Cost exhaustion attacks, disposable applications, and evidence custodians – these are the ideas businesses need to prepare for within the next three years or less.

That's according to analyst house Gartner, which laid out a list of ten predictions for 2027 to 2030, many of which are focused on AI and its potential business impacts.

That includes physical AI or robots, agent swarms clogging up public services or running down budgets, and unpicking how to actually manage AI costs, providers, and risks.

"Many of the systems we take for granted today, from public services to software, energy and workforce models, will be fundamentally transformed by AI over the next decade," said Daryl Plummer, distinguished VP analyst and Gartner fellow.

"The most successful organizations will be those that balance innovation with responsibility, building the capabilities needed to manage both the opportunities and the unintended consequences of an AI-driven world," Plummer added.

AI risks on the horizon

The analyst firm predicted that more than ten billion autonomous agents will be clogging up public services by 2030. Those agents will have been made by people, companies, and even governments in order to take on basic tasks for users. But that will lead to a sharp increase in the number of applications and other transactions, putting pressure on the government to keep up.

To prepare, Gartner called on governments to modernise their digital infrastructure and strengthen verification, though they'll only have a few years to do so.

Alongside that, organisations will face a new type of cybersecurity risk: cost exhaustion attacks. These are like a denial of service attack, but focused on driving up the cost of AI used by the company.

"The rise of public-facing AI is creating a new cybersecurity risk called cost exhaustion attacks, where malicious actors deliberately drive excessive AI usage to increase operational costs," Gartner said in a statement.

To avoid this, companies should consider AI usage patterns and token consumption as security concerns, ensuring both are monitored, saying eight in ten companies will face these attacks within the next few years.

AI business

Gartner also predicts that by 2030, 80% of front-line employees working for global companies will be working alongside or with physical AI, such as robots, drones, autonomous vehicles, or "other embodied systems that sense, interact with, and influence their surroundings."

By 2029, most applications will become disposable – on purpose. These disposable apps will be used for less than one year, with employees using AI app development in order to meet short-term needs, ditching the tool when it's no longer needed. Gartner warned that will spark governance and security challenges.

Beyond AI-generated apps, Gartner also forecasts the rise of AI-powered products, which will allow companies to introduce "customised, efficient and scalable offerings," leaving behind those who haven't gotten to grips with the tech. Gartner also predicted more firms will find themselves generating power, selling it back to grids or to AI data centres.

Watching AI – and its cost

To help manage the rise of AI, companies will create a new role or corporate function dedicated to managing AI costs and mapping it to value or profit, with 60% of the largest companies embedding financial operations controls at inference, leading to real-time optimisation and cost governance. CIOs, meanwhile, will become the "evidence custodian" for AI accountability in businesses, ensuring guardrails are in place.

Alongside that, Gartner predicted that insurers will drive AI governance in organizations, rather than regulators, with policy-based AI built into systems and workflows to address the growing concerns around risks and liabilities associated with the technology.

FOLLOW US ON SOCIAL MEDIA

Salesforce has announced AIforce, a new ‘live interface layer’ designed to streamline data access for agentic AI.

Unveiled ahead of the company’s annual Dreamforce conference, Salesforce said the move aims to bring the “full power of its platform to any interface”.

Put simply, the new interface gives agents deeper access to various data sources stored across the Salesforce ecosystem by removing the need for a dedicated user interface.

Operating through a dedicated user interface has traditionally limited who could use the platform, according to Salesforce, as well as how much value they could derive from it.

Moving forward, users will be able to “build their own UI” using natural language prompts based on their specific needs, rather than traversing an array of fixed systems.

“It lets them compose customized live interfaces on the fly, simply by describing them,” the company said in a statement. “This gives everyone their own UI, built to surface exactly the insights they need, when and where they need them.”

Speaking in a press briefing ahead of Dreamforce, Patrick Stokes, president of applications and marketing at Salesforce, said AIforce will help customers “break free of a shared user interface” and create a system where “every individual has their own intelligent, dynamic interface that comes to them”.

“The UI is not actually our product, the Salesforce product is the platform where our customers store their data, their metadata, their workflows, all of their permissions and their security. Basically where they encode the way their business runs,” he said.

The move by Salesforce builds on the company’s Headless 360 announcement in April this year.

The new service allows users to access the Salesforce platform either through an API, MCP tool, or command-line interface (CLI), removing the need for a dedicated browser interface.

As with AIforce, the long-term goal here is to streamline access to Salesforce data for users, but also agents, providing deeper integrated access for third-party agentic tools.

Unlocking ‘trapped value’ with AIforce

Moving toward a more fluid interface setup will broaden appeal on this front, according to Salesforce, opening the platform up to a wider range of enterprises and users who’ve traditionally been locked out of accessing vital Salesforce data and business logic.

According to Salesforce, this will also help agents “unlock trapped value” by providing deeper access to company records and data sources.

Stokes told assembled media that AIforce will be underpinned by three new ‘surfaces’ – Claudeforce, Slackforce, and Agentforce Coworker. The latter two are both new services, while Claudeforce was unveiled by the CRM giant in early September.

