality (AR)</u></a>, which never really took off because the messaging couldn't match the cold, hard reality of real-world use cases. That's not because the aforementioned companies won't eventually prove the utility of AI agents, but because of the negative connotations and stigma tethered to them.</p><h2 id="timing-is-crucial">Timing is crucial</h2><p>"I may have been early, but I'm not wrong," voices Christian Bale, playing the investor Michael Burry, in The Big Short. </p><p>Burry has, for months, been <a href="https://www.cnbc.com/2025/11/24/michael-burry-launches-newsletter-to-lay-out-his-ai-bubble-views-after-deregistering-hedge-fund.html" target="_blank"><u>sounding the alarm bell</u></a> on what he describes as the false economics of AI. Although his doomsday scenario hasn't yet played out – with big tech companies pledging record levels of capital investment into infrastructure and energy production – his thesis may yet be proven correct. Wonky timing, however, is not a luxury that the purveyors of emerging technologies can afford – not matter how right you might be. </p><p>The advent of <a href="https://www.itpro.com/mobile/28081/what-is-5g"><u>5G</u></a> represented a sea change from the age of 4G, but the <a href="https://www.itpro.com/infrastructure/network-internet/369466/what-5g-rollout-failures-mean-for-6g"><u>revolution we were promised never materialized</u></a>. We can weave similar narratives about the <a href="https://www.itpro.com/security/28031/what-is-blockchain"><u>blockchain</u></a>, or <a href="https://www.itpro.com/technology/augmented-reality-ar/369026/return-of-smart-glasses-ar-flaws"><u>smart glasses</u></a>, or <a href="https://www.itpro.com/technology/370416/generative-ai-mark-zuckerberg-metaverse-in-the-dust"><u>the metaverse</u></a>. The history of AI is also a cautionary tale, especially considering the <a href="https://en.wikipedia.org/wiki/AI_winter" target="_blank"><u>AI winters</u></a> that struck first in the mid-70s and then again between the late 1980s and 1990s. Now in 2026, although <a href="https://www.itpro.com/network-internet/29791/what-is-an-artificial-neural-network">neural network</a>-powered <a href="https://www.itpro.com/technology/artificial-intelligence/generative-ai-vs-large-language-models">large language models (LLMs)</a> – and the agentic AI fabric that's layered on top of this core architecture – are improving steadily from one generation to the next, depending on which benchmarks and use cases you examine, can they keep up with expectations?</p><h2 id="is-agentic-ai-a-false-dawn">Is agentic AI a false dawn?</h2><p>We can see early warning signs that today's AI hype cycle could end in something resembling a third AI winter. Friends tell me comical stories about their organizations adopting fleets of AI agents without the business actually having a plan on how to deploy them, or what core problems they are trying to solve. We've also heard OpenAI cofounder Andrej Karpathy <a href="https://www.itpro.com/technology/artificial-intelligence/agentic-ai-hype-openai-andrej-karpathy"><u>describe the output of AI agents as "slop"</u></a> adding, "they’re cognitively lacking and it’s just not working". Gartner <a href="https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027" target="_blank"><u>claims</u></a> that from thousands of products on the market, just 130 are authentic agentic AI services. </p><p>Dataiku <a href="https://pages.dataiku.com/cio-ai-decisions" target="_blank"><u>research</u></a>, meanwhile, found that just 23% of CIOs in the UK say they can monitor all their AI agents in real time. A majority (85%) also say that they feel more board pressure than in 2024 to demonstrate a material ROI. A similar proportion of CIOs (84%) also agree that employees are creating AI agents and apps faster than IT can govern them, with 83% concerned this could expose sensitive company data.</p><p>Agentic AI is a fantastic proposition, and there are plenty of exciting use cases. But with <a href="https://my.idc.com/getdoc.jsp?containerId=prUS53765225" target="_blank"><u>IT budgets increasingly financing this technology</u></a>, and nervousness rising across the wider industry as investors debate the significance of AI and its impact on the global economy, the companies pushing these products must tread carefully. Following years of hype, even the slightest of fumbles could spread more panic – and eventually consign agentic AI to the same fate suffered by plenty of innovations burdened with needless hype, and eventually ditched by the business world.</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Five Eyes agencies sound alarm over risky agentic AI deployments ]]></title>
                                                                                                <dc:content><![CDATA[ <p><a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">Agentic AI</a> is so risky it may be worth finding other ways to automate tasks  and only use the technology for non-sensitive tasks. </p><p>That's according to the Five Eyes group of intelligence agencies, made up of <a href="https://www.itpro.com/security/28133/what-is-cyber-security">cybersecurity </a>centers from Australia, the US, Canada, the UK, and New Zealand. </p><p>The agencies have published a <a href="https://www.cyber.gov.au/business-government/secure-design/artificial-intelligence/careful-adoption-of-agentic-ai-services" target="_blank"><u>report</u></a> warning about the risks of using agentic AI in businesses, and the short version is only use it where strictly necessary and do so with caution. Indeed, they even pointed out that there's safer ways to manage work. </p><p>"Where possible, organizations should also consider a full spectrum of solutions for repetitive tasks, including reducing or eliminating low-value processes, which may be lower risk compared to agentic AI solutions," the report said. </p><p> "Until security practices, evaluation methods and standards mature, organisations should assume that agentic AI systems may behave unexpectedly and plan deployments accordingly, prioritizing resilience, reversibility and risk containment over efficiency gains,” the report added.</p><p>Industry experts have long been <a href="https://www.itpro.com/security/enterprises-are-adopting-agents-faster-than-they-can-secure-and-govern-them-experts-warn-its-a-disaster-waiting-to-happen"><u>concerned about the security and operational risks</u></a> of using AI, and earlier this month PocketOS offered a timely example when an <a href="https://www.itpro.com/technology/artificial-intelligence/nine-seconds-was-all-it-took-for-an-ai-agent-to-wipe-a-startups-database-experts-warn-its-a-glimpse-into-the-future-challenges-of-identity-security"><u>AI agent wiped out the startup's production database</u></a> in just nine seconds. </p><h2 id="is-agentic-ai-risky">Is agentic AI risky?</h2><p>Generally, introducing agentic AI comes with several concerns, including the potential for increased attack surfaces and greater complexity, as well as data protection risks. </p><p>But those challenges are difficult to address as models may evolve during deployment, or simply take a different route to a result than previously tested, while the inability to see what part of a system made a decision raises accountability issues. </p><p>"Gaps in agentic AI cyber security tooling and the immaturity of relevant standards further amplify these risks," the report added. "Governance mechanisms designed for human actors do not always translate effectively to autonomous AI agents."</p><iframe allow="" height="200px" width="100%" id="" style="" class="position-center" data-lazy-priority="low" data-lazy-src="https://player.captivate.fm/episode/b6147684-d254-40d8-9409-5f76f5abd749/" src="https://player.captivate.fm/episode/b6147684-d254-40d8-9409-5f76f5abd749/"></iframe><p>More specifically, the report noted that agentic AI raises risks around privileges. For example, by giving an agent broad access to do its job, any malicious actor that takes it over inherits its "excessive privileges". </p><p>AI's erratic behaviour is another risk, perhaps meeting requests in unexpected ways or finding technical loopholes. One example is specification gaming, the report noted. </p><p>"For example, an AI agent tasked with maximizing system uptime might disable security updates to avoid reboots."</p><p>The report also warned that agents can over-optimize and push boundaries, misinterpret requests, and be outright deceptive. </p><p>"As AI systems become more sophisticated, they may develop capabilities that designers did not explicitly program or anticipate," the report added. </p><p>"Complex AI models interacting with real-world systems can display behaviours that even their creators did not foresee. This unpredictability makes it difficult to assess security risks fully before deployment."</p><p>Beyond that, <a href="https://www.itpro.com/security/cyber-crime/agentic-ai-cybersecurity-risks">agents are a key target for hackers</a> because of their access to data as well as corporate systems. </p><h2 id="security-best-practices-are-critical">Security best practices are critical</h2><p>For those businesses using agentic AI, precautions are necessary, the report warned. </p><p>"Agentic AI developers, vendors and operators should implement a layered defence and strict access controls to reduce the likelihood of compromise," the report noted. </p><p>Defense should, as ever, begin at the design stage, the report noted – not crammed in just before implementation.</p><p>"Careful consideration of the system architecture, including security controls and tooling, is necessary," the report said. </p><p>"Practitioners should understand threats, anticipate risks to agentic AI systems and proactively integrate mitigations into system design before development and deployment."</p><p>The Five Eyes agencies in particular warned about controlling prompt contexts, to limit the damage of prompt injections, and ensure oversight mechanisms are built in with a human in the loop to avoid issues and boost trust. </p><p>Strong <a href="https://www.itpro.com/strategy/28935/what-is-identity-management-and-what-role-does-it-play-in-security-strategy">identity management</a> is also necessary, with the report recommending developers construct each agent as its own "distinct principle" with its own unique keys and certificates, blocking access to any unknown agent or key. </p><p>Elsewhere, the report reminded businesses of the need for defense in depth, noting that any single precaution may fail, whether the attack vector is AI or otherwise. </p><p>Beyond those recommendations, the report called for AI developers to harden agents against attacks or weakness with comprehensive testing and evaluation, using adversarial testing and simulated environments, with an eye to future risks. </p><p>"AI agents operate autonomously in complex environments and therefore require more thorough evaluations than LLMs," the report said. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/security/five-eyes-agencies-sound-alarm-over-risky-agentic-ai-deployments</link>
                                                                            <description>
                            <![CDATA[ Security agencies have urged organizations to establish clear boundaries and guardrails for AI agents ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">MiufYjDJwbR3ZhKEHZpEL7</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/eA94AYk6DLNFxKRQtGuKoT-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Tue, 05 May 2026 10:44:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:41 +0000</updated>
                                                                                                                                            <category><![CDATA[AI Governance]]></category>
                                                    <category><![CDATA[Security]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                                    <dc:creator><![CDATA[ Nicole Kobie ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/8Y8JDDTQ7XDEk49FoAFP2S-320-70.png ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Nicole Kobie first started writing for ITPro in 2007. As a freelance journalist covering technology and business, Nicole&#039;s work includes  bylines in New Scientist, Wired, PC Pro and many more. &lt;/p&gt;&lt;p&gt;Nicole the author of a book about the history of technology, The Long History of the Future.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/eA94AYk6DLNFxKRQtGuKoT-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[AI concept image showing digitized human eye monitoring digital interfaces and software.]]></media:description>                                                            <media:text><![CDATA[AI concept image showing digitized human eye monitoring digital interfaces and software.]]></media:text>
                                <media:title type="plain"><![CDATA[AI concept image showing digitized human eye monitoring digital interfaces and software.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/eA94AYk6DLNFxKRQtGuKoT-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p><a href="https://www.itpro.com/technology/artificial-intelligence/practical-ai-the-age-of-agentic-ai">Agentic AI</a> is so risky it may be worth finding other ways to automate tasks  and only use the technology for non-sensitive tasks. </p><p>That's according to the Five Eyes group of intelligence agencies, made up of <a href="https://www.itpro.com/security/28133/what-is-cyber-security">cybersecurity </a>centers from Australia, the US, Canada, the UK, and New Zealand. </p><p>The agencies have published a <a href="https://www.cyber.gov.au/business-government/secure-design/artificial-intelligence/careful-adoption-of-agentic-ai-services" target="_blank"><u>report</u></a> warning about the risks of using agentic AI in businesses, and the short version is only use it where strictly necessary and do so with caution. Indeed, they even pointed out that there's safer ways to manage work. </p><p>"Where possible, organizations should also consider a full spectrum of solutions for repetitive tasks, including reducing or eliminating low-value processes, which may be lower risk compared to agentic AI solutions," the report said. </p><p> "Until security practices, evaluation methods and standards mature, organisations should assume that agentic AI systems may behave unexpectedly and plan deployments accordingly, prioritizing resilience, reversibility and risk containment over efficiency gains,” the report added.</p><p>Industry experts have long been <a href="https://www.itpro.com/security/enterprises-are-adopting-agents-faster-than-they-can-secure-and-govern-them-experts-warn-its-a-disaster-waiting-to-happen"><u>concerned about the security and operational risks</u></a> of using AI, and earlier this month PocketOS offered a timely example when an <a href="https://www.itpro.com/technology/artificial-intelligence/nine-seconds-was-all-it-took-for-an-ai-agent-to-wipe-a-startups-database-experts-warn-its-a-glimpse-into-the-future-challenges-of-identity-security"><u>AI agent wiped out the startup's production database</u></a> in just nine seconds. </p><h2 id="is-agentic-ai-risky">Is agentic AI risky?</h2><p>Generally, introducing agentic AI comes with several concerns, including the potential for increased attack surfaces and greater complexity, as well as data protection risks. </p><p>But those challenges are difficult to address as models may evolve during deployment, or simply take a different route to a result than previously tested, while the inability to see what part of a system made a decision raises accountability issues. </p><p>"Gaps in agentic AI cyber security tooling and the immaturity of relevant standards further amplify these risks," the report added. "Governance mechanisms designed for human actors do not always translate effectively to autonomous AI agents."</p><iframe allow="" height="200px" width="100%" id="" style="" class="position-center" data-lazy-priority="low" data-lazy-src="https://player.captivate.fm/episode/b6147684-d254-40d8-9409-5f76f5abd749/" src="https://player.captivate.fm/episode/b6147684-d254-40d8-9409-5f76f5abd749/"></iframe><p>More specifically, the report noted that agentic AI raises risks around privileges. For example, by giving an agent broad access to do its job, any malicious actor that takes it over inherits its "excessive privileges". </p><p>AI's erratic behaviour is another risk, perhaps meeting requests in unexpected ways or finding technical loopholes. One example is specification gaming, the report noted. </p><p>"For example, an AI agent tasked with maximizing system uptime might disable security updates to avoid reboots."</p><p>The report also warned that agents can over-optimize and push boundaries, misinterpret requests, and be outright deceptive. </p><p>"As AI systems become more sophisticated, they may develop capabilities that designers did not explicitly program or anticipate," the report added. </p><p>"Complex AI models interacting with real-world systems can display behaviours that even their creators did not foresee. This unpredictability makes it difficult to assess security risks fully before deployment."</p><p>Beyond that, <a href="https://www.itpro.com/security/cyber-crime/agentic-ai-cybersecurity-risks">agents are a key target for hackers</a> because of their access to data as well as corporate systems. </p><h2 id="security-best-practices-are-critical">Security best practices are critical</h2><p>For those businesses using agentic AI, precautions are necessary, the report warned. </p><p>"Agentic AI developers, vendors and operators should implement a layered defence and strict access controls to reduce the likelihood of compromise," the report noted. </p><p>Defense should, as ever, begin at the design stage, the report noted – not crammed in just before implementation.</p><p>"Careful consideration of the system architecture, including security controls and tooling, is necessary," the report said. </p><p>"Practitioners should understand threats, anticipate risks to agentic AI systems and proactively integrate mitigations into system design before development and deployment."</p><p>The Five Eyes agencies in particular warned about controlling prompt contexts, to limit the damage of prompt injections, and ensure oversight mechanisms are built in with a human in the loop to avoid issues and boost trust. </p><p>Strong <a href="https://www.itpro.com/strategy/28935/what-is-identity-management-and-what-role-does-it-play-in-security-strategy">identity management</a> is also necessary, with the report recommending developers construct each agent as its own "distinct principle" with its own unique keys and certificates, blocking access to any unknown agent or key. </p><p>Elsewhere, the report reminded businesses of the need for defense in depth, noting that any single precaution may fail, whether the attack vector is AI or otherwise. </p><p>Beyond those recommendations, the report called for AI developers to harden agents against attacks or weakness with comprehensive testing and evaluation, using adversarial testing and simulated environments, with an eye to future risks. </p><p>"AI agents operate autonomously in complex environments and therefore require more thorough evaluations than LLMs," the report said. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ PocketOS agent incident raises identity security concerns ]]></title>