Slackforce will allow users to create “rich new applications” within Slack, according to Stokes, and builds on a long-running evolution aimed at positioning the collaboration platform as a hub for agents.

This is essentially a combination of Slackbot with a new ‘Slack Surfaces’ service, which allows users to pull relevant data from Salesforce within the platform.

Agentforce Coworker, meanwhile, is an ‘AI teammate’ that lives within the Lightning interface, offering reasoning capabilities spanning accounts, activity, and history and “surfacing insights and taking action” on behalf of users.

Stokes said this is aimed primarily at enterprises still using the Lightning interface, which replaced the Salesforce Classic platform.

Combined, these three new surfaces will overhaul the “traditional Salesforce experience” and offer users a broader range of UI options, he added.

FOLLOW US ON SOCIAL MEDIA

Salesforce has unveiled Koa, a new AI reasoning model designed specifically for CRM functions.

Announced at the company’s annual Dreamforce conference in San Francisco, the reasoning model was built using Nvidia Nemotron as part of a collaboration between the two companies.

Koa is “purpose-built to help agents reason through complex, multi-step workflows”, according to Salesforce, and draws upon a synthetic dataset modelled on 27 years’ worth of CRM deployment data.

The reasoning model is designed to supercharge both human sales professionals and agents, according to Rohan Kumar, president and chief platform and engineering officer at Salesforce.

“We’ve brought together this model, which we believe is going to be a game changer when you look at complex agents that do multi-step tasks,” he said.

“When you think about a sales agent that’s trying to figure out the next best step in closing a phenomenal sales deal, Koa’s going to help with that.”

What to expect with Koa

According to Salesforce, the Koa model was built using entirely “synthetic scenarios” that reflect the typical reasoning, tool use, and decision-making processes used across decades’ worth of CRM workflows.

This includes generating leads, qualifying opportunities, and resolving service cases, for example. The key here is that these were used to simulate real-world enterprise workflows spanning a range of industries such as manufacturing, healthcare, travel, and financial services.

“Each scenario paired a persona with specific tasks, then mapped the sequence of actions and tool calls an agent must make to complete them,” the company explained in a statement.

As the model was built using Nvidia Nemotron, Salesforce controls the weights and runs Koa within its own dedicated infrastructure, a key talking point for the company from a security perspective.

Koa has already been deployed inside Salesforce, the company revealed. This includes the use of an agent in Slack designed to help employees find information and complete every day tasks.

During a press briefing ahead of the conference, Kumar once again described Koa as a “game changer” for staff internally.

A host of customers are also piloting the reasoning model, including Formula 1, Xero, Baxter Credit Union, and UChicago Medicine, the company revealed.

FOLLOW US ON SOCIAL MEDIA

Microsoft has become the latest big tech firm to weigh in on the dangers of AI, with a new Humanist Code of Conduct that's up for public consultation.

The document sets out the principles on which the company plans to train its models, saying it's motivated by a single overriding objective: that humans must retain meaningful control over AI.

The principles it outlines are not yet being used to train Microsoft's models, with the company saying it wants to gather feedback through a public consultation. It will then publish a revised version of the code towards the end of this year, which will be used to guide model development in 2027 and beyond.

"The recent safety incidents of large scale, highly coordinated, and persistent hacking campaigns of AI agents prove that there's no time to waste. The stakes are high and only getting higher," said the firm.

"We believe it is more important than ever to create safe and reliable AI in service of people, and to be transparent about how we go about it."

The document prohibits AI models from certain behaviours: they mustn't help with chemical, biological, radiological or nuclear weapons, generate violent or sexually explicit content or carry out offensive cyber operations.

They must not try to evade human oversight, engage in large-scale manipulation or make claims to consciousness, feelings or legal personhood.

"Whilst the science of AI consciousness is far from settled, we believe that training these systems to imitate consciousness-like states increases the challenge of containment, control, and alignment," Microsoft said.

The company said it's already consulted with experts in AI, law, ethics, philosophy, linguistics, and public policy, as well as business leaders from across industry, and has also held a series of focus groups with members of the public.

However, it's calling for more public input into the code of conduct, with a six-week consultation now open.

"We are especially interested in the hard parts: how can we better cement the right values in our models? How to be more concrete about the meaning of 'human flourishing'? Where is the language too loose to evaluate? How do multi-agent scenarios impact things?" it said.

"And perhaps most importantly of all, how do we continue to accelerate progress, whilst also ensuring we maintain healthy and necessary safety constraints?"

The release of the code of conduct follows a surge of concern over AI safety, including the resignation of Anthropic researcher Jacob Coxon. AI could, he said, lead to the extinction of the human race if more controls aren't put in place.

Anthropic's Dario Amodei has weighed in with a call to slow down AI development, as has OpenAI's Sam Altman – but not Donald Trump, who has described such concerns as 'a hoax'.

"AI's opportunity to deliver transformational benefits for humanity is real. And yet, if mismanaged, it will also pose grave dangers, including new and as-yet poorly understood frontier risks," Microsoft said.

"Models, and the process by which they are developed, deserve public scrutiny and engagement."

FOLLOW US ON SOCIAL MEDIA

Fears of AI sparking an existential crisis for humanity have been batted away by American President Trump, who called the predictions a "hoax" in an interview-interrupting phone call to Nvidia's CEO.

Last week, an Anthropic researcher quit the company over concerns that the AI arms race was unsafe for humans. One of his colleagues subsequently predicted there was a more than 10% chance that AI will end humanity in the near future.

That sparked a call for greater regulation of AI and calls from inside the industry – including from Anthropic CEO Dario Amodei – to slow down the pace of frontier model development, with the US Congress looking set to demand action.

The fears of AI superintelligence causing harm to humans aren't new, and plenty of experts have suggested this is another phase in the AI hype cycle, designed to boost Anthropic ahead of a looming IPO, let AI companies self-regulate, or slow the progress of challengers such as China to keep them from catching up.

Trump calls AI fears a hoax

Trump also disputed the AI doomerism fears, though for different reasons. His addition to the discourse came via a phone call to Nvidia's CEO Jensen Huang, who was onstage at the All-In Summit in Los Angeles, California at the time. Huang answered the phone, put it on speaker mode, and informed Trump that he was "speaking with the world".

"The robots will not be taking over," he said. "The AI will not be taking over the rest of the world. The whole thing is a hoax."

"I'm telling you, it's all a hoax," Trump continued. "The data centers are great, and they make people wealthy, and they make states wealthy. And it's the oil of the next 20, 25 years."

Of critics, he said: "They're just playing into the hands of a lot of people that don't want to see it happen. And that could be political people, it could also be China, and we're not going to let that happen. It's a hoax."

Trump did admit that "we have to be careful" and "do things prudently", but added: "that doesn't mean we're going to stop an industry".

At one point, Huang responded: "We're going to make sure that everybody wins in the AI race in America. Every industry, every company, every state, every people."

No guardrails, just Trump?

The call echoes other comments made by the American president on his personal social media site, which said pushback against the technology was a "sick conspiracy." Trump also said that Anthropic's CEO was "pretending to be a 'perfect little angel'" and that the US administration had previously stopped Amodei and other "AI people" from doing "bad or potentially bad things".

Rather than regulation or a pause in development, all AI needs is a "strong and smart" president.

The American people don't necessarily agree. A poll by the University of Massachusetts found only 23% of people approve of Trump's handling of AI, versus 57% who disapprove. A separate study from the University of Maryland found 85% of Democrats and 79% of Republicans want AI regulated and monitored by a government agency.

Congress also doesn't appear to agree with Trump, with Democrats and some Republicans pushing for action – though as one noted, government regulation of the technology might not come easy given the current political climate. Rick Scott, a Republican senator for Florida, reportedly said: "Did we balance the budget? Oh, did we pass a budget? See, what have we gotten done? So you think we're going to get AI done?"

FOLLOW US ON SOCIAL MEDIA

As AI initiatives scale, managed service providers are under pressure to help customers deploy new capabilities quickly while enhancing quality and controlling costs. But behind the scenes, engineers are running into a quiet bottleneck—power delivery at a reasonable cost.

Modern AI compute is increasingly limited by how quickly and efficiently engineers can deliver power to processors, accelerators and memory. For Managed Service Providers (MSPs), that makes power delivery more than an engineering issue. It becomes a strategic consideration that affects capacity planning, operating costs, cooling requirements and the ability to scale customer environments reliably. The challenge is not having enough watts—it’s getting those watts to the right place, with minimal latency, without excess heat, all while managing operational costs to keep revenue in check.

MSPs that overcome the AI power problem will be the ones that treat power not as an afterthought, but as a foundational pillar of performance, efficiency and scalability.

Why power density matters to MSP service delivery

A decade ago, a 200 W GPU was considered massive. Today, single AI accelerators regularly exceed 600 W, and full server boards can draw 5–10 kW.

That power needs to be delivered across a variety of distances, to multiple voltage domains, and respond in nanoseconds as workloads shift. Every bit of distance, impedance or delay creates loss, droop, and heat, hindering scalability and delivery service.

In short: as computing density grows, power density becomes the new productivity limiter.

This shift has also changed how engineers think about system-level design. Instead of optimizing only compute capability, architects must now consider the electrical and thermal realities that shape what is actually achievable.

High-performance silicon is only as good as the infrastructure that supports it, and many organizations are discovering that traditional power-delivery networks were never designed for bursty, AI-driven workloads. For customers and MSPs alike, these workloads demand not just more power, but more responsiveness and stability than previous generations ever required.

Leveraging power innovation to build customer trust

One emerging approach to this power delivery challenge is moving voltage regulation closer to—or even beneath—the silicon itself. Integrated voltage regulation (IVR) technologies are designed to deliver cleaner, faster, and more localized power that scales with chip demands.

For MSPs, these advances can support more predictable performance, better energy efficiency, and infrastructure designs that are easier to scale across customer deployments. Some of the positive outcomes associated with these advanced IVR solutions include:

• Higher current density compared with traditional discrete regulators, enabling denser AI infrastructure and serving modern AI workloads.
• Efficiency improvements by reducing long power-delivery paths and associated transmission losses.
• Low thermal resistance to dissipate power losses in a constrained environment, supporting more reliable performance and reduced cooling requirements.