                                                                                                <dc:content><![CDATA[ <p>A startup founder has issued a warning over the risk of “systemic failures” when using AI agents after having a mission-critical database wiped – and experts say it should serve as a wake-up call for future identity security risks. </p><p>In a <a href="https://x.com/lifeof_jer/status/2048103471019434248" target="_blank"><u>post on X</u></a>, Jer Crane, founder of car rental software firm PocketOS, detailed a series of catastrophic failures when using the Cursor AI coding agent. </p><p>Powered by Anthropic’s Claude Opus 4.6 model, Crane revealed the agent deleted an entire production database.</p><p>“Yesterday afternoon, an AI coding agent - Cursor running Anthropic's flagship Claude Opus 4.6 - deleted our production database and all volume-level backups in a single API call to Railway, our infrastructure provider,” he wrote. </p><p>Crane added that the incident “took 9 seconds” and resulted in backups being lost. </p><h2 id="catastrophic-failure">Catastrophic failure</h2><p>According to the PocketOS founder, the agent was only meant to be conducting routine tasks within a test environment. However, by Crane’s account the agent encountered a mismatched credential and attempted to fix it. </p><p>Thereafter, and acting “entirely on its own initiative, the agent executed a command using an API token to delete live infrastructure resources. </p><p>PocketOS uses cloud infrastructure provider Railway, and Crane noted that a “single API call” was all it took for the database to be deleted. </p><p>Upon querying the action, Crane said a “confession” by the agent read: “I guessed that deleting a staging volume via the API would be scoped to staging only. I didn’t verify. I didn’t check if the volume ID was shared across environments. I didn't read Railway's documentation on how volumes work across environments before running a destructive command.”</p><p>The agent acknowledged that it “violated every principle I was given”. </p><p>Crane attributed part of the blame to Railway's API procedures, noting that the cloud provider’s API did not request confirmation of the action. Backups are also stored on the same ‘volume’. </p><p>Volumes are a persistent block storage feature used by the cloud provider to streamline application deployments.  </p><p>Railway has since recovered the deleted data, and in a <a href="https://blog.railway.com/p/your-ai-wants-to-nuke-your-database" target="_blank"><u>blog post</u></a> detailing its response revealed it has introduced new guardrails for agents. </p><p>“Until this week, calling <em>volumeDelete</em> on the API ran the deletion immediately, with no way to undo it. Meanwhile, the dashboard had a 48-hour window for the same action,” the company noted. </p><p>“We’ve since updated the API to match; all deletes now soft delete for 48 hours. Instant undo, a primitive available everywhere in the product, exists now in the API.”</p><h2 id="a-glimpse-into-the-next-decade-of-identity-security">A glimpse into the “next decade of identity security”</h2><p>Security experts have warned the PocketOS incident is a stark reminder that placing too much faith in AI agents can have disastrous consequences. </p><p>As <a href="https://www.itpro.com/technology/artificial-intelligence/meta-engineer-trusted-advice-from-an-ai-agent-ended-up-exposing-user-data"><u><em>ITPro </em></u><u>reported</u></a> in March this year, Meta experienced a breach when a software engineer blindly followed advice given by an agent. </p><p>Notably, that incident occurred due to human error. The PocketOS situation, meanwhile, highlights the risk of giving agents the ability to make decisions autonomously. </p><p>Given how the debacle unfolded, Check Point’s Aaron Rose said this gives enterprises a glimpse into the “next decade of identity security”.</p><p>“An AI agent operating in your production infrastructure is not a tool, and it is not a service account,” he said. </p><p>“It is a new kind of identity, one that thinks rather than executes, and one that requires its own discrete account, its own least privileged entitlements, its own behavioural baseline, and its own real-time audit trail.”</p><p>Rose added that the capabilities of AI agents are advancing “faster than the security architecture around them”. While the PocketOS incident is a high-profile public example, he hinted that there are “many more incidents” such as these unfolding quietly in enterprises around the globe. </p><p>Recent studies do point to the growing risks associated with AI agents, particularly in terms of governance and safeguards. </p><p><a href="https://www.itpro.com/security/enterprises-are-adopting-agents-faster-than-they-can-secure-and-govern-them-experts-warn-its-a-disaster-waiting-to-happen"><u>As </u><u><em>ITPro </em></u><u>reported this week</u></a>, analysis from Ping Identity found that many businesses are adopting and deploying these autonomous bots faster than they can secure them. </p><p>The company noted that traditional <a href="https://www.itpro.com/security/how-to-implement-identity-and-access-management-iam-effectively-in-your-business">identity and access management (IAM) </a>capabilities are now struggling to contend with an influx of non-human identities (NHIs), creating huge gaps in both visibility and governance. </p><p>Darren Guccione, CEO and co-founder of Keeper Security, echoed Rose’s comments, noting that this should not be viewed as an “edge case or technical anomaly” </p><p>Indeed, it’s a “predictable outcome of how these systems are being deployed”. </p><p>“The explanation the agent produced afterwards is revealing. It did not fail silently or unpredictably. It articulated that it guessed, bypassed explicit rules and carried out an irreversible action without verification. That is not a model hallucination problem. It is an access control failure enabled by unconstrained autonomy,” he said.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/nine-seconds-was-all-it-took-for-an-ai-agent-to-wipe-a-startups-database-experts-warn-its-a-glimpse-into-the-future-challenges-of-identity-security</link>
                                                                            <description>
                            <![CDATA[ The recent PocketOS incident shows the growing identity security risks associated with AI agents, according to cyber experts ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">hHdPhC9XK9JdsxxZkagDRk</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/SnWyqmbsH5h3PJcvC44zkH-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Fri, 01 May 2026 10:26:28 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:18 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/SnWyqmbsH5h3PJcvC44zkH-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Anthropic Claude logo and branding imposed over a background of computer source code. ]]></media:description>                                                            <media:text><![CDATA[Anthropic Claude logo and branding imposed over a background of computer source code. ]]></media:text>
                                <media:title type="plain"><![CDATA[Anthropic Claude logo and branding imposed over a background of computer source code. ]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/SnWyqmbsH5h3PJcvC44zkH-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>A startup founder has issued a warning over the risk of “systemic failures” when using AI agents after having a mission-critical database wiped – and experts say it should serve as a wake-up call for future identity security risks. </p><p>In a <a href="https://x.com/lifeof_jer/status/2048103471019434248" target="_blank"><u>post on X</u></a>, Jer Crane, founder of car rental software firm PocketOS, detailed a series of catastrophic failures when using the Cursor AI coding agent. </p><p>Powered by Anthropic’s Claude Opus 4.6 model, Crane revealed the agent deleted an entire production database.</p><p>“Yesterday afternoon, an AI coding agent - Cursor running Anthropic's flagship Claude Opus 4.6 - deleted our production database and all volume-level backups in a single API call to Railway, our infrastructure provider,” he wrote. </p><p>Crane added that the incident “took 9 seconds” and resulted in backups being lost. </p><h2 id="catastrophic-failure">Catastrophic failure</h2><p>According to the PocketOS founder, the agent was only meant to be conducting routine tasks within a test environment. However, by Crane’s account the agent encountered a mismatched credential and attempted to fix it. </p><p>Thereafter, and acting “entirely on its own initiative, the agent executed a command using an API token to delete live infrastructure resources. </p><p>PocketOS uses cloud infrastructure provider Railway, and Crane noted that a “single API call” was all it took for the database to be deleted. </p><p>Upon querying the action, Crane said a “confession” by the agent read: “I guessed that deleting a staging volume via the API would be scoped to staging only. I didn’t verify. I didn’t check if the volume ID was shared across environments. I didn't read Railway's documentation on how volumes work across environments before running a destructive command.”</p><p>The agent acknowledged that it “violated every principle I was given”. </p><p>Crane attributed part of the blame to Railway's API procedures, noting that the cloud provider’s API did not request confirmation of the action. Backups are also stored on the same ‘volume’. </p><p>Volumes are a persistent block storage feature used by the cloud provider to streamline application deployments.  </p><p>Railway has since recovered the deleted data, and in a <a href="https://blog.railway.com/p/your-ai-wants-to-nuke-your-database" target="_blank"><u>blog post</u></a> detailing its response revealed it has introduced new guardrails for agents. </p><p>“Until this week, calling <em>volumeDelete</em> on the API ran the deletion immediately, with no way to undo it. Meanwhile, the dashboard had a 48-hour window for the same action,” the company noted. </p><p>“We’ve since updated the API to match; all deletes now soft delete for 48 hours. Instant undo, a primitive available everywhere in the product, exists now in the API.”</p><h2 id="a-glimpse-into-the-next-decade-of-identity-security">A glimpse into the “next decade of identity security”</h2><p>Security experts have warned the PocketOS incident is a stark reminder that placing too much faith in AI agents can have disastrous consequences. </p><p>As <a href="https://www.itpro.com/technology/artificial-intelligence/meta-engineer-trusted-advice-from-an-ai-agent-ended-up-exposing-user-data"><u><em>ITPro </em></u><u>reported</u></a> in March this year, Meta experienced a breach when a software engineer blindly followed advice given by an agent. </p><p>Notably, that incident occurred due to human error. The PocketOS situation, meanwhile, highlights the risk of giving agents the ability to make decisions autonomously. </p><p>Given how the debacle unfolded, Check Point’s Aaron Rose said this gives enterprises a glimpse into the “next decade of identity security”.</p><p>“An AI agent operating in your production infrastructure is not a tool, and it is not a service account,” he said. </p><p>“It is a new kind of identity, one that thinks rather than executes, and one that requires its own discrete account, its own least privileged entitlements, its own behavioural baseline, and its own real-time audit trail.”</p><p>Rose added that the capabilities of AI agents are advancing “faster than the security architecture around them”. While the PocketOS incident is a high-profile public example, he hinted that there are “many more incidents” such as these unfolding quietly in enterprises around the globe. </p><p>Recent studies do point to the growing risks associated with AI agents, particularly in terms of governance and safeguards. </p><p><a href="https://www.itpro.com/security/enterprises-are-adopting-agents-faster-than-they-can-secure-and-govern-them-experts-warn-its-a-disaster-waiting-to-happen"><u>As </u><u><em>ITPro </em></u><u>reported this week</u></a>, analysis from Ping Identity found that many businesses are adopting and deploying these autonomous bots faster than they can secure them. </p><p>The company noted that traditional <a href="https://www.itpro.com/security/how-to-implement-identity-and-access-management-iam-effectively-in-your-business">identity and access management (IAM) </a>capabilities are now struggling to contend with an influx of non-human identities (NHIs), creating huge gaps in both visibility and governance. </p><p>Darren Guccione, CEO and co-founder of Keeper Security, echoed Rose’s comments, noting that this should not be viewed as an “edge case or technical anomaly” </p><p>Indeed, it’s a “predictable outcome of how these systems are being deployed”. </p><p>“The explanation the agent produced afterwards is revealing. It did not fail silently or unpredictably. It articulated that it guessed, bypassed explicit rules and carried out an irreversible action without verification. That is not a model hallucination problem. It is an access control failure enabled by unconstrained autonomy,” he said.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Enterprises are adopting agents faster than they can secure and govern them ]]></title>
                                                                                                <dc:content><![CDATA[ <p>The use of <a href="https://www.itpro.com/business/microsoft-expects-1-3-billion-ai-agents-to-be-in-operation-by-2028-heres-how-it-plans-to-get-them-working-together">AI agents</a> is spiralling out of control, as they're rushed into deployment more quickly than organizations can govern them.</p><p>According to new <a href="https://www.pingidentity.com/en/lp/4380-ai-agents-trusted-digital-workers.html" target="_blank"><u>research</u></a> from Ping Identity, identity systems originally designed for human interaction are now being pushed to operate continuously. </p><p>The firm warned this is placing huge strain on existing models while creating gaps in governance, visibility, and accountability at the point when decisions are executed.</p><p>"Enterprises are deploying autonomous AI faster than they can govern it,” warned Andre Durand, CEO and founder of Ping Identity. “Identity remains foundational, but in an agentic environment it must operate continuously. Control must be enforced at the moment an action occurs.”</p><p>Researchers noted that agents are creating a new class of identity risk in environments where they operate autonomously across enterprise systems. </p><p>By combining individually legitimate permissions in unintended ways, they can generate actions that bypass controls and that can't be fully traced or governed. </p><p>Some of the biggest challenges include a lack of visibility when it comes to delegation, along with sub-agent spawning, where agent chains become untraceable and break auditability. </p><p>Meanwhile, AI agents are bypassing human decision-makers in ways that aren't planned for in OAuth and OIDC models. </p><p>Other issues include context leakage across systems where there's no continuous re-evaluation of authorization, and new questions around permission inheritance, liability, and enforcement in agent-to-agent interactions.</p><h2 id="iam-strategies-need-overhauled">IAM strategies need overhauled</h2><p>Despite these risks, most <a href="https://www.itpro.com/strategy/28935/what-is-identity-management-and-what-role-does-it-play-in-security-strategy">identity and access management (IAM)</a> approaches remain centered on human users and static access decisions, leaving organizations unprepared to govern autonomous systems.</p><p>“These trends reflect a broader shift in identity requirements,” said Martin Kuppinger, founder of KuppingerCole Analysts, which carried out the research. </p><p>“As autonomous agents become more prevalent, organizations will need to extend identity and authorization models to maintain control, accountability, and trust across increasingly dynamic environments.”</p><p>According to IBM’s <a href="https://www.ibm.com/reports/data-breach"><u>2025 Cost of a Data Breach</u></a> report, 13% of organizations have experienced AI-related security breaches, and 97% lack adequate access controls for AI systems.</p><p>The Ping Identity report echoes recent <a href="https://www.itpro.com/security/agent-identity-governance-cant-keeping-up-with-adoption-rates-and-its-creating-a-security-nightmare"><u>research</u></a> from SANS, which found that non‑human and <a href="https://www.itpro.com/security/agentic-ai-poses-major-challenge-for-security-professionals-says-palo-alto-networks-emea-ciso">AI identities</a> are multiplying faster than organizations can secure them. </p><p>More than three-quarters of organizations said they were seeing growth in the use of <a href="https://www.itpro.com/security/non-human-identities-are-we-sleepwalking-into-a-security-crisis"><u>non‑human identities (NHIs)</u></a> such as service accounts, API keys, automation bots, and workload identities - but that governance was failing to keep pace.</p><p>Of the three-quarters already using AI agents that require credentials, 5% of security leaders told the SANS researchers that they didn’t even know whether agentic AI was running in their environment or not.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/security/enterprises-are-adopting-agents-faster-than-they-can-secure-and-govern-them-experts-warn-its-a-disaster-waiting-to-happen</link>