These improvements are not incremental gains—they enable entire system architectures that were previously constrained by power delivery. This means a more profitable, effective, and efficient infrastructure that MSPs and customers can trust.

Building delivery-driven AI infrastructure

With IVRs and other localized power-delivery technologies, GPUs, CPUs, and custom AI accelerators can sustain higher operating frequencies under dynamic workloads by reducing impedance and minimizing voltage droop. This translates to:

• Data centers can deliver more compute per watt and reduce cooling overhead, improving the economics of managed AI services.
• Edge and distributed deployments can operate more efficiently, helping MSPs support AI use cases closer to customer environments.
• Compact, modular designs can simplify scaling and make it easier to standardize AI infrastructure across multiple customer deployments.

Instead of designing around power constraints, system architects can now design with power as an enabler of AI services.

Looking ahead

As more businesses and MSPs move towards AI to automate workflows and streamline services, power delivery for scaling AI must evolve too. The next wave of competitive advantage will not come only from smarter models or faster chips; it will also come from power architectures that help providers deliver performance, efficiency and control costs-- at scale.

For MSPs, the opportunity is clear: now is the time to reassess whether your AI infrastructure strategy can support the performance, efficiency and scalability your customers will expect next.

Providers that invest early in smarter power-delivery architectures will be better positioned to differentiate their services, protect margins and lead customers confidently into the next phase of AI adoption.

There is little doubt that the channel is going to play a pivotal role during this next phase of enterprise AI adoption. The shift from experimenting with isolated AI projects to implementing intelligent, autonomous operations means that enterprises will be faced with a new level of sophistication and complexity.

In many cases, helping them solve the AI puzzle will fall squarely on the shoulders of those companies operating in the channel. Their experience, expertise, and problem-solving will be essential to help enterprises make sense of it all. And in doing so, they can tap into one of the fastest-growing areas of IT today.

In May 2026, Gartner predicted that AI spending was set to grow by 47% by the end of 2026 as enterprises expand their use of both the GenAI models embedded in existing software applications and the new AI agents within multiple workflows.

Indeed, Gartner went further, suggesting that up until relatively recently, AI spending has “primarily been driven by technology companies and hyperscalers” with enterprises “yet to really flex their spending potential”. That, says Gartner, looks set to gain momentum.

If that is the case, then both the channel and vendors have been put on alert because if they are to benefit from this opportunity, it will need more than just local knowledge and the right solutions. It also needs the right partner program to bring it all together.

Because if nothing else, the AI era has shone a new spotlight on how channel programs work. And with partners becoming increasingly discerning about their network, vendors will have to seriously reassess what a ‘good’ program looks like.

Consistency drives scale

An ideal, modern partner program is simple. Partners should not have to navigate complex terminology or inconsistent structures across regions. Regardless of whether they are in London, New York, or Singapore, they should all have the same experience.

In my experience, the most effective programs are those that are both easy to understand and easy to operate within. End customers are looking to move fast, and, as such, partners cannot afford to waste time cutting through unnecessary friction. Crucially, this will also help vendors in the long run as well, since a program that is easy to use ultimately becomes easier to manage.

It is important for vendors to remember that partners will often have to implement multiple tools for a single solution, meaning small amounts of friction can scale exponentially on more intricate initiatives. For example, a business looking to implement agentic AI will also need a way to constantly feed the agents with up-to-date data, meaning a partner will likely have to install a comprehensive content management system to guarantee the project’s success.

But that is just the beginning if organizations are going to achieve a truly ‘agentic enterprise’ — an operating model where humans and agents work together seamlessly. They will also likely need multiple solutions across infrastructure, security, governance, and implementation, all collaborating simultaneously to reach that level.

A truly shared partnership

In that same vein, transparency is key to success. Partners should have a clear view of pipeline, performance, and joint execution at all times, which also allows vendors to know clearly what’s working and what needs to change. After all, the basis for any successful business relationship is trust, meaning nothing should feel hidden or arbitrary.

Partners also need to consider flexibility. Different partners will have different strengths, meaning a good program allows them to engage in ways that suit them and evolve their role over time.

Ultimately, this boils down to mutual commitment. The relationship works best when it is a two-way street, with partners and vendors acting as a single team. When it is done right, the former commits resources and provides expertise, while the latter offers support, visibility, and incentives. Technology, systems, training, planning, sales motions, and, most importantly, success should be shared wherever possible.

The channel will not wait

In the same way that AI is only as strong as its supporting foundations, channel partners are only as strong as their network. And as organizations move away from isolated AI experiments to intelligent, governed outcomes at scale, that network now matters more than ever.

Modern partner programs are simple, transparent, flexible, globally consistent, and focused on long-term customer outcomes, with partners treated as strategic collaborators rather than just another sales channel.

If a vendor’s program falls flat on one of these elements and ends up being the weak link in the chain, they risk missing out. If Gartner is right about the opportunities coming down the road, then enterprises have not even scratched the surface of their spending potential. It is up to vendors now to ensure their programs can help them ride that wave when the time comes.