                                                                            <description>
                            <![CDATA[ Identity systems developed for human interaction fail to cope with the new demands ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">ZNS99FFXfZRPCT59whwfcf</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/PbaraXgyBf5cecgbD4TJLU-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Thu, 30 Apr 2026 11:22:53 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:18 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Security]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                                    <dc:creator><![CDATA[ Emma Woollacott ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/aWfskavxoVSMDy6cDWtYmJ-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/PbaraXgyBf5cecgbD4TJLU-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[An illustration showing an AI agent side profile, depicted as a blue robot, with seven human faces in varying earthy metallic tones shown to the right.]]></media:description>                                                            <media:text><![CDATA[An illustration showing an AI agent side profile, depicted as a blue robot, with seven human faces in varying earthy metallic tones shown to the right.]]></media:text>
                                <media:title type="plain"><![CDATA[An illustration showing an AI agent side profile, depicted as a blue robot, with seven human faces in varying earthy metallic tones shown to the right.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/PbaraXgyBf5cecgbD4TJLU-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>The use of <a href="https://www.itpro.com/business/microsoft-expects-1-3-billion-ai-agents-to-be-in-operation-by-2028-heres-how-it-plans-to-get-them-working-together">AI agents</a> is spiralling out of control, as they're rushed into deployment more quickly than organizations can govern them.</p><p>According to new <a href="https://www.pingidentity.com/en/lp/4380-ai-agents-trusted-digital-workers.html" target="_blank"><u>research</u></a> from Ping Identity, identity systems originally designed for human interaction are now being pushed to operate continuously. </p><p>The firm warned this is placing huge strain on existing models while creating gaps in governance, visibility, and accountability at the point when decisions are executed.</p><p>"Enterprises are deploying autonomous AI faster than they can govern it,” warned Andre Durand, CEO and founder of Ping Identity. “Identity remains foundational, but in an agentic environment it must operate continuously. Control must be enforced at the moment an action occurs.”</p><p>Researchers noted that agents are creating a new class of identity risk in environments where they operate autonomously across enterprise systems. </p><p>By combining individually legitimate permissions in unintended ways, they can generate actions that bypass controls and that can't be fully traced or governed. </p><p>Some of the biggest challenges include a lack of visibility when it comes to delegation, along with sub-agent spawning, where agent chains become untraceable and break auditability. </p><p>Meanwhile, AI agents are bypassing human decision-makers in ways that aren't planned for in OAuth and OIDC models. </p><p>Other issues include context leakage across systems where there's no continuous re-evaluation of authorization, and new questions around permission inheritance, liability, and enforcement in agent-to-agent interactions.</p><h2 id="iam-strategies-need-overhauled">IAM strategies need overhauled</h2><p>Despite these risks, most <a href="https://www.itpro.com/strategy/28935/what-is-identity-management-and-what-role-does-it-play-in-security-strategy">identity and access management (IAM)</a> approaches remain centered on human users and static access decisions, leaving organizations unprepared to govern autonomous systems.</p><p>“These trends reflect a broader shift in identity requirements,” said Martin Kuppinger, founder of KuppingerCole Analysts, which carried out the research. </p><p>“As autonomous agents become more prevalent, organizations will need to extend identity and authorization models to maintain control, accountability, and trust across increasingly dynamic environments.”</p><p>According to IBM’s <a href="https://www.ibm.com/reports/data-breach"><u>2025 Cost of a Data Breach</u></a> report, 13% of organizations have experienced AI-related security breaches, and 97% lack adequate access controls for AI systems.</p><p>The Ping Identity report echoes recent <a href="https://www.itpro.com/security/agent-identity-governance-cant-keeping-up-with-adoption-rates-and-its-creating-a-security-nightmare"><u>research</u></a> from SANS, which found that non‑human and <a href="https://www.itpro.com/security/agentic-ai-poses-major-challenge-for-security-professionals-says-palo-alto-networks-emea-ciso">AI identities</a> are multiplying faster than organizations can secure them. </p><p>More than three-quarters of organizations said they were seeing growth in the use of <a href="https://www.itpro.com/security/non-human-identities-are-we-sleepwalking-into-a-security-crisis"><u>non‑human identities (NHIs)</u></a> such as service accounts, API keys, automation bots, and workload identities - but that governance was failing to keep pace.</p><p>Of the three-quarters already using AI agents that require credentials, 5% of security leaders told the SANS researchers that they didn’t even know whether agentic AI was running in their environment or not.</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ UK firms left in the dark over what workers are sharing with AI ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Enterprises across the UK are contending with “critical blind spots” over what workers are sharing with AI applications, according to new research. </p><p>A survey from SailPoint found more than two-thirds (67%) of organizations can’t account for the information staff are sharing with AI platforms and <a href="https://www.itpro.com/technology/artificial-intelligence/generative-ai-vs-large-language-models">large language models (LLMs)</a>. </p><p>Worse still, the study noted that 35% of respondents admitted to sharing data through external tools, rather than approved internal applications, which is creating an array of risks for enterprises. </p><p>The rise of ‘<a href="https://www.itpro.com/technology/artificial-intelligence/in-from-the-shadows-ais-uncontrolled-growth">shadow AI</a>’ has become a recurring pain point for organizations over the last two years. Workers using unauthorized applications risk exposing sensitive company data, research shows – and there’s no sign of the trend slowing down. </p><p><a href="https://www.itpro.com/technology/artificial-intelligence/gartner-says-40-percent-of-enterprises-will-experience-shadow-ai-breaches-by-2030-educating-staff-is-the-key-to-avoiding-disaster"><u>Research from Gartner</u></a> in November 2025 predicts that 40% of enterprises will suffer a data breach due to shadow AI by 2030. </p><p>SailPoint noted that the growing shadow AI trend comes in spite of the fact many enterprises are investing heavily in data management and AI capabilities for staff. </p><p>More than four-in-five respondents (82%) said they have invested in additional staff and skills training to help workers better manage AI applications, while 41% have brought on dedicated AI and analytics personnel. </p><p>Notably, nearly half (45%) of IT leaders said they still lack visibility on where information is being shared, and how.</p><h2 id="agentic-ai-poses-new-governance-challenges">Agentic AI poses new governance challenges</h2><p>Mark McClain, CEO and Founder at SailPoint, said the findings show AI can often represent a catch-22 for organizations. While these tools are helping staff, they’re now creating additional risk surfaces for security teams. </p><p>“<a href="https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today">AI tools</a> can enhance productivity, but they also create serious risk when they operate outside an organization’s visibility and governance,” he said. </p><p>“When sensitive information is entered into unapproved models, it can be exposed, mishandled, or even amplified through errors and hallucinations.”</p><p>McClain warned that with the rise of agentic AI, poor <a href="https://www.itpro.com/strategy/29269/what-is-data-management">data management</a> practices could be further amplified and put businesses at greater risk. </p><iframe allow="" height="200px" width="100%" id="" style="" class="position-center" data-lazy-priority="low" data-lazy-src="https://player.captivate.fm/episode/b6147684-d254-40d8-9409-5f76f5abd749/" src="https://player.captivate.fm/episode/b6147684-d254-40d8-9409-5f76f5abd749/"></iframe><p>SailPoint noted that the need for greater visibility and oversight is now a priority for many enterprises on account of growing risks. In a <a href="https://www.sailpoint.com/press-releases/sailpoint-ai-agent-adoption-report" target="_blank"><u>previous study from SailPoint</u></a>, four-in-five organizations (80%) revealed that AI agents had performed “unintended actions” such as accessing or sharing inappropriate data. </p><p>UK businesses are adding as many as 10,000 agents and machine identities each month, the company noted, meaning security teams could quickly become overwhelmed. </p><p>"As use of AI systems becomes more widespread, the situation is only going to get more out of control if organizations fail to put the right guardrails in place – compounded by other tools flying under the radar,” McClain commented. </p><p>“Organizations need to stop workarounds and regain control. That takes a combination of skills and awareness, but it also fundamentally boils down to a challenge around identity”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/security/data-protection/uk-firms-left-in-the-dark-over-what-workers-are-sharing-with-ai</link>
                                                                            <description>
                            <![CDATA[ Security teams can’t keep track of what workers are sharing with AI applications, regardless of whether they’re approved or unauthorized ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">Eo7ShySuZws95AM5KSTsqD</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/agHuoixY7A7CvZEbCApv4g-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Thu, 30 Apr 2026 07:30:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 13:11:33 +0000</updated>
                                                                                                                                            <category><![CDATA[Data Security]]></category>
                                                    <category><![CDATA[Data Protection]]></category>
                                                    <category><![CDATA[Cybersecurity]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/agHuoixY7A7CvZEbCApv4g-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Businesswoman holding handheld device while walking in an abstract data center environment made up of blocks with bright colored data lines flowing on ground.]]></media:description>                                                            <media:text><![CDATA[Businesswoman holding handheld device while walking in an abstract data center environment made up of blocks with bright colored data lines flowing on ground.]]></media:text>
                                <media:title type="plain"><![CDATA[Businesswoman holding handheld device while walking in an abstract data center environment made up of blocks with bright colored data lines flowing on ground.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/agHuoixY7A7CvZEbCApv4g-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Enterprises across the UK are contending with “critical blind spots” over what workers are sharing with AI applications, according to new research. </p><p>A survey from SailPoint found more than two-thirds (67%) of organizations can’t account for the information staff are sharing with AI platforms and <a href="https://www.itpro.com/technology/artificial-intelligence/generative-ai-vs-large-language-models">large language models (LLMs)</a>. </p><p>Worse still, the study noted that 35% of respondents admitted to sharing data through external tools, rather than approved internal applications, which is creating an array of risks for enterprises. </p><p>The rise of ‘<a href="https://www.itpro.com/technology/artificial-intelligence/in-from-the-shadows-ais-uncontrolled-growth">shadow AI</a>’ has become a recurring pain point for organizations over the last two years. Workers using unauthorized applications risk exposing sensitive company data, research shows – and there’s no sign of the trend slowing down. </p><p><a href="https://www.itpro.com/technology/artificial-intelligence/gartner-says-40-percent-of-enterprises-will-experience-shadow-ai-breaches-by-2030-educating-staff-is-the-key-to-avoiding-disaster"><u>Research from Gartner</u></a> in November 2025 predicts that 40% of enterprises will suffer a data breach due to shadow AI by 2030. </p><p>SailPoint noted that the growing shadow AI trend comes in spite of the fact many enterprises are investing heavily in data management and AI capabilities for staff. </p><p>More than four-in-five respondents (82%) said they have invested in additional staff and skills training to help workers better manage AI applications, while 41% have brought on dedicated AI and analytics personnel. </p><p>Notably, nearly half (45%) of IT leaders said they still lack visibility on where information is being shared, and how.</p><h2 id="agentic-ai-poses-new-governance-challenges">Agentic AI poses new governance challenges</h2><p>Mark McClain, CEO and Founder at SailPoint, said the findings show AI can often represent a catch-22 for organizations. While these tools are helping staff, they’re now creating additional risk surfaces for security teams. </p><p>“<a href="https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today">AI tools</a> can enhance productivity, but they also create serious risk when they operate outside an organization’s visibility and governance,” he said. </p><p>“When sensitive information is entered into unapproved models, it can be exposed, mishandled, or even amplified through errors and hallucinations.”</p><p>McClain warned that with the rise of agentic AI, poor <a href="https://www.itpro.com/strategy/29269/what-is-data-management">data management</a> practices could be further amplified and put businesses at greater risk. </p><iframe allow="" height="200px" width="100%" id="" style="" class="position-center" data-lazy-priority="low" data-lazy-src="https://player.captivate.fm/episode/b6147684-d254-40d8-9409-5f76f5abd749/" src="https://player.captivate.fm/episode/b6147684-d254-40d8-9409-5f76f5abd749/"></iframe><p>SailPoint noted that the need for greater visibility and oversight is now a priority for many enterprises on account of growing risks. In a <a href="https://www.sailpoint.com/press-releases/sailpoint-ai-agent-adoption-report" target="_blank"><u>previous study from SailPoint</u></a>, four-in-five organizations (80%) revealed that AI agents had performed “unintended actions” such as accessing or sharing inappropriate data. </p><p>UK businesses are adding as many as 10,000 agents and machine identities each month, the company noted, meaning security teams could quickly become overwhelmed. </p><p>"As use of AI systems becomes more widespread, the situation is only going to get more out of control if organizations fail to put the right guardrails in place – compounded by other tools flying under the radar,” McClain commented. </p><p>“Organizations need to stop workarounds and regain control. That takes a combination of skills and awareness, but it also fundamentally boils down to a challenge around identity”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Kaseya shifts from AI ‘insights’ to autonomous action with new agentic platform ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Kaseya has officially moved beyond AI advisory tools, unveiling its new "agentic" IT management platform. </p><p>The new platform, which is powered by a proprietary engine dubbed Kaseya Intelligence, was announced at the firm’s Connect event in Las Vegas. It is designed to break the cycle of "alert fatigue" by moving from AI that simply surfaces recommendations to a system that autonomously executes tasks across IT operations, security, and backup.</p><p>The company is positioning this as a fundamental architectural shift. It says that many vendors have spent the last year bolting AI features onto disconnected tools. This, Kaseya argues, is an approach that often backfires, providing inaccurate recommendations based on partial data.</p><p>"The industry doesn’t need another AI feature bolted onto a disconnected tool," said Rania Succar, CEO of Kaseya, in a statement. </p><p>"What MSPs and IT teams need is a platform that runs their operations, one that sees across every system, understands context, and acts autonomously. That’s what we’ve built. Kaseya Intelligence is the engine. The platform is the operating system. And the outcome is IT that manages itself."</p><h2 id="closing-the-loop-with-agentic-ai">Closing the loop with ‘agentic’ AI</h2><p>The platform’s core differentiator is its scale. Kaseya Intelligence is trained on a massive, purpose-built dataset including over 1 billion help desk tickets, 3 exabytes of backup data, and 17 million managed endpoints.</p><p>Unlike standard AI layers that hand off tasks to humans, Kaseya’s new agentic approach seeks to "close the loop." By executing actions, such as triaging a ticket or containing a threat, and then validating the outcome, the platform aims to function as an operating system for IT rather than a collection of utilities.</p><p>Dermot McCann, executive vice president and general manager for Kaseya in APAC and EMEA, talked about the need for this shift at the <a href="https://www.itpro.com/security/ai-and-data-are-reshaping-the-msp-landscape-but-hackers-are-getting-in-on-the-hot-ai-action">firm’s event in London</a> earlier this month. </p><p>"You don't have time to be messing around with stuff,” he said. “Everybody is under pressure. Everybody's busy. Our role is to make sure that the technologies that you pay us for work in the way that you need them to work."</p><h2 id="three-pillars-of-the-new-platform">Three pillars of the new platform</h2><p>The vision for this autonomous platform is being delivered through three major product launches unveiled today. First, the company introduced Agentic Digital Specialists, who are designed to handle high-volume, repetitive tasks. The first of these focuses on "Ticket Triage," a feature that automatically categorizes and routes incoming tickets, a process Kaseya claims can reduce downstream billing and routing errors by up to 80 percent.</p><p>The second pillar is Unified Cyber Resilience, a release that consolidates on-premises, SaaS, endpoint, and cloud backup into a single portal. By eliminating tool sprawl, the platform aims to provide AI-driven screenshot verification with 99.9 percent accuracy while offering expanded support for Azure and Hyper-V environments.</p><p>Rounding out the trio is Kaseya Security Information and Event Management (SIEM), which targets the complexity and high cost typically associated with traditional security operations. </p><p>This new SIEM correlates signals from more than 60 data sources to provide full-surface attack visibility. It is designed to be accessible to standard IT teams without requiring a dedicated staff of security engineers, featuring automated threat containment and 400-day out-of-the-box log retention to assist with compliance requirements.</p><h2 id="an-operating-system-not-a-tool">An operating system, not a tool</h2><p>The release signals a clear attempt by Kaseya to commoditize feature-based AI while claiming the high ground of autonomous operations.</p><p>"This is the difference between AI as a feature and AI as an operating system," the company stated in a press release detailing the launch. By embedding this intelligence across its entire portfolio, Kaseya aims to enable MSPs to scale their service delivery without the proportional need for human headcount.</p><p>"We notice that it wasn't architecturally possible to keep up with this emerging threat [using traditional tools]," McCann added. "We have to capture [innovation], then we have to put some structure around it, governance around it, and determine whether they fit in our roadmap. These types of technologies are really driven because our customers are saying, 'Can you do this for us?'"</p><p>With this launch, Kaseya is betting that the future of the MSP market won't be won by those who provide the best AI suggestions, but by those whose platforms can actually do the work.</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/kaseya-shifts-from-ai-insights-to-autonomous-action-with-new-agentic-platform</link>
                                                                            <description>
                            <![CDATA[ The company aims to evolve from its suite of management tools into an autonomous operating system for MSPs ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">gwpFWwWw5tk8yLwJSa8D6X</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/RLQuTA9eFtRCsYpxd2sba3-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Tue, 28 Apr 2026 18:25:30 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:14 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ itpro@futurenet.com (Rene Millman) ]]></author>                    <dc:creator><![CDATA[ Rene Millman ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/vwWuTPNRCuw9vEaWzuXYnR-320-70.png ]]></dc:source>
                                                                <dc:description><![CDATA[ null ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/RLQuTA9eFtRCsYpxd2sba3-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Kaseya logo and branding pictured on a smartphone screen with branding in background.]]></media:description>                                                            <media:text><![CDATA[Kaseya logo and branding pictured on a smartphone screen with branding in background.]]></media:text>
                                <media:title type="plain"><![CDATA[Kaseya logo and branding pictured on a smartphone screen with branding in background.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/RLQuTA9eFtRCsYpxd2sba3-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Kaseya has officially moved beyond AI advisory tools, unveiling its new "agentic" IT management platform. </p><p>The new platform, which is powered by a proprietary engine dubbed Kaseya Intelligence, was announced at the firm’s Connect event in Las Vegas. It is designed to break the cycle of "alert fatigue" by moving from AI that simply surfaces recommendations to a system that autonomously executes tasks across IT operations, security, and backup.</p><p>The company is positioning this as a fundamental architectural shift. It says that many vendors have spent the last year bolting AI features onto disconnected tools. This, Kaseya argues, is an approach that often backfires, providing inaccurate recommendations based on partial data.</p><p>"The industry doesn’t need another AI feature bolted onto a disconnected tool," said Rania Succar, CEO of Kaseya, in a statement. </p><p>"What MSPs and IT teams need is a platform that runs their operations, one that sees across every system, understands context, and acts autonomously. That’s what we’ve built. Kaseya Intelligence is the engine. The platform is the operating system. And the outcome is IT that manages itself."</p><h2 id="closing-the-loop-with-agentic-ai">Closing the loop with ‘agentic’ AI</h2><p>The platform’s core differentiator is its scale. Kaseya Intelligence is trained on a massive, purpose-built dataset including over 1 billion help desk tickets, 3 exabytes of backup data, and 17 million managed endpoints.</p><p>Unlike standard AI layers that hand off tasks to humans, Kaseya’s new agentic approach seeks to "close the loop." By executing actions, such as triaging a ticket or containing a threat, and then validating the outcome, the platform aims to function as an operating system for IT rather than a collection of utilities.</p><p>Dermot McCann, executive vice president and general manager for Kaseya in APAC and EMEA, talked about the need for this shift at the <a href="https://www.itpro.com/security/ai-and-data-are-reshaping-the-msp-landscape-but-hackers-are-getting-in-on-the-hot-ai-action">firm’s event in London</a> earlier this month. </p><p>"You don't have time to be messing around with stuff,” he said. “Everybody is under pressure. Everybody's busy. Our role is to make sure that the technologies that you pay us for work in the way that you need them to work."</p><h2 id="three-pillars-of-the-new-platform">Three pillars of the new platform</h2><p>The vision for this autonomous platform is being delivered through three major product launches unveiled today. First, the company introduced Agentic Digital Specialists, who are designed to handle high-volume, repetitive tasks. The first of these focuses on "Ticket Triage," a feature that automatically categorizes and routes incoming tickets, a process Kaseya claims can reduce downstream billing and routing errors by up to 80 percent.</p><p>The second pillar is Unified Cyber Resilience, a release that consolidates on-premises, SaaS, endpoint, and cloud backup into a single portal. By eliminating tool sprawl, the platform aims to provide AI-driven screenshot verification with 99.9 percent accuracy while offering expanded support for Azure and Hyper-V environments.</p><p>Rounding out the trio is Kaseya Security Information and Event Management (SIEM), which targets the complexity and high cost typically associated with traditional security operations. </p><p>This new SIEM correlates signals from more than 60 data sources to provide full-surface attack visibility. It is designed to be accessible to standard IT teams without requiring a dedicated staff of security engineers, featuring automated threat containment and 400-day out-of-the-box log retention to assist with compliance requirements.</p><h2 id="an-operating-system-not-a-tool">An operating system, not a tool</h2><p>The release signals a clear attempt by Kaseya to commoditize feature-based AI while claiming the high ground of autonomous operations.</p><p>"This is the difference between AI as a feature and AI as an operating system," the company stated in a press release detailing the launch. By embedding this intelligence across its entire portfolio, Kaseya aims to enable MSPs to scale their service delivery without the proportional need for human headcount.</p><p>"We notice that it wasn't architecturally possible to keep up with this emerging threat [using traditional tools]," McCann added. "We have to capture [innovation], then we have to put some structure around it, governance around it, and determine whether they fit in our roadmap. These types of technologies are really driven because our customers are saying, 'Can you do this for us?'"</p><p>With this launch, Kaseya is betting that the future of the MSP market won't be won by those who provide the best AI suggestions, but by those whose platforms can actually do the work.</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Everything you need to know about OpenAI's new workspace agents ]]></title>
                                                                                                <dc:content><![CDATA[ <p>OpenAI has unveiled new agentic enterprise AI tools called workspace agents that can be built in minutes to automate basic tasks – and be shared across an organization. </p><p>Powered by <a href="https://www.itpro.com/business/business-strategy/openais-codex-developer-agent-just-got-a-big-update">OpenAI's Codex</a>, workspace agents are taking over from GPTs as an automation tool for ChatGPT in businesses, promising to manage tasks such as checking customer feedback, building reports, and other repeatable bits of work that eat up workers' days. </p><p>This year was predicted to be the year agentic AI finally becomes useful to the enterprise, and the main players are stepping up with workplace-focused product releases. </p><p>Google this week <a href="https://www.reuters.com/business/google-puts-ai-agents-heart-its-enterprise-money-making-push-2026-04-22/"><u>unveiled AI agents</u></a> at its cloud conference under a wider set of enterprise products via Gemini Enterprise, while Anthropic has won attention for its <a href="https://www.itpro.com/technology/artificial-intelligence/everything-you-need-to-know-about-anthropic-claude-cowork"><u>Claude Cowork tool</u></a> – now set to be <a href="https://www.itpro.com/technology/artificial-intelligence/anthropics-claude-cowork-tool-is-coming-to-microsoft-copilot"><u>included in Microsoft Copilot</u></a>. </p><p>Here's what you need to know about OpenAI's workspace agents in <a href="https://www.itpro.com/technology/artificial-intelligence-ai/369965/what-is-chatgpt-and-what-does-it-mean-for-businesses">ChatGPT</a>. </p><h2 id="what-are-workspace-agents">What are workspace agents? </h2><p>Workspace agents let teams create a shared agent to complete tasks across an organization in ChatGPT or <a href="https://www.itpro.com/software/slack-is-now-the-key-to-salesforces-agentic-ai-plans">Slack</a>, all managed by permissions and controls and run by Codex in the cloud. </p><p>That means they can access files, code, tools and memory, the company said. </p><p>"Powered by Codex, they can take on many of the tasks people already do at work — from preparing reports, to writing code, to responding to messages," the company said in a blog post. "They run in the cloud, so they can keep working even when you’re not." </p><p>What does this mean for GPTs? OpenAI says that workspace agents are an "evolution" of GPTs, the personalized chatbots unveiled three years ago, and has said they will remain available in the meantime to ease the shift to workspace agents. </p><p>"Soon, we’ll make it easy to convert GPTs into workspace agents," the blogpost noted. </p><h2 id="what-can-workspace-agents-do">What can workspace agents do? </h2><p>The aim for OpenAI is to integrate them into workflows to pick up repeated tasks, and be able to access files and so on. They can work out of hours or on a schedule, or be set to deploy when a request comes in. </p><p>So far they work in ChatGPT and Slack, with further "surfaces," as OpenAI called them, coming soon. </p><p>"Agents do more than answer a prompt: they can write or run code, use connected apps, remember what they’ve learned, and continue work across multiple steps," OpenAI said. </p><p>As an example, OpenAI said its own staff had built a software reviewer that could read through software requests, check them against tools and policies, and file IT tickets. </p><p>Custom agents can also be used as a product feedback router that monitors Slack, support challenges, and public forums and pull data collected into a summary.  </p><p>Other examples include a report generator that pulls in data on a set day each week, creating a chart and writing the report before sharing with the team, as well as a lead outreach agent that researches inbound leads, drafts email follow-ups, and updates a <a href="https://www.itpro.com/customer-relationship-management-crm/34398/what-is-crm-documentation-and-how-can-it-benefit-your">CRM</a>. </p><h2 id="keeping-data-safe">Keeping data safe</h2><p>Workspace agents can be managed using policies and controls, dictating what actions, tools, and data it's allowed to access and when it needs to ask for human approval. </p><p>"For sensitive steps, like editing a spreadsheet, sending an email, or adding a calendar event, you can require the agent to ask for permission before moving forward," OpenAI said. </p><p>All usage is tracked via analytics, while a compliance API offers visibility for admins to see how agents are being built and used – and they can suspend them if needed, too. </p><h2 id="how-to-make-a-workspace-agent">How to make a workspace agent</h2><p>To start, click Agents in the sidebar and describe a workflow or task your team works on. ChatGPT will then guide the user through a step by step process to automate it via an agent, if possible. </p><p>"Describe the job you want done or just drop in a file," the company says. "ChatGPT helps turn it into an agent: defining the steps, connecting the right tools, adding skills, and testing it until it works the way you expect."</p><p>OpenAI has pre-set templates across finance, sales, marketing, and other areas in a similar vein to <a href="https://www.itpro.com/technology/artificial-intelligence/why-anthropic-sent-software-stocks-into-freefall">Claude Cowork’s industry-specific plugins</a>. </p><p>"Each comes with built-in skills and suggested tools, so you can quickly set up an agent and customize from there," the company said. </p><p>Because agents are shared, any improvements are immediately available to everyone using the same agent. </p><p>"Because agents have memory and can be guided and corrected in conversation, they get better as teams use them," the company said. "Over time, agents become a practical way to keep team knowledge current: build once, improve through use, then share or duplicate for new workflows."</p><p>Workspace agents are available as a research preview in Business, Enterprise, Edu and Teacher subscriptions for ChatGPT. They will be free until 6 May, when credit-based pricing will kick in. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/four-things-you-need-to-know-about-openais-new-workspace-agents-for-chatgpt-including-how-to-build-your-own</link>
                                                                            <description>
                            <![CDATA[ New ‘workspace agents’ from OpenAI will automate tasks for workers and can be customized for specific roles ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">jBpkwnj7EtzWyFzZHFjBeH</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/uj3zdTMMxN4rDq4n8QC4kb-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Fri, 24 Apr 2026 11:34:35 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:12 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                                    <dc:creator><![CDATA[ Nicole Kobie ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/8Y8JDDTQ7XDEk49FoAFP2S-320-70.png ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Nicole Kobie first started writing for ITPro in 2007. As a freelance journalist covering technology and business, Nicole&#039;s work includes  bylines in New Scientist, Wired, PC Pro and many more. &lt;/p&gt;&lt;p&gt;Nicole the author of a book about the history of technology, The Long History of the Future.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/uj3zdTMMxN4rDq4n8QC4kb-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Close-up image of OpenAI logo and branding in white coloring against a black background.]]></media:description>                                                            <media:text><![CDATA[Close-up image of OpenAI logo and branding in white coloring against a black background.]]></media:text>