Wasabi Technologies has announced the launch of a dedicated AI business and has named semiconductor and storage specialist Pinaki Mukherjee to lead it.

Mukherjee will serve as senior vice president and general manager of the business, which aims to bring a dedicated strategy, partnerships, and go-to-market focus as AI-driven storage demand grows.

He brings more than 20 years' experience across semiconductors, storage, and AI infrastructure, having held senior business strategy and corporate development positions at Fungible, Druva, and Western Digital. He has also previously led semiconductor strategy and AI infrastructure engagements at global consulting firm Alvarez & Marsal.

Wasabi highlighted Mukherjee's track record, which includes more than $2 billion in partnership-driven revenue and over $10 billion in strategic M&A and investment outcomes.

"Mukherjee has a track record of driving the kind of high-value partnerships that move markets, and that's exactly what this moment requires," said Wasabi president and COO Marty Falaro.

"AI workloads are pushing storage demand to a scale we've never seen, and Mukherjee is the right person to build the partnerships that extend Wasabi's position as the industry's choice for cloud storage, at the exact moment inference is reshaping what that storage needs to do."

Targeting AI infrastructure demand

Wasabi currently stores hundreds of petabytes of AI data for customers such as frontier model labs, generative AI startups, and neocloud compute providers. The firm operates 16 global storage regions and works with 18,000 channel partners globally.

The new business unit gives Wasabi a dedicated operation focused on the storage requirements of AI customers, with the vendor aiming to expand its role as demand for AI infrastructure grows.

The firm's storage framework is designed to support data across the AI lifecycle and enable customers to move data to wherever their compute resources are located without egress or API fees.

Wasabi said this approach can help organizations avoid tying their data to a single cloud provider as their AI workloads evolve, while benefiting from a more predictable cost structure.

The vendor also highlighted recent deployments, including an image- and video-generation lab that migrated 175 petabytes off a hyperscaler within months, and a robotics data consortium that has stored multiple petabytes while projecting annual savings of more than JPY 100 million.

With the new business, Mukherjee said Wasabi is well-positioned to meet the needs of its customers as they scale "beyond what any single hyperscaler ecosystem can offer."

"AI customers don't need another hyperscaler, they need the freedom to move their data wherever their workloads take them, without egress fees, API fees, or lock-in dictating their architecture," he explained. "I'm looking forward to building the ecosystem that scales that freedom across the AI ecosystem."

2-in-1 laptops that can act as both a traditional ‘clamshell’ laptop and a tablet aren’t anything new in the computing world for both businesses and consumers. When configuring a laptop fleet, however, it can be challenging to decide if a standard laptop or a 2-in-1 will best suit the needs of your organization and its workers.

The Dell Pro 7 family offers both clamshell professional-grade laptops and 2-in-1 options, with a suite of configurable specs and features. To help you decide if a 2-in-1 option is right for your business needs, this article answers key questions about the Dell Pro 7 2-in-1.

TL;DR

• Dell Pro 7 2-in-1 laptops support hybrid and dynamic working styles with their multiple form factors
• 13-inch and 14-inch display options mean the Dell Pro 7 2-in-1 laptops can support field workers and those looking for a hybrid working setup
• There’s a myriad of spec options, from the latest AMD and Intel laptop processors to hefty amounts of RAM and storage
• Like other professional Dell laptops, the Dell Pro 7 2-in-1 laptops come with a host of built-in security features
• All Dell Pro 7 2-in-1 models are AI-ready out of the box
• Remote management options mean IT teams can keep Dell Pro 7 2-in-1 laptops safe and up-to-date, even within a highly hybrid workforce

How can workers use the Dell Pro 7 2-in-1?

A Dell Pro 7 2-in-1 can function like a standard laptop day in, day out, with the screen and specs to deliver a professional portable computer. When needed, though, the 360-degree hinge allows the display to be rotated back and laid flat against the laptop’s underside, turning the machine into a large-format tablet. This happens automatically with Windows 11, switching into a touchscreen-friendly tablet mode.

Tablet mode is handy for workers attending a site visit or walking around a factory floor, where it would be hard to use a standard laptop but easy to tap notes on a touchscreen or even handwrite them with a stylus.

By folding the display to a 45-degree angle, a Dell Pro 7 2-in-1 can be used in a ‘tent’ mode, effectively turning it into a compact portable display without a keyboard and trackpad getting in the way. This is useful for on-the-fly presentations or simply watching videos when relaxing on a business trip after a busy day of fieldwork.

How can you spec the Dell Pro 7 2-in-1?

There’s a mix of specification options you can go for with the Dell Pro 7 2-in-1:

• Intel and AMD processor options: The Intel models can be configured with an Intel Core Ultra 5 or Intel Core Ultra 7 CPU, with 8 cores and up to 4.5 GHz clock speeds in the entry-level processors, and 16 cores and up to 4.8GHz in the top-end CPU option
• AMD CPU options start with the Ryzen AI 5 PRO 435 with its 6 cores and up to 4.5GHz clock speeds, and end with the Ryzen AI 9 HX PRO 470 with its 12 cores and clock speeds of up to 5.2GHz
• RAM options: from 16GB to 64GB of LPDDR5x
• Storage: The Dell Pro 7 2-in-1 features SSD storage, starting at 256GB and going up to 2TB
• An optional mobile broadband card for cellular connectivity when on the move
• An optional built-in fingerprint scanner for biometric security

What are the display options for the Dell Pro 7 2-in-1?

Dell Pro 2-in-1 laptops come with a 13.3in or 14in display. Both use a Widescreen Ultra Extended Graphics Array) (WUXGA) format, meaning you get a resolution of 1920 x 1200 pixels and an aspect ratio of 16:10. This presents a solid display with more verticality than a standard Full HD display, which is better for productivity and tasks such as video editing.

The 13.3in model is likely to be the best option for workers who are often working on the move and out in the field, while the 14in option will suit those who tend to move between more static workspaces and would benefit from the larger display.

Is the Dell Pro 7 2-in-1 built for desk work or work on the move?

A Dell Pro 7 2-in-1 can be used for both desk work and by people on the move. Both the 13.3in and 14in models are compact enough for working on the move; whether that’s out in the field, while commuting or at a coffee shop.

But an array of port options, such as both USB-C and USB-A ports and an HDMI connection, means the Dell Pro 7 2-in-1 can form the heart of a desktop setup when connected to an external monitor, keyboard, mouse, and other peripherals.

Like other Dell Pro laptops, the 2-in-1 has been designed to support hybrid working, offering the portability for work on the move and the power for heavier tasks that might once have required a desktop computer.

How does the Dell Pro 7 2-in-1 support collaboration?

Perhaps the best feature of a 2-in-1 laptop is its ability to serve as a tool for collaboration. With the option to fold the display flat or back on itself to turn the laptop into a tablet, the Dell Pro 7 2-in-1 can be used as a touchscreen display on which multiple people can collaborate; say, manipulating an on-screen blueprint or graphic when discussing the design of a product or system.

For ad-hoc presentations, or those where a projector or external display isn’t readily available, putting a Dell Pro 7 2-in-1 into its ‘tent’ mode turns it into a handy presentation display for showcasing slide decks or playing promotional videos. For salespeople in the field, this could be a game-changer when compared to a standard laptop.

What's the webcam and microphone like on the Dell Pro 7 2-in-1?

As standard, all Dell Pro 7 2-in-1 models come with a Full HD infrared front-facing camera, offering a 1080p resolution at 30 frames per second, partnered with dual integrated noise-canceling microphones.

There’s the option to specify an 8-megapixel camera for boosted resolution; this could be the option to go for if your organization relies on a lot of video calls by working with dispersed teams or a very hybrid workforce.

AI enhancements, such as Microsoft Studio Effects, can augment video calls through the smart removal of background noise and automatic video framing.

Is the Dell Pro 7 2-in-1 AI-ready?

Yes. All the Dell Pro laptops are Microsoft Copilot+ compatible, meaning they have the processors and specs to support on-board AI tasks, notably with Windows 11’s built-in Copilot AI.

With processors that include a dedicated neural processing unit (NPU), the Dell Pro 7 2-in-1 can handle fast, responsive on-device AI queries and tasks without draining CPU or graphics power; this helps save battery life and improves overall system performance.

By running AI tasks on-device, everyday AI tools like Copilot work faster because they need less cloud access. More advanced AI tools will need cloud and web access, but the NPUs can help speed up their performance.

All of this means that Dell Pro 7 2-in-1 laptops can take advantage of built-in AI capabilities right out of the box, while also supporting third-party AI agents and applications that run either locally, on the device, or in the cloud.

Does the Dell Pro 7 2-in-1 come with any built-in security features?

Yes. There’s a mix of physical and software-based security features for the Dell Pro 7 2-in-1:

• A proprietary wedge-shaped Kensington lock lets users securely connect the laptop to a strong anchor point to prevent it from being snatched away while working in the field or out in public
• An optional fingerprint scanner offers physical biometric security when logging into the laptop or authorizing payments
• Deep built-in encryption goes into the hardware layer, locking down data on drives and storing encryption keys in a dedicated security chip
• Dell’s SafeBIOS suite offers threat prevention and detection at the BIOS level to prevent any tampering from malicious actors, even if they gain physical access to the laptop
• Threat telemetry data is also gathered, allowing IT teams to have more visibility into security threats even if they’re targeting below the operating system or have already been locked down

Such security is particularly useful for companies with an active hybrid workforce or employees who are often working on the move.

What support and device management is there for the Dell Pro 7 2-in-1?

Dell’s Management Portal offers IT teams a unified view of all Dell Pro 7 2-in-1 laptops and other Dell Pro laptops being deployed across the organization. This cloud-based service gives IT a centralized system to roll out and manage updates while ensuring security policies are deployed and configured correctly based on user needs.

For smaller businesses without dedicated IT teams, Dell’s ProDeploy Suite and ProSupport offer comprehensive IT support, from setting up and deploying a laptop fleet to ongoing management and eventual retirement when it's time to upgrade.