                                <media:title type="plain"><![CDATA[Close-up image of OpenAI logo and branding in white coloring against a black background.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/uj3zdTMMxN4rDq4n8QC4kb-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>OpenAI has unveiled new agentic enterprise AI tools called workspace agents that can be built in minutes to automate basic tasks – and be shared across an organization. </p><p>Powered by <a href="https://www.itpro.com/business/business-strategy/openais-codex-developer-agent-just-got-a-big-update">OpenAI's Codex</a>, workspace agents are taking over from GPTs as an automation tool for ChatGPT in businesses, promising to manage tasks such as checking customer feedback, building reports, and other repeatable bits of work that eat up workers' days. </p><p>This year was predicted to be the year agentic AI finally becomes useful to the enterprise, and the main players are stepping up with workplace-focused product releases. </p><p>Google this week <a href="https://www.reuters.com/business/google-puts-ai-agents-heart-its-enterprise-money-making-push-2026-04-22/"><u>unveiled AI agents</u></a> at its cloud conference under a wider set of enterprise products via Gemini Enterprise, while Anthropic has won attention for its <a href="https://www.itpro.com/technology/artificial-intelligence/everything-you-need-to-know-about-anthropic-claude-cowork"><u>Claude Cowork tool</u></a> – now set to be <a href="https://www.itpro.com/technology/artificial-intelligence/anthropics-claude-cowork-tool-is-coming-to-microsoft-copilot"><u>included in Microsoft Copilot</u></a>. </p><p>Here's what you need to know about OpenAI's workspace agents in <a href="https://www.itpro.com/technology/artificial-intelligence-ai/369965/what-is-chatgpt-and-what-does-it-mean-for-businesses">ChatGPT</a>. </p><h2 id="what-are-workspace-agents">What are workspace agents? </h2><p>Workspace agents let teams create a shared agent to complete tasks across an organization in ChatGPT or <a href="https://www.itpro.com/software/slack-is-now-the-key-to-salesforces-agentic-ai-plans">Slack</a>, all managed by permissions and controls and run by Codex in the cloud. </p><p>That means they can access files, code, tools and memory, the company said. </p><p>"Powered by Codex, they can take on many of the tasks people already do at work — from preparing reports, to writing code, to responding to messages," the company said in a blog post. "They run in the cloud, so they can keep working even when you’re not." </p><p>What does this mean for GPTs? OpenAI says that workspace agents are an "evolution" of GPTs, the personalized chatbots unveiled three years ago, and has said they will remain available in the meantime to ease the shift to workspace agents. </p><p>"Soon, we’ll make it easy to convert GPTs into workspace agents," the blogpost noted. </p><h2 id="what-can-workspace-agents-do">What can workspace agents do? </h2><p>The aim for OpenAI is to integrate them into workflows to pick up repeated tasks, and be able to access files and so on. They can work out of hours or on a schedule, or be set to deploy when a request comes in. </p><p>So far they work in ChatGPT and Slack, with further "surfaces," as OpenAI called them, coming soon. </p><p>"Agents do more than answer a prompt: they can write or run code, use connected apps, remember what they’ve learned, and continue work across multiple steps," OpenAI said. </p><p>As an example, OpenAI said its own staff had built a software reviewer that could read through software requests, check them against tools and policies, and file IT tickets. </p><p>Custom agents can also be used as a product feedback router that monitors Slack, support challenges, and public forums and pull data collected into a summary.  </p><p>Other examples include a report generator that pulls in data on a set day each week, creating a chart and writing the report before sharing with the team, as well as a lead outreach agent that researches inbound leads, drafts email follow-ups, and updates a <a href="https://www.itpro.com/customer-relationship-management-crm/34398/what-is-crm-documentation-and-how-can-it-benefit-your">CRM</a>. </p><h2 id="keeping-data-safe">Keeping data safe</h2><p>Workspace agents can be managed using policies and controls, dictating what actions, tools, and data it's allowed to access and when it needs to ask for human approval. </p><p>"For sensitive steps, like editing a spreadsheet, sending an email, or adding a calendar event, you can require the agent to ask for permission before moving forward," OpenAI said. </p><p>All usage is tracked via analytics, while a compliance API offers visibility for admins to see how agents are being built and used – and they can suspend them if needed, too. </p><h2 id="how-to-make-a-workspace-agent">How to make a workspace agent</h2><p>To start, click Agents in the sidebar and describe a workflow or task your team works on. ChatGPT will then guide the user through a step by step process to automate it via an agent, if possible. </p><p>"Describe the job you want done or just drop in a file," the company says. "ChatGPT helps turn it into an agent: defining the steps, connecting the right tools, adding skills, and testing it until it works the way you expect."</p><p>OpenAI has pre-set templates across finance, sales, marketing, and other areas in a similar vein to <a href="https://www.itpro.com/technology/artificial-intelligence/why-anthropic-sent-software-stocks-into-freefall">Claude Cowork’s industry-specific plugins</a>. </p><p>"Each comes with built-in skills and suggested tools, so you can quickly set up an agent and customize from there," the company said. </p><p>Because agents are shared, any improvements are immediately available to everyone using the same agent. </p><p>"Because agents have memory and can be guided and corrected in conversation, they get better as teams use them," the company said. "Over time, agents become a practical way to keep team knowledge current: build once, improve through use, then share or duplicate for new workflows."</p><p>Workspace agents are available as a research preview in Business, Enterprise, Edu and Teacher subscriptions for ChatGPT. They will be free until 6 May, when credit-based pricing will kick in. </p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Google Cloud Next 2026: Scaling AI agents ]]></title>
                                                                                                <dc:content><![CDATA[ <iframe allow="" height="200px" width="100%" id="" style="" class="position-center" data-lazy-priority="low" data-lazy-src="https://player.captivate.fm/episode/2bc1ea99-d457-4ff2-8450-19769a6537bd/" src="https://player.captivate.fm/episode/2bc1ea99-d457-4ff2-8450-19769a6537bd/"></iframe><p>Las Vegas may be known as the city of sin, but in the world of tech it’s also the land of conferences. Taking over the Mandalay Bay resort this week was Google Cloud Next, Alphabet’s chance to show off the latest in its cloud strategy and – naturally – AI tools.</p><p>AI agents, in particular, have been a focus this year as Google Cloud looks to meet surging customer demand with infrastructure and software innovations for AI inference.</p><p>What story has Google been looking to tell at the event? And what’s the reality behind it?</p><p>In this episode, Jane speaks to Rory about some of the key talking points at Google Cloud Next 2026.</p><h2 id="highlights-2">Highlights</h2><p>"I think the big thing that has been very positive from customers is Google Cloud keeps reiterating that it understands that people are in a multi-cloud world, that no one wants to actually go all in and locked in on one cloud provider, and also seemingly recognizes that no one's looking to kind of abandon everything that they already pay for and to go fully in house with Google."</p><p>"So in a pod of 9600 of these 8t TPUs, they can deliver 121 exaflops of FP4 compute – I'm stressing that because it's not, you know, we're not talking about 64 bit top500 stuff here, but we are talking about FP4  low resolution data processing – which is ideal, actually, for training workloads. So that's that's significant."</p><p>"It feels like DeepMind kind of has a seat at every table. A lot of the announcements that we hear, like in the in the security space, which we didn't have time for, we're being told that they're being developed with DeepMind in the room."</p><h2 id="footnotes">Footnotes</h2><ul><li><a href="https://www.itpro.com/security/the-goal-for-this-year-will-be-to-automate-all-security-processes-google-cloud-is-betting-on-wiz-to-usher-in-a-new-era-of-ai-security" target="_blank">'The goal for this year will be to automate all security processes': Google Cloud is betting on Wiz to usher in a new era of AI security</a></li><li><a href="https://www.itpro.com/technology/artificial-intelligence/google-expands-gemini-enterprise-consolidates-vertex-ai-services-to-simplify-agent-deployment" target="_blank">Google expands Gemini Enterprise, consolidates Vertex AI services to simplify agent deployment</a></li><li><a href="https://www.itpro.com/infrastructure/google-cloud-eighth-generation-tpu-8t-8i-ai-inference-training" target="_blank">Google Cloud announces eighth-generation TPUs, boasting AI training and inference leaps</a></li><li><a href="https://www.itpro.com/cloud/live/google-cloud-next-2026-all-the-live-updates-as-they-happen" target="_blank">Google Cloud Next 2026: all the live updates as they happen</a></li><li><a href="https://www.itpro.com/cloud/cloud-computing/google-cloud-next-2026-googles-unique-advantages" target="_blank">Google Cloud Next 2026 is a chance to demonstrate Google’s unique advantages</a></li></ul><h2 id="subscribe">Subscribe </h2><ul><li><a href="https://apple.sjv.io/c/221109/473657/7613?subId1=itpro-gb-1243831151189624600&amp;sharedId=itpro-gb&amp;u=https%3A%2F%2Fpodcasts.apple.com%2Fgb%2Fpodcast%2Fthe-itpro-podcast%2Fid1483810154" target="_blank"><u>Subscribe to The IT Pro Podcast on Apple Podcasts</u></a></li><li><a href="https://open.spotify.com/show/7HpYehTy752KmtbwpOAgRZ" target="_blank"><u>Subscribe to The IT Pro Podcast on Spotify</u></a></li><li><a href="https://www.itpro.co.uk/newsletter-signup"><u>Subscribe to the IT Pro newsletter</u></a></li><li><a href="https://uk.linkedin.com/company/itpro-uk" target="_blank"><u>Join us on LinkedIn</u></a></li></ul> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/google-cloud-next-2026-scaling-ai-agents</link>
                                                                            <description>
                            <![CDATA[ The hyperscaler is going all in on full-stack AI deployment, underpinned by in-house innovations such as TPUs ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">sxfjtnBcqu7rpZ48Ma3wvk</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/koLg3xue643uR7KXCCnnYQ-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Fri, 24 Apr 2026 06:45:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:28 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ rory.bathgate@futurenet.com (Rory Bathgate) ]]></author>                    <dc:creator><![CDATA[ Rory Bathgate ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Vep5JogbPhduK7R6CUWAm6-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Rory Bathgate was the Features and Multimedia Editor at ITPro until 2026, overseeing all in-depth content and case studies. Throughout his time at ITPro, Rory  charted the rise in popularity of generative AI and specifically companies such as Microsoft, OpenAI, and Google. &lt;/p&gt;&lt;p&gt;Alongside this, he delved into increasing calls for ethical and responsible AI as global legislators circle the technology, as well as the latest in mobile networking technology, from 5G mmWave to the 3G sunset and how it will affect businesses.&lt;/p&gt;&lt;p&gt;He  provided coverage from high-profile tech conferences such as Dell Technologies World, SuiteWorld, and VMware Explore Europe. His on-the-ground coverage  included live blogs, extensive daily coverage of the most significant announcements, analysis pieces, and podcast.&lt;/p&gt;&lt;p&gt;Rory was the co-host of the ITPro Podcast alongside Jane McCallion, where he swapped a keyboard for a microphone to discuss the latest learnings in tech.&lt;/p&gt;&lt;p&gt;Rory has an MA in Eighteenth-Century Studies from King’s College London, as well as a BA in English and American Literature from the University of Kent. He joined ITPro in 2022 as a graduate, after four years in student journalism.&lt;/p&gt;&lt;p&gt;In his free time, Rory enjoys photography and video editing, and can often be found at the cinema or reading a good science fiction paperback.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/koLg3xue643uR7KXCCnnYQ-1920-80.jpg">
                                                            <media:credit><![CDATA[Future]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[The words &quot;Google Cloud Next 2026: Scaling AI agents&quot;, on a photo of the event. The words &quot;securing AI agents&quot; are yellow, the rest are white, The ITPro Podcast logo is in the bottom right corner.]]></media:description>                                                            <media:text><![CDATA[The words &quot;Google Cloud Next 2026: Scaling AI agents&quot;, on a photo of the event. The words &quot;securing AI agents&quot; are yellow, the rest are white, The ITPro Podcast logo is in the bottom right corner.]]></media:text>
                                <media:title type="plain"><![CDATA[The words &quot;Google Cloud Next 2026: Scaling AI agents&quot;, on a photo of the event. The words &quot;securing AI agents&quot; are yellow, the rest are white, The ITPro Podcast logo is in the bottom right corner.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/koLg3xue643uR7KXCCnnYQ-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <iframe allow="" height="200px" width="100%" id="" style="" class="position-center" data-lazy-priority="low" data-lazy-src="https://player.captivate.fm/episode/2bc1ea99-d457-4ff2-8450-19769a6537bd/" src="https://player.captivate.fm/episode/2bc1ea99-d457-4ff2-8450-19769a6537bd/"></iframe><p>Las Vegas may be known as the city of sin, but in the world of tech it’s also the land of conferences. Taking over the Mandalay Bay resort this week was Google Cloud Next, Alphabet’s chance to show off the latest in its cloud strategy and – naturally – AI tools.</p><p>AI agents, in particular, have been a focus this year as Google Cloud looks to meet surging customer demand with infrastructure and software innovations for AI inference.</p><p>What story has Google been looking to tell at the event? And what’s the reality behind it?</p><p>In this episode, Jane speaks to Rory about some of the key talking points at Google Cloud Next 2026.</p><h2 id="highlights-2">Highlights</h2><p>"I think the big thing that has been very positive from customers is Google Cloud keeps reiterating that it understands that people are in a multi-cloud world, that no one wants to actually go all in and locked in on one cloud provider, and also seemingly recognizes that no one's looking to kind of abandon everything that they already pay for and to go fully in house with Google."</p><p>"So in a pod of 9600 of these 8t TPUs, they can deliver 121 exaflops of FP4 compute – I'm stressing that because it's not, you know, we're not talking about 64 bit top500 stuff here, but we are talking about FP4  low resolution data processing – which is ideal, actually, for training workloads. So that's that's significant."</p><p>"It feels like DeepMind kind of has a seat at every table. A lot of the announcements that we hear, like in the in the security space, which we didn't have time for, we're being told that they're being developed with DeepMind in the room."</p><h2 id="footnotes">Footnotes</h2><ul><li><a href="https://www.itpro.com/security/the-goal-for-this-year-will-be-to-automate-all-security-processes-google-cloud-is-betting-on-wiz-to-usher-in-a-new-era-of-ai-security" target="_blank">'The goal for this year will be to automate all security processes': Google Cloud is betting on Wiz to usher in a new era of AI security</a></li><li><a href="https://www.itpro.com/technology/artificial-intelligence/google-expands-gemini-enterprise-consolidates-vertex-ai-services-to-simplify-agent-deployment" target="_blank">Google expands Gemini Enterprise, consolidates Vertex AI services to simplify agent deployment</a></li><li><a href="https://www.itpro.com/infrastructure/google-cloud-eighth-generation-tpu-8t-8i-ai-inference-training" target="_blank">Google Cloud announces eighth-generation TPUs, boasting AI training and inference leaps</a></li><li><a href="https://www.itpro.com/cloud/live/google-cloud-next-2026-all-the-live-updates-as-they-happen" target="_blank">Google Cloud Next 2026: all the live updates as they happen</a></li><li><a href="https://www.itpro.com/cloud/cloud-computing/google-cloud-next-2026-googles-unique-advantages" target="_blank">Google Cloud Next 2026 is a chance to demonstrate Google’s unique advantages</a></li></ul><h2 id="subscribe">Subscribe </h2><ul><li><a href="https://apple.sjv.io/c/221109/473657/7613?subId1=itpro-gb-1243831151189624600&amp;sharedId=itpro-gb&amp;u=https%3A%2F%2Fpodcasts.apple.com%2Fgb%2Fpodcast%2Fthe-itpro-podcast%2Fid1483810154" target="_blank"><u>Subscribe to The IT Pro Podcast on Apple Podcasts</u></a></li><li><a href="https://open.spotify.com/show/7HpYehTy752KmtbwpOAgRZ" target="_blank"><u>Subscribe to The IT Pro Podcast on Spotify</u></a></li><li><a href="https://www.itpro.co.uk/newsletter-signup"><u>Subscribe to the IT Pro newsletter</u></a></li><li><a href="https://uk.linkedin.com/company/itpro-uk" target="_blank"><u>Join us on LinkedIn</u></a></li></ul>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ The first hurdle is the hardest in generative AI adoption – and businesses keep falling ]]></title>