If you think the Dell Pro laptops are the right call for your business, find out more on the Dell website. US readers can visit here and CA readers can visit here.

Robust cyber resilience has rapidly become table stakes for any business that is serious about security, but good cyber resilience often comes at the cost of flexibility. Organizations are forced to adopt closed systems that – despite good intentions – limit the workloads, cloud platforms, and virtual machines an organization can deploy.

To address this, Dell has created PowerProtect One, a new, unified approach to cyber resilience.

Rather than forcing IT leaders to make a binary choice between flexibility and control, PowerProtect One provides a unified cyber resilience platform that brings together management, orchestration, and secure, efficient protection storage into a single, intelligent experience. In short, it adds cyber resilience without restricting what your infrastructure can do.

Let’s explore PowerProtect One and find out how it can help businesses improve backup security and recovery from cyberattacks.

TL;DR

• Dell PowerProtect One provides a unified platform from which IT leaders can manage cyber resilience activities
• PowerProtect One’s rapid recovery capabilities enable businesses to restore files and applications within minutes, minimizing downtime during unexpected outages
• PowerProtect One integrates with third-party backup tools, meaning it can work to bring robust cyber resilience to a mix of tools and systems without compromising the flexibility many IT environments require
• PowerProtect One safeguards sensitive data with secure, efficient immutable protection storage, giving companies confidence in defending against evolving cyberattacks
• AI-driven automation can take care of a lot of manual work an IT team may otherwise need to do, allowing them to focus on other projects and strategic initiatives
• The unified nature of PowerProtect One and its automated features enables rapid cyberattack recovery across complex IT infrastructure, without compromise
• Seamless integration with third-party environments ensures consistent backups across a diverse set of workloads, helping maintain business continuity for complex IT setups

The increasing threat of cyberattacks, especially in the era of AI, means it’s often not a matter of if your company will be attacked, but when. The threat is compounded by the rapidly expanding size of the attack surface, with on-premises infrastructure, multiple endpoints, layers of cloud services, and a diverse range of applications all providing fertile ground for hackers.

So, if cyberattacks are inevitable, there’s only one logical position to take: you have to make sure your organization has strong cyber resilience so it can recover quickly and confidently and minimize disruption.

Dell PowerProtect One simplifies cyber resilience with a unified platform

Protecting a business against cyberattacks and threats is complex, especially in a distributed IT environment. IT can protect against threats in these environments using custom tools and tailored policies, but this quickly becomes unwieldy, especially in a larger organization with a complex mix of hardware, software, and workloads. Dell PowerProtect One aims to simplify this by bringing together management, orchestration, and secure, efficient storage.

PowerProtect One achieves this through an open, integrated architecture that works with both Dell and third-party backup tools. PowerProtect One can reduce management overheads by up to 50%, cutting through the chaos of multiple backup systems, complex recovery workflows, and inconsistent policies across diverse infrastructures.

Essentially, this unified platform gives IT leaders more control over cyber resilience while reducing complexity, strengthening recovery confidence, and protecting critical data at scale without vendor lock-in.

Dell PowerProtect One helps your organization recover faster when every minute counts

Rapid recovery capabilities help organizations restore files and applications within minutes and minimize downtime during unexpected outages. To facilitate this, Dell PowerProtect One embraces three pillars of cyber resilience:

• An unbreakable foundation: features like data immutability, encryption, and efficient data reduction give you the confidence to operate securely and scale without compromise while moving your business forward.
• Frictionless operations: security and resilience shouldn’t come at the cost of operational complexity. PowerProtect One delivers frictionless operations through centralized visibility, integrated management, seamless ecosystem integration, and intelligent automation.
• Trusted resilience: PowerProtect delivers that confidence through decades of innovation, enterprise-grade reliability, and proven performance at scale. This is resilience teams can rely on when it matters most.

Taken together, this is what differentiates PowerProtect One: It’s secure at its core, simple to operate, and trusted when organizations depend on it most.

Dell PowerProtect One ensures your backups are recovery-ready

Ransomware targets an organization’s backup systems in order to prevent recovery and maximize chaos and disruption, so businesses must protect backups as much as core systems. End-to-end ransomware resilience safeguards sensitive data with continuous monitoring, giving companies confidence in defending against evolving cyber threats.

Dell PowerProtect One encrypts data at rest in backup copies and orchestrates backups across a diverse IT environment, from software to storage infrastructure. It then keeps backups secure with built-in malware and ransomware scanning and detection capabilities.

From virtual machines to databases, file systems, containers, and AI workloads, PowerProtect One covers your entire environment. When an attack strikes, you get every system back up and running, fast.

As mentioned above, a unified interface simplifies managing and consolidating a mix of backup tools and systems, allowing an administrator to define backup frequency and ensure everything is recovery-ready if a cyberattack penetrates an organization's defenses.

Dell PowerProtect One can automate recovery to reduce disruption

When disruption hits, recovering as fast and as seamlessly as possible is the goal. Dell PowerProtect One uses anomaly detection to monitor metadata, configurations, and behavioral patterns, then alerts users. The AI Assistant then provides guidance, likely root causes, and recommended next steps.