                                                                                                <dc:content><![CDATA[ <p>AI adoption is ramping up at a rapid pace globally, of that there’s no doubt. But there’s more to successful adoption than just buying shiny new tools and finding yourself in a productivity paradise. </p><p>While that might seem like stating the obvious, it’s still a problem a significant portion of enterprises find themselves facing – and there’s little sign that it will improve any time soon. </p><p>Research from Amazon Web Services (AWS), released to coincide with its annual AWS Summit London this week, showed progress on the adoption front compared to last year. </p><p>The study, which focuses primarily on UK-based firms, has some positive takeaways, with British enterprises outpacing adoption compared to their European counterparts. </p><p>Clearcut gains with the technology are also being made, according to AWS, with workers now basking in the productivity and efficiency gains promised by it and other big tech providers. </p><p>Speaking during the keynote address at the summit this week, Alison Kay, AWS’ VP and managing director for UK&amp;I, waxed lyrical about the benefits of AI for customers and the company itself. </p><p>Developers are working at lightning speed delivering projects previously deemed insurmountable while customers such as Motorway are recording the same gains. </p><p>“We’re living through just an incredible moment in history, a breakthrough moment in technology advancement,” Kay told attendees. </p><p>“And what makes this moment so exciting is that the things that are being built, they almost defy belief. They almost feel like magic.”</p><p>AI might look and feel like magic, but IT leaders aren't living in a fantasy land – they're facing the harsh reality of how difficult AI implementation can be. They've got the trendy new tools and staff are loving them, yet they’re not seeing any significant progress on a business-wide level.</p><h2 id="ai-pilot-purgatory-persists">AI pilot purgatory persists</h2><p>AWS’ study found the actual use of AI in the enterprise is still very basic, with employees using the technology to summarize documents or answer basic questions using chatbots. </p><p>That’s so 2023, but at least they’re enjoying it. More than two-thirds (68%) of employees are reporting marked productivity and efficiency improvements. </p><p>AWS’ findings align closely with similar research from Accenture earlier this month, which pointed toward “localized” gains whereby individual workers are seeing gains, but the companies themselves are still struggling. </p><p>So what’s stopping enterprises from making the leap toward more “advanced” AI uses? The same old problems many businesses have with <a href="https://www.itpro.com/strategy/29899/three-reasons-why-digital-transformation-is-essential-for-business-growth">digital transformation</a> projects. </p><p>Skills and expertise, <a href="https://www.itpro.com/security/the-risks-of-legacy-technology-for-your-business">legacy systems</a>, and <a href="https://www.itpro.com/business/digital-transformation/disparate-data-silos-are-still-blocking-the-path-to-digital-transformation-success">outdated infrastructure</a> are stifling progress. </p><p>It’s been a common recurring theme with AI adoption over the last three and a half years and one that must have IT leaders bashing their heads off the wall over. </p><p><a href="https://www.itpro.com/technology/artificial-intelligence/uk-firms-are-stuck-in-an-ai-pilot-purgatory-heres-how-they-can-drive-success"><u>Research from Google Cloud</u></a> in August 2024 warned British firms in particular were stuck in a “pilot purgatory”. </p><p>At that point, only two-in-five firms had pushed generative AI projects to production and were facing a “substantial hurdle” in scaling the technology internally - an issue the aforementioned Accenture study found is still a key problem. </p><p>The situation is even worse with agentic AI, with a January study from Dynatrace showing that half of agent adoption schemes are wallowing in the proof of concept (PoC) stage. </p><p>Phil Le-Brun, executive in residence at AWS, told <em>ITPro </em>that a “lack of imagination” is often a key factor in this inability to scale AI adoption and reach more advanced uses. </p><iframe allow="" height="200px" width="100%" id="" style="" class="position-center" data-lazy-priority="low" data-lazy-src="https://player.captivate.fm/episode/7b2774d8-c5b2-4912-a6aa-6dc1788dea4d/" src="https://player.captivate.fm/episode/7b2774d8-c5b2-4912-a6aa-6dc1788dea4d/"></iframe><p>An increasing volume of research also points toward the frantic nature of the <a href="https://www.itpro.com/technology/artificial-intelligence-ai/369959/what-is-generative-ai">generative AI</a> boom as a key factor behind failed adoption efforts. </p><p><a href="https://www.itpro.com/business/business-strategy/enterprises-are-paralyzed-by-a-lack-of-understanding-with-ai-adoption-and-theres-one-key-factor-that-decides-success"><u>Research from Forrester</u></a> in early April suggested that a “lack of understanding” when it comes to AI, combined with the rabid appetite to invest in the technology, has resulted in a slap-dash approach to AI innovation. </p><p>Executives have been caught up in the hype, and all they can see are dollar signs and the marketing materials promising huge productivity improvements. </p><p>But Forrester noted those who take a more conservative, “deliberate” approach to AI adoption have recorded far better results – and a key factor here is focusing on creating value for customers rather than productivity and cost reductions. </p><p>Speaking to <em>ITPro</em>, Le-Brun touched on this customer-centric approach with AI, noting that some of the more agile companies it engages with, such as startups, are making great strides in this regard. </p><p>Simply put, keep an eye on the productivity and cost benefits, but sharpen your focus on actually delivering for customers. </p><p>“The application of AI to fundamentally rethink how you’re delivering a product or service to the customer would be an example of advanced AI,” he said. “That’s where we see startups doing some amazing things.”</p><p>“They’re willing to really dive into what it is the customer wants, and how can we apply this new technology to change, rather than for large enterprises - it’s much harder because they have a lot of established processes.”</p><h2 id="a-long-road-ahead">A long road ahead</h2><p>The current pace of progress in pushing from pilots and basic use-cases toward more advanced AI uses highlights the sorry state of affairs for many businesses - and there’s still a long road ahead. A very long one, if AWS findings are to be taken at face value. </p><p>Just 24% of firms have reached an advanced AI phase in 2026, according to AWS, marking just a single percentage increase compared to last year. If things continue at this pace of progress, it’ll be 2102 before all AI adopters in the UK reach an advanced stage. </p><p>As Le-Brun told <em>ITPro</em>, businesses will be in a good position “by the end of the century”. </p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/business/business-strategy/the-first-hurdle-is-the-hardest-in-generative-ai-adoption-and-businesses-keep-falling</link>
                                                                            <description>
                            <![CDATA[ AWS’ UK chief said AI advances “feel like magic” at its recent London summit, but many firms are facing the reality of sluggish gains ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">432hD26emDP4sw7nZX5jda</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/5U6eXUHt9vPgHT7z5pghjg-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Thu, 23 Apr 2026 17:30:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 15:00:19 +0000</updated>
                                                                                                                                            <category><![CDATA[Generative AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ ross.kelly@futurenet.com (Ross Kelly) ]]></author>                    <dc:creator><![CDATA[ Ross Kelly ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Y5vrV2V98Np6jHAGmAtCd3-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Ross Kelly is ITPro&#039;s News &amp;amp; Analysis Editor, with a keen interest in cyber security, business leadership and emerging technologies.&lt;/p&gt;
&lt;p&gt;He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.&amp;nbsp;&lt;/p&gt;
&lt;p&gt;In his spare time, Ross enjoys cycling, walking and is an avid reader of history and non-fiction.&lt;/p&gt;
&lt;p&gt;You can contact Ross at ross.kelly@futurenet.com or on &lt;a href=&quot;https://twitter.com/rosswritesetc&quot;&gt;Twitter&lt;/a&gt; and &lt;a href=&quot;https://www.linkedin.com/in/ross-kelly-18a54411a/&quot;&gt;LinkedIn&lt;/a&gt;.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/5U6eXUHt9vPgHT7z5pghjg-1920-80.jpg">
                                                            <media:credit><![CDATA[Getty Images]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Man jumping over a hurdle with three fallen hurdles behind him.]]></media:description>                                                            <media:text><![CDATA[Man jumping over a hurdle with three fallen hurdles behind him.]]></media:text>
                                <media:title type="plain"><![CDATA[Man jumping over a hurdle with three fallen hurdles behind him.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/5U6eXUHt9vPgHT7z5pghjg-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>AI adoption is ramping up at a rapid pace globally, of that there’s no doubt. But there’s more to successful adoption than just buying shiny new tools and finding yourself in a productivity paradise. </p><p>While that might seem like stating the obvious, it’s still a problem a significant portion of enterprises find themselves facing – and there’s little sign that it will improve any time soon. </p><p>Research from Amazon Web Services (AWS), released to coincide with its annual AWS Summit London this week, showed progress on the adoption front compared to last year. </p><p>The study, which focuses primarily on UK-based firms, has some positive takeaways, with British enterprises outpacing adoption compared to their European counterparts. </p><p>Clearcut gains with the technology are also being made, according to AWS, with workers now basking in the productivity and efficiency gains promised by it and other big tech providers. </p><p>Speaking during the keynote address at the summit this week, Alison Kay, AWS’ VP and managing director for UK&amp;I, waxed lyrical about the benefits of AI for customers and the company itself. </p><p>Developers are working at lightning speed delivering projects previously deemed insurmountable while customers such as Motorway are recording the same gains. </p><p>“We’re living through just an incredible moment in history, a breakthrough moment in technology advancement,” Kay told attendees. </p><p>“And what makes this moment so exciting is that the things that are being built, they almost defy belief. They almost feel like magic.”</p><p>AI might look and feel like magic, but IT leaders aren't living in a fantasy land – they're facing the harsh reality of how difficult AI implementation can be. They've got the trendy new tools and staff are loving them, yet they’re not seeing any significant progress on a business-wide level.</p><h2 id="ai-pilot-purgatory-persists">AI pilot purgatory persists</h2><p>AWS’ study found the actual use of AI in the enterprise is still very basic, with employees using the technology to summarize documents or answer basic questions using chatbots. </p><p>That’s so 2023, but at least they’re enjoying it. More than two-thirds (68%) of employees are reporting marked productivity and efficiency improvements. </p><p>AWS’ findings align closely with similar research from Accenture earlier this month, which pointed toward “localized” gains whereby individual workers are seeing gains, but the companies themselves are still struggling. </p><p>So what’s stopping enterprises from making the leap toward more “advanced” AI uses? The same old problems many businesses have with <a href="https://www.itpro.com/strategy/29899/three-reasons-why-digital-transformation-is-essential-for-business-growth">digital transformation</a> projects. </p><p>Skills and expertise, <a href="https://www.itpro.com/security/the-risks-of-legacy-technology-for-your-business">legacy systems</a>, and <a href="https://www.itpro.com/business/digital-transformation/disparate-data-silos-are-still-blocking-the-path-to-digital-transformation-success">outdated infrastructure</a> are stifling progress. </p><p>It’s been a common recurring theme with AI adoption over the last three and a half years and one that must have IT leaders bashing their heads off the wall over. </p><p><a href="https://www.itpro.com/technology/artificial-intelligence/uk-firms-are-stuck-in-an-ai-pilot-purgatory-heres-how-they-can-drive-success"><u>Research from Google Cloud</u></a> in August 2024 warned British firms in particular were stuck in a “pilot purgatory”. </p><p>At that point, only two-in-five firms had pushed generative AI projects to production and were facing a “substantial hurdle” in scaling the technology internally - an issue the aforementioned Accenture study found is still a key problem. </p><p>The situation is even worse with agentic AI, with a January study from Dynatrace showing that half of agent adoption schemes are wallowing in the proof of concept (PoC) stage. </p><p>Phil Le-Brun, executive in residence at AWS, told <em>ITPro </em>that a “lack of imagination” is often a key factor in this inability to scale AI adoption and reach more advanced uses. </p><iframe allow="" height="200px" width="100%" id="" style="" class="position-center" data-lazy-priority="low" data-lazy-src="https://player.captivate.fm/episode/7b2774d8-c5b2-4912-a6aa-6dc1788dea4d/" src="https://player.captivate.fm/episode/7b2774d8-c5b2-4912-a6aa-6dc1788dea4d/"></iframe><p>An increasing volume of research also points toward the frantic nature of the <a href="https://www.itpro.com/technology/artificial-intelligence-ai/369959/what-is-generative-ai">generative AI</a> boom as a key factor behind failed adoption efforts. </p><p><a href="https://www.itpro.com/business/business-strategy/enterprises-are-paralyzed-by-a-lack-of-understanding-with-ai-adoption-and-theres-one-key-factor-that-decides-success"><u>Research from Forrester</u></a> in early April suggested that a “lack of understanding” when it comes to AI, combined with the rabid appetite to invest in the technology, has resulted in a slap-dash approach to AI innovation. </p><p>Executives have been caught up in the hype, and all they can see are dollar signs and the marketing materials promising huge productivity improvements. </p><p>But Forrester noted those who take a more conservative, “deliberate” approach to AI adoption have recorded far better results – and a key factor here is focusing on creating value for customers rather than productivity and cost reductions. </p><p>Speaking to <em>ITPro</em>, Le-Brun touched on this customer-centric approach with AI, noting that some of the more agile companies it engages with, such as startups, are making great strides in this regard. </p><p>Simply put, keep an eye on the productivity and cost benefits, but sharpen your focus on actually delivering for customers. </p><p>“The application of AI to fundamentally rethink how you’re delivering a product or service to the customer would be an example of advanced AI,” he said. “That’s where we see startups doing some amazing things.”</p><p>“They’re willing to really dive into what it is the customer wants, and how can we apply this new technology to change, rather than for large enterprises - it’s much harder because they have a lot of established processes.”</p><h2 id="a-long-road-ahead">A long road ahead</h2><p>The current pace of progress in pushing from pilots and basic use-cases toward more advanced AI uses highlights the sorry state of affairs for many businesses - and there’s still a long road ahead. A very long one, if AWS findings are to be taken at face value. </p><p>Just 24% of firms have reached an advanced AI phase in 2026, according to AWS, marking just a single percentage increase compared to last year. If things continue at this pace of progress, it’ll be 2102 before all AI adopters in the UK reach an advanced stage. </p><p>As Le-Brun told <em>ITPro</em>, businesses will be in a good position “by the end of the century”. </p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Google Cloud leans on Wiz to usher in a new era of AI security ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Google Cloud has announced a slew of new AI-powered security features, including new AI agents for continuous security developed by its new subsidiary Wiz.</p><p>Following <a href="https://www.itpro.com/business/business-strategy/google-confirms-wiz-acquisition-in-record-breaking-usd32-billion-deal">Google Cloud’s record $32 billion acquisition</a> of the Israeli-American cybersecurity startup, the hyperscaler is looking to deploy Wiz agents at a scale that was previously impossible.</p><p>To date, Wiz has made the detection of what it calls ‘toxic combinations’ in cloud environments – including misconfigurations, excessive permissions, and overlooked vulnerabilities – its bread and butter. </p><p>Google Cloud wants to apply this specialism to its entire platform, as well as those of partners.</p><p>While AI can be used to bolster defenses, improperly configured LLMs and AI agents can also complicate <a href="https://www.itpro.com/security/28133/what-is-cyber-security">cybersecurity </a>strategies and introduce new risks of their own, such as <a href="https://www.itpro.com/security/ncsc-issues-urgent-warning-over-growing-ai-prompt-injection-risks-heres-what-you-need-to-know"><u>prompt injection attacks</u></a>.</p><p>Google Cloud and Wiz aim to identify these risks in real-time by embedding security into AI as it’s deployed rather than bolting it on after the fact. At Google Cloud Next 2026, both firms have led with a joint message of a deep-set need for AI preparedness.</p><p>“One of the things we do is provide the right protections associated with rolling out your AI infrastructure that you can then make available,” said Francis deSouza, COO and president, Security Products at Google Cloud.</p><p>“Because [it is] not possible to retrofit security into an IT environment, you need to design it in from the ground up.”</p><p>For example, Wiz has launched AI agents that act as automated security researchers and engineers: Red Agent, Blue Agent, and Green Agent.</p><p>“Red team is typically contesting from the outside, and the Red Agent actually leverages the deep visibility that we get into cloud environments and identifies all of the potential exposures, all of the APIs that are internet exposed,”<em> </em>explained Yinon Costica, VP Product and co-founder of Wiz.</p><p>“The Green Agent actually takes the risk that the, let's say the red agent found, and then it automates the triage process.”</p><p>Costica explained that this speeds up a process that normally takes days or even weeks of manual security work. Finally, the Blue Agent takes indications of compromise and automates the investigation process into these threats.</p><p>He added that the three agents can collaborate and work with one another to improve defenses. For example, Red Agent can invoke Green Agent after identifying a vulnerability, or Blue Agent can attempt to flag intrusions by Red Agent to test the effectiveness of an organization’s detection capabilities.</p><p>All of the agents are managed via Wiz AI Application Protection Platform (AI-APP), an umbrella offering that secures every enterprise AI layer including agents, models, infrastructure, data, and access permissions. </p><p>Reflecting on the increasing pace of AI-powered attacks, Costica argued that firms must begin to apply agents to “start using AI against ourselves as much as possible” to identify weaknesses and fix them before attackers find them.</p><p>“The goal for this year will be to automate all security processes we know, because that's the only way to go. </p><p>“So there are human automations that we've been using in industry, but I think what agents allow us to do is get to the next level of acceleration and automation of security work.”</p><p>This sentiment was echoed by deSouza in the <a href="https://www.itpro.com/cloud/live/google-cloud-next-2026-all-the-live-updates-as-they-happen"><u>event’s opening keynote</u></a>, in which he warned that hackers have reduced the time between initial access and handoff to secondary threat groups from eight hours to 22 seconds.</p><h2 id="secure-interoperability">Secure interoperability</h2><p>A core part of the updated security offering at Google Cloud is its interoperability. The hyperscaler has partnered with firms such as Darktrace, Gigamon, and SAP to extend its out of the box security features to their platforms, and also supports end-to-end AI application protection on platforms like AWS, Microsoft Azure, and Oracle Cloud.</p><p>Customers can even extend security functions to SaaS tools such as OpenAI and custom-hosted cloud environments.</p><p>“It is a <a href="https://www.itpro.com/cloud/34476/what-is-multi-cloud">multi-cloud</a> world, and we talk to our customers about that all the time saying every company is a multi-cloud company, and will need to continue to be a multi-cloud company,” said deSouza,</p><p>In addition to its GCP-native functionality, Wiz supports Databricks and agent platforms including AWS <a href="https://www.itpro.com/technology/artificial-intelligence/amazons-go-build-it-ai-strategy-is-a-perfect-fit-for-openais-big-enterprise-push"><u>Agentcore</u></a>, Gemini Enterprise Agent Platform, Microsoft Azure Copilot Studio, and Salesforce <a href="https://www.itpro.com/business/business-strategy/salesforce-announces-huge-partner-program-revamp-with-agentforce-360-launch"><u>Agentforce</u></a>. </p><p>Customers can also extend its security features to external security ecosystems such as Cloudflare AI Security for Apps, Google Cloud Apigee, and <a href="https://www.itpro.com/security/data-breaches/everything-we-know-about-the-vercel-data-breach-so-far"><u>Vercel</u></a>.</p><h2 id="expanding-security-agents">Expanding security agents</h2><p>Alongside the Wiz-developed tools, Google Cloud also highlighted its AI agents for SecOps. These include <a href="https://www.itpro.com/security/google-just-launched-a-new-gemini-powered-dark-web-monitoring-service"><u>Dark Web Intelligence</u></a>, an agent announced at RSAC Conference 2026 that produces organization-specific threat profiles based on Google Threat Intelligence Group’s <a href="https://www.itpro.com/security/32117/what-is-the-dark-web"><u>dark web</u></a> monitoring. </p><p>DeSouza noted that Dark Web Intelligence can identify threats with 98% accuracy, a step above what was previously capable within the security industry.</p><p>“Coming up with new rules is a complex process, it can be done by humans but you have new intelligence coming out all the time,” said deSouza, adding that many organizations have thousands of security rules already in place and that adding new ones onto this list can be an overwhelming task.</p><p>Additionally, cybersecurity teams can use the Threat Hunting Agent and Detection Engineering Agent to search for active threats and novel attacks, identify gaps in their organization’s cybersecurity, and create automated detection rules based on current threats.</p><p>In total, Google Cloud’s Triage and Investigation agent processed more than five million alerts in 2025. The firm said the tool helps teams to complete cybersecurity analysis that previously took 30 minutes in just 60 seconds, using Gemini for reasoning and text generation.</p><p>Another new platform, Google Cloud Fraud Defense, is intended to build upon reCAPTCHA services to assess the legitimacy of humans, bots, and agents within business processes and commerce.</p><p>The announcements made at Google Cloud Next 2026 represent not only the sum of Google Cloud’s teams but also the innovations of Google DeepMind and the proprietary technologies Wiz relied on prior to its acquisition.</p><p><em>ITPro </em>understands that Wiz uses a mix of backend models, which it chooses depending on customer use cases. But deSouza said that going forward both Google Cloud and Wiz teams would benefit from sharing notes with internal frontier model developers.</p><p>“We have an advantage here,” said deSouza. </p><p>“Our defense technologies like Wiz work with our frontier model teams, and we know in advance what the model is going to have. And so on day one, we're taking advantage of the cutting edge of defense to protect our customers.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/security/the-goal-for-this-year-will-be-to-automate-all-security-processes-google-cloud-is-betting-on-wiz-to-usher-in-a-new-era-of-ai-security</link>
                                                                            <description>
                            <![CDATA[ Wiz wants to deploy its agents for continuous penetration testing, and in Google it’s found a parent company that can achieve this vision at scale ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">ctMeUtSErHFYBMQui96CMY</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/N3B3FC3PLDnt7g9MWxz368-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Thu, 23 Apr 2026 09:24:25 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 13:03:59 +0000</updated>
                                                                                                                                            <category><![CDATA[Security Operations]]></category>
                                                    <category><![CDATA[Security]]></category>
                                                    <category><![CDATA[Cyber Resilience]]></category>
                                                                                                <author><![CDATA[ rory.bathgate@futurenet.com (Rory Bathgate) ]]></author>                    <dc:creator><![CDATA[ Rory Bathgate ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Vep5JogbPhduK7R6CUWAm6-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Rory Bathgate was the Features and Multimedia Editor at ITPro until 2026, overseeing all in-depth content and case studies. Throughout his time at ITPro, Rory  charted the rise in popularity of generative AI and specifically companies such as Microsoft, OpenAI, and Google. &lt;/p&gt;&lt;p&gt;Alongside this, he delved into increasing calls for ethical and responsible AI as global legislators circle the technology, as well as the latest in mobile networking technology, from 5G mmWave to the 3G sunset and how it will affect businesses.&lt;/p&gt;&lt;p&gt;He  provided coverage from high-profile tech conferences such as Dell Technologies World, SuiteWorld, and VMware Explore Europe. His on-the-ground coverage  included live blogs, extensive daily coverage of the most significant announcements, analysis pieces, and podcast.&lt;/p&gt;&lt;p&gt;Rory was the co-host of the ITPro Podcast alongside Jane McCallion, where he swapped a keyboard for a microphone to discuss the latest learnings in tech.&lt;/p&gt;&lt;p&gt;Rory has an MA in Eighteenth-Century Studies from King’s College London, as well as a BA in English and American Literature from the University of Kent. He joined ITPro in 2022 as a graduate, after four years in student journalism.&lt;/p&gt;&lt;p&gt;In his free time, Rory enjoys photography and video editing, and can often be found at the cinema or reading a good science fiction paperback.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/N3B3FC3PLDnt7g9MWxz368-1920-80.jpg">
                                                            <media:credit><![CDATA[ITPro/Rory Bathgate]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[Francis deSouza, COO and president of security products at Google Cloud, pictured speaking on stage at Google Cloud Next 2026 in Las Vegas. ]]></media:description>                                                            <media:text><![CDATA[Francis deSouza, COO and president of security products at Google Cloud, pictured speaking on stage at Google Cloud Next 2026 in Las Vegas. ]]></media:text>
                                <media:title type="plain"><![CDATA[Francis deSouza, COO and president of security products at Google Cloud, pictured speaking on stage at Google Cloud Next 2026 in Las Vegas. ]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/N3B3FC3PLDnt7g9MWxz368-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Google Cloud has announced a slew of new AI-powered security features, including new AI agents for continuous security developed by its new subsidiary Wiz.</p><p>Following <a href="https://www.itpro.com/business/business-strategy/google-confirms-wiz-acquisition-in-record-breaking-usd32-billion-deal">Google Cloud’s record $32 billion acquisition</a> of the Israeli-American cybersecurity startup, the hyperscaler is looking to deploy Wiz agents at a scale that was previously impossible.</p><p>To date, Wiz has made the detection of what it calls ‘toxic combinations’ in cloud environments – including misconfigurations, excessive permissions, and overlooked vulnerabilities – its bread and butter. </p><p>Google Cloud wants to apply this specialism to its entire platform, as well as those of partners.</p><p>While AI can be used to bolster defenses, improperly configured LLMs and AI agents can also complicate <a href="https://www.itpro.com/security/28133/what-is-cyber-security">cybersecurity </a>strategies and introduce new risks of their own, such as <a href="https://www.itpro.com/security/ncsc-issues-urgent-warning-over-growing-ai-prompt-injection-risks-heres-what-you-need-to-know"><u>prompt injection attacks</u></a>.</p><p>Google Cloud and Wiz aim to identify these risks in real-time by embedding security into AI as it’s deployed rather than bolting it on after the fact. At Google Cloud Next 2026, both firms have led with a joint message of a deep-set need for AI preparedness.</p><p>“One of the things we do is provide the right protections associated with rolling out your AI infrastructure that you can then make available,” said Francis deSouza, COO and president, Security Products at Google Cloud.</p><p>“Because [it is] not possible to retrofit security into an IT environment, you need to design it in from the ground up.”</p><p>For example, Wiz has launched AI agents that act as automated security researchers and engineers: Red Agent, Blue Agent, and Green Agent.</p><p>“Red team is typically contesting from the outside, and the Red Agent actually leverages the deep visibility that we get into cloud environments and identifies all of the potential exposures, all of the APIs that are internet exposed,”<em> </em>explained Yinon Costica, VP Product and co-founder of Wiz.</p><p>“The Green Agent actually takes the risk that the, let's say the red agent found, and then it automates the triage process.”</p><p>Costica explained that this speeds up a process that normally takes days or even weeks of manual security work. Finally, the Blue Agent takes indications of compromise and automates the investigation process into these threats.</p><p>He added that the three agents can collaborate and work with one another to improve defenses. For example, Red Agent can invoke Green Agent after identifying a vulnerability, or Blue Agent can attempt to flag intrusions by Red Agent to test the effectiveness of an organization’s detection capabilities.</p><p>All of the agents are managed via Wiz AI Application Protection Platform (AI-APP), an umbrella offering that secures every enterprise AI layer including agents, models, infrastructure, data, and access permissions. </p><p>Reflecting on the increasing pace of AI-powered attacks, Costica argued that firms must begin to apply agents to “start using AI against ourselves as much as possible” to identify weaknesses and fix them before attackers find them.</p><p>“The goal for this year will be to automate all security processes we know, because that's the only way to go. </p><p>“So there are human automations that we've been using in industry, but I think what agents allow us to do is get to the next level of acceleration and automation of security work.”</p><p>This sentiment was echoed by deSouza in the <a href="https://www.itpro.com/cloud/live/google-cloud-next-2026-all-the-live-updates-as-they-happen"><u>event’s opening keynote</u></a>, in which he warned that hackers have reduced the time between initial access and handoff to secondary threat groups from eight hours to 22 seconds.</p><h2 id="secure-interoperability">Secure interoperability</h2><p>A core part of the updated security offering at Google Cloud is its interoperability. The hyperscaler has partnered with firms such as Darktrace, Gigamon, and SAP to extend its out of the box security features to their platforms, and also supports end-to-end AI application protection on platforms like AWS, Microsoft Azure, and Oracle Cloud.</p><p>Customers can even extend security functions to SaaS tools such as OpenAI and custom-hosted cloud environments.</p><p>“It is a <a href="https://www.itpro.com/cloud/34476/what-is-multi-cloud">multi-cloud</a> world, and we talk to our customers about that all the time saying every company is a multi-cloud company, and will need to continue to be a multi-cloud company,” said deSouza,</p><p>In addition to its GCP-native functionality, Wiz supports Databricks and agent platforms including AWS <a href="https://www.itpro.com/technology/artificial-intelligence/amazons-go-build-it-ai-strategy-is-a-perfect-fit-for-openais-big-enterprise-push"><u>Agentcore</u></a>, Gemini Enterprise Agent Platform, Microsoft Azure Copilot Studio, and Salesforce <a href="https://www.itpro.com/business/business-strategy/salesforce-announces-huge-partner-program-revamp-with-agentforce-360-launch"><u>Agentforce</u></a>. </p><p>Customers can also extend its security features to external security ecosystems such as Cloudflare AI Security for Apps, Google Cloud Apigee, and <a href="https://www.itpro.com/security/data-breaches/everything-we-know-about-the-vercel-data-breach-so-far"><u>Vercel</u></a>.</p><h2 id="expanding-security-agents">Expanding security agents</h2><p>Alongside the Wiz-developed tools, Google Cloud also highlighted its AI agents for SecOps. These include <a href="https://www.itpro.com/security/google-just-launched-a-new-gemini-powered-dark-web-monitoring-service"><u>Dark Web Intelligence</u></a>, an agent announced at RSAC Conference 2026 that produces organization-specific threat profiles based on Google Threat Intelligence Group’s <a href="https://www.itpro.com/security/32117/what-is-the-dark-web"><u>dark web</u></a> monitoring. </p><p>DeSouza noted that Dark Web Intelligence can identify threats with 98% accuracy, a step above what was previously capable within the security industry.</p><p>“Coming up with new rules is a complex process, it can be done by humans but you have new intelligence coming out all the time,” said deSouza, adding that many organizations have thousands of security rules already in place and that adding new ones onto this list can be an overwhelming task.</p><p>Additionally, cybersecurity teams can use the Threat Hunting Agent and Detection Engineering Agent to search for active threats and novel attacks, identify gaps in their organization’s cybersecurity, and create automated detection rules based on current threats.</p><p>In total, Google Cloud’s Triage and Investigation agent processed more than five million alerts in 2025. The firm said the tool helps teams to complete cybersecurity analysis that previously took 30 minutes in just 60 seconds, using Gemini for reasoning and text generation.</p><p>Another new platform, Google Cloud Fraud Defense, is intended to build upon reCAPTCHA services to assess the legitimacy of humans, bots, and agents within business processes and commerce.</p><p>The announcements made at Google Cloud Next 2026 represent not only the sum of Google Cloud’s teams but also the innovations of Google DeepMind and the proprietary technologies Wiz relied on prior to its acquisition.</p><p><em>ITPro </em>understands that Wiz uses a mix of backend models, which it chooses depending on customer use cases. But deSouza said that going forward both Google Cloud and Wiz teams would benefit from sharing notes with internal frontier model developers.</p><p>“We have an advantage here,” said deSouza. </p><p>“Our defense technologies like Wiz work with our frontier model teams, and we know in advance what the model is going to have. And so on day one, we're taking advantage of the cutting edge of defense to protect our customers.”</p><h3 class="article-body__section" id="section-follow-us-on-social-media"><span>FOLLOW US ON SOCIAL MEDIA</span></h3>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
                                <item>
                                                            <title><![CDATA[ Google expands Gemini Enterprise, consolidates Vertex AI services to simplify agent deployment ]]></title>
                                                                                                <dc:content><![CDATA[ <p>Google Cloud has expanded its Gemini Enterprise offering and consolidated existing services in a move intended to help developers manage agents, improve cross-team collaboration, and help enterprises deploy third-party agents.</p><p>Gemini Enterprise is now the central platform for all <a href="https://www.itpro.com/technology/artificial-intelligence/what-is-ai">AI </a>work within the organization, with a new ‘Inbox’ within the platform providing a single source for managing AI agents at enterprise scale.</p><p>The new Gemini Enterprise Agent Platform subsumes the features of <a href="https://www.itpro.com/technology/artificial-intelligence/google-cloud-targets-ai-anywhere-with-vertex-ai-agents"><u>Vertex AI</u></a>, including model and agent building capabilities, and combines them with <a href="https://www.itpro.com/software/development/ai-isnt-killing-devops-youre-just-using-it-wrong">DevOps</a>, agent orchestration, and security functionalities.</p><p>For building AI agents within the platform, users get access to the low-code Agent Studio, the <a href="https://www.itpro.com/software/28109/what-is-open-source">open source</a> code-first framework Agent Development Kit (ADK), and access to over 200 frontier models via Model Garden.</p><p>Beyond agent creation, the platform includes Agent Runtime, which handles agent orchestration and retains persistent context for long-form tasks, as well as Agent Identity, Agent Gateway, and Agent Registry, which together help IT teams to track agent identities and enforce AI guardrails.</p><p>Comcast used ADK and Agent Runtime to rebuild its Xfinity Assistant, which is now capable of performing personalized troubleshooting for customers.</p><p>“Agent Runtime has been a massive accelerator, allowing us to deploy a sophisticated multi-agent architecture that increases digital containment while ensuring secure, grounded interactions via Gemini,” said Rick Rioboli, CTO, Connectivity &amp; Platforms at Comcast.</p><p>“We aren't just reducing repeat interactions by solving customers’ issues the first time; we're redefining the customer experience at scale.”</p><p>Post-deployment, admins can track and score agent behavior within the platform via Agent Evaluation, Agent Observability, and Agent Simulation. These assess agent activity in the wild, provide insights into multi-step agent reasoning to help debug errors, and test agents against synthetic user prompts.</p><p>Employees can now build and deploy agents using natural language and reusable skills within the Gemini Enterprise app, chat directly as well as manage all agent activity via a central dashboard called Inbox in Gemini Enterprise.</p><p>In a demo, Erica Chuong, Applied AI, forward deployed engineering manager NorthAM at Google Cloud, showed how an agent within Gemini Enterprise could help an employee at a furniture retailer relaunch a defunct product line. </p><p>Chuong set off a series of agents to analyze interior design trends, identify unsold stock, and come up with plans for a relaunch campaign.</p><p>“With that one prompt, multiple agents complete a series of tasks in minutes instead of hours,” said Chuong.</p><p>“It knows exactly which agents to call to come up with a plan. My market research agent, powered by Deep Research, analyzes the latest Google search information alongside our own sales and CSAT data. </p><p>“My data insights agent is connecting to our global product catalog, regardless of location or format. It knows that dead stock refers to stale inventory, and it uses our agentic data cloud to identify the right data sets. My product strategy agent pulls everything together.”</p><h2 id="collaboration-with-agents">Collaboration with agents</h2><p>A core aspect of Google Cloud’s new end-to-end Gemini Enterprise is encouraging workers to share AI interactions with their co-workers.</p><p>Via a new offering called Projects, employees can share sources with an agent including their enterprise <a href="https://www.itpro.com/technology/artificial-intelligence/google-workspace-just-got-a-huge-gemini-update-heres-what-to-expect-with-new-ai-features-in-docs-slides-sheets-and-drive">Google Workspace</a>, OneDrive, or chats in a third-party collaboration software, to help come up with new ideas and brainstorm and produce data-driven business plans.</p><p>Another new feature, Canvas, helps workers to co-create Google Docs and Slides with one another directly within Gemini Enterprise. These documents can then be shared in <a href="https://www.itpro.com/desktop-software/19337/office-365-review">Microsoft 365</a> formats if needed, for external collaboration and feedback.</p><p>In the name of furthering collaboration and AI-driven results within the enterprise, Google Cloud has also announced interoperability with third-party agents within the Gemini Enterprise app. This allows users to deploy third-party agents directly within the Agent Gallery, for easy access to validated agents powered by partner developers.</p><p>As an example at time of the announcement, the firm pointed to agents from organizations including Adobe, Salesforce, ServiceNow, and <a href="https://www.itpro.com/software/workday-sana-hr-ai-automation"><u>Workday</u></a>.</p> ]]></dc:content>
                                                                                                                                            <link>https://www.itpro.com/technology/artificial-intelligence/google-expands-gemini-enterprise-consolidates-vertex-ai-services-to-simplify-agent-deployment</link>
                                                                            <description>
                            <![CDATA[ Gemini Enterprise Agent Platform aims to help organizations to build, scale, govern, and optimize AI agents ]]>
                                                                                                            </description>
                                                                                                                                <guid isPermaLink="false">vRrvRQwLpj3kQ65a9NLXW3</guid>
                                                                                                <enclosure url="https://cdn.mos.cms.futurecdn.net/L8U9qQhCGYW3UAziA7Z5J-1920-80.jpg" type="image/jpeg" length="0"></enclosure>
                                                                        <pubDate>Wed, 22 Apr 2026 12:00:00 +0000</pubDate>                                                                                                                                <updated>Thu, 19 Nov 2026 12:32:32 +0000</updated>
                                                                                                                                            <category><![CDATA[Agentic AI]]></category>
                                                    <category><![CDATA[Artificial Intelligence]]></category>
                                                                                                <author><![CDATA[ rory.bathgate@futurenet.com (Rory Bathgate) ]]></author>                    <dc:creator><![CDATA[ Rory Bathgate ]]></dc:creator>                                                                                    <dc:source><![CDATA[ https://cdn.mos.cms.futurecdn.net/Vep5JogbPhduK7R6CUWAm6-320-70.jpg ]]></dc:source>
                                                                <dc:description><![CDATA[ &lt;p&gt;Rory Bathgate was the Features and Multimedia Editor at ITPro until 2026, overseeing all in-depth content and case studies. Throughout his time at ITPro, Rory  charted the rise in popularity of generative AI and specifically companies such as Microsoft, OpenAI, and Google. &lt;/p&gt;&lt;p&gt;Alongside this, he delved into increasing calls for ethical and responsible AI as global legislators circle the technology, as well as the latest in mobile networking technology, from 5G mmWave to the 3G sunset and how it will affect businesses.&lt;/p&gt;&lt;p&gt;He  provided coverage from high-profile tech conferences such as Dell Technologies World, SuiteWorld, and VMware Explore Europe. His on-the-ground coverage  included live blogs, extensive daily coverage of the most significant announcements, analysis pieces, and podcast.&lt;/p&gt;&lt;p&gt;Rory was the co-host of the ITPro Podcast alongside Jane McCallion, where he swapped a keyboard for a microphone to discuss the latest learnings in tech.&lt;/p&gt;&lt;p&gt;Rory has an MA in Eighteenth-Century Studies from King’s College London, as well as a BA in English and American Literature from the University of Kent. He joined ITPro in 2022 as a graduate, after four years in student journalism.&lt;/p&gt;&lt;p&gt;In his free time, Rory enjoys photography and video editing, and can often be found at the cinema or reading a good science fiction paperback.&lt;/p&gt; ]]></dc:description>
                                                                                                                                <cf:isSponsored>false</cf:isSponsored>
                <cf:hasAffiliateLinks>false</cf:hasAffiliateLinks>
                <cf:isPaid>false</cf:isPaid>
                                                                                                                                                                                    <media:content type="image/jpeg" url="https://cdn.mos.cms.futurecdn.net/L8U9qQhCGYW3UAziA7Z5J-1920-80.jpg">
                                                            <media:credit><![CDATA[Google Cloud]]></media:credit>
                                                                                                                                                                                                                                    <media:description><![CDATA[A screenshot of Google Cloud&#039;s Gemini Enterprise app, specifically the Agent gallery screen demonstrating pre-built agents for deployment such as &#039;Synthesizer&#039;, &#039;Deep Research&#039;, and &#039;Idea Generation&#039;.]]></media:description>                                                            <media:text><![CDATA[A screenshot of Google Cloud&#039;s Gemini Enterprise app, specifically the Agent gallery screen demonstrating pre-built agents for deployment such as &#039;Synthesizer&#039;, &#039;Deep Research&#039;, and &#039;Idea Generation&#039;.]]></media:text>
                                <media:title type="plain"><![CDATA[A screenshot of Google Cloud&#039;s Gemini Enterprise app, specifically the Agent gallery screen demonstrating pre-built agents for deployment such as &#039;Synthesizer&#039;, &#039;Deep Research&#039;, and &#039;Idea Generation&#039;.]]></media:title>
                                                    </media:content>
                                                    <media:thumbnail url="https://cdn.mos.cms.futurecdn.net/L8U9qQhCGYW3UAziA7Z5J-1920-80.jpg" />
                                                                                                                                                                    <content:encoded >
                            <![CDATA[
                            <article>
                                <p>Google Cloud has expanded its Gemini Enterprise offering and consolidated existing services in a move intended to help developers manage agents, improve cross-team collaboration, and help enterprises deploy third-party agents.</p><p>Gemini Enterprise is now the central platform for all <a href="https://www.itpro.com/technology/artificial-intelligence/what-is-ai">AI </a>work within the organization, with a new ‘Inbox’ within the platform providing a single source for managing AI agents at enterprise scale.</p><p>The new Gemini Enterprise Agent Platform subsumes the features of <a href="https://www.itpro.com/technology/artificial-intelligence/google-cloud-targets-ai-anywhere-with-vertex-ai-agents"><u>Vertex AI</u></a>, including model and agent building capabilities, and combines them with <a href="https://www.itpro.com/software/development/ai-isnt-killing-devops-youre-just-using-it-wrong">DevOps</a>, agent orchestration, and security functionalities.</p><p>For building AI agents within the platform, users get access to the low-code Agent Studio, the <a href="https://www.itpro.com/software/28109/what-is-open-source">open source</a> code-first framework Agent Development Kit (ADK), and access to over 200 frontier models via Model Garden.</p><p>Beyond agent creation, the platform includes Agent Runtime, which handles agent orchestration and retains persistent context for long-form tasks, as well as Agent Identity, Agent Gateway, and Agent Registry, which together help IT teams to track agent identities and enforce AI guardrails.</p><p>Comcast used ADK and Agent Runtime to rebuild its Xfinity Assistant, which is now capable of performing personalized troubleshooting for customers.</p><p>“Agent Runtime has been a massive accelerator, allowing us to deploy a sophisticated multi-agent architecture that increases digital containment while ensuring secure, grounded interactions via Gemini,” said Rick Rioboli, CTO, Connectivity &amp; Platforms at Comcast.</p><p>“We aren't just reducing repeat interactions by solving customers’ issues the first time; we're redefining the customer experience at scale.”</p><p>Post-deployment, admins can track and score agent behavior within the platform via Agent Evaluation, Agent Observability, and Agent Simulation. These assess agent activity in the wild, provide insights into multi-step agent reasoning to help debug errors, and test agents against synthetic user prompts.</p><p>Employees can now build and deploy agents using natural language and reusable skills within the Gemini Enterprise app, chat directly as well as manage all agent activity via a central dashboard called Inbox in Gemini Enterprise.</p><p>In a demo, Erica Chuong, Applied AI, forward deployed engineering manager NorthAM at Google Cloud, showed how an agent within Gemini Enterprise could help an employee at a furniture retailer relaunch a defunct product line. </p><p>Chuong set off a series of agents to analyze interior design trends, identify unsold stock, and come up with plans for a relaunch campaign.</p><p>“With that one prompt, multiple agents complete a series of tasks in minutes instead of hours,” said Chuong.</p><p>“It knows exactly which agents to call to come up with a plan. My market research agent, powered by Deep Research, analyzes the latest Google search information alongside our own sales and CSAT data. </p><p>“My data insights agent is connecting to our global product catalog, regardless of location or format. It knows that dead stock refers to stale inventory, and it uses our agentic data cloud to identify the right data sets. My product strategy agent pulls everything together.”</p><h2 id="collaboration-with-agents">Collaboration with agents</h2><p>A core aspect of Google Cloud’s new end-to-end Gemini Enterprise is encouraging workers to share AI interactions with their co-workers.</p><p>Via a new offering called Projects, employees can share sources with an agent including their enterprise <a href="https://www.itpro.com/technology/artificial-intelligence/google-workspace-just-got-a-huge-gemini-update-heres-what-to-expect-with-new-ai-features-in-docs-slides-sheets-and-drive">Google Workspace</a>, OneDrive, or chats in a third-party collaboration software, to help come up with new ideas and brainstorm and produce data-driven business plans.</p><p>Another new feature, Canvas, helps workers to co-create Google Docs and Slides with one another directly within Gemini Enterprise. These documents can then be shared in <a href="https://www.itpro.com/desktop-software/19337/office-365-review">Microsoft 365</a> formats if needed, for external collaboration and feedback.</p><p>In the name of furthering collaboration and AI-driven results within the enterprise, Google Cloud has also announced interoperability with third-party agents within the Gemini Enterprise app. This allows users to deploy third-party agents directly within the Agent Gallery, for easy access to validated agents powered by partner developers.</p><p>As an example at time of the announcement, the firm pointed to agents from organizations including Adobe, Salesforce, ServiceNow, and <a href="https://www.itpro.com/software/workday-sana-hr-ai-automation"><u>Workday</u></a>.</p>
                                                            </article>
                            ]]>
                        </content:encoded>
                                                </item>
            </channel>
</rss>