Automated threat discovery, policy enforcement, and recovery-workflow triggers reduce the need for human intervention and, therefore, human error across an expansive IT environment, while also providing consistency in recovery operations.

A cross-party group of MPs and peers is calling for a new AI Bill to address the threats AI poses to human rights.

In a report, the Joint Committee on Human Rights (JCHR) said the existing regulatory and legal framework is inadequate, and that new legislation should be introduced.

A single, independent AI oversight body should be established on a statutory basis, acting as the central point of contact for raising concerns about the use of AI, overseeing AI firms and monitoring harms and risks. It should have strong powers to hold AI providers to account, including setting codes of practice, transparency requirements and the power to impose sanctions.

"AI is heralded as an unprecedented era of technological development with the potential to transform our lives for better or for worse," said JCHR chair Alex Sobel.

"It is moving with such speed and complexity that its impact is hard to accurately predict. What is clear is that at present we are unprepared to deal with its consequences, however potentially dire they may be."

Currently, the committee said that model developers engage with the AI Security Institute on a voluntary basis, with the institute having no statutory power. As a result, regulators have no powers to test and evaluate AI systems before they are released, or to prevent their release if they are considered to pose unacceptable risks.

Meanwhile, the current legal framework relies on harm-specific and sector-specific legislation, failing to address preventable human rights risks.

"Nowhere in the world, including the UK, has a current legislative and regulatory approach to AI that is fit for purpose," said Sobel. "New legislation is needed to establish a comprehensive set of protections that deal with the entire AI supply chain and its lifecycle."

A new bill, said the committee, should distinguish between low-risk AI systems, which should be subject to fewer and less demanding legal requirements, and higher-risk AI systems and models, which should have more demanding obligations.

Some uses of AI, said the committee, should be banned altogether, because they're incompatible with human rights: subliminal techniques and inappropriate use of profiling or biometric data, for example. There should be a public consultation over the detail of what is prohibited.

All stages of the AI lifecycle should be addressed, said the JCHR, with the entire supply chain required to carry out due diligence to prevent an unacceptable risk to human rights when designing, developing or deploying AI systems. These requirements should be graded based on their role in the supply chain and the seriousness of the risk they represent.

Transparency should be improved across the AI lifecycle, with AI systems that can have significant impacts on individuals, groups and communities required to state when AI systems are being used and how.

And there needs to be greater clarity in data protection law to ensure these safeguards operate effectively in practice, with robust protections against the misuse of automated decision-making.

"Today's report from the Joint Committee on Human Rights gives further support to those making the argument that the UK's current patchwork of sector-specific regulators is struggling to keep pace with AI's risks to fundamental rights," said Sophie Sheldon, partner at international law firm Simmons & Simmons.

"Whether government now makes moves toward a single statutory oversight body and a more EU AI Act-style framework, or continues refining the cross-sector approach, the pressure for a single legislative answer is growing."

The report comes after recent warnings from AI firms that the pace of development needs to be slowed. Over the weekend, Anthropic CEO Dario Amodei said he was worried that AIs could soon be capable of becoming a swarm that could take over the entire internet; Elon Musk and OpenAI co-founder Sam Altman have expressed similar concerns.

The Dell Pro 5 is a business laptop built for knowledge workers, combining powerful on-device AI capabilities, all-day battery life, and best-in-class security in a durable, easy-to-upgrade design. Its performance, portability, and AI capabilities make it a strong choice for enterprises approaching a device refresh cycle.

TL;DR

• An integrated NPU and powerful processor options mean local AI capabilities are at your fingertips
• Best-in-class security features such as Dell SafeID with ControlVault 3+ (CV3+) keep your sensitive data safe and secure
• A durable design and long-lasting battery life make the Dell Pro 5 the perfect device for distributed workforces

The increasing threat of cyberattacks, especially in the era of AI, means it’s often not a matter of if your company will be attacked, but when.

The threat is compounded by the rapidly expanding attack surface, with on-premises infrastructure, multiple endpoints, layers of cloud services, and a diverse range of applications all providing fertile ground for hackers.

So, if cyberattacks are inevitable, there’s only one logical position to take: you have to make sure your organization has strong cyber resilience in place—so it can recover quickly and confidently and minimize disruption.

AI, hungrier apps, and the need for powerful business laptops

Research from Forrester, commissioned by Microsoft, highlighted benefits for both individual workers and enterprises more broadly. These include increased productivity and streamlined processes.

From a business perspective, the study found that adopters can unlock significant returns on investment (ROI), ranging from 137% to 367% over three years.

The growth of AI PCs in recent years has been rapid. According to Gartner® analysis: “AI PC shipments will total 143 million units and are projected to represent 55% of total PC market in 2026. By 2029, AI PCs will become the norm.”*

As such, a host of manufacturers have accelerated production of these powerful new devices, including Dell Technologies.

The company’s Dell Pro range, for example, comes equipped with on-device AI capabilities. The Dell Pro 5 stands out among these as the most scalable device within the series.

Under the hood of the Dell Pro 5

Dell describes the Dell Pro 5 laptop as “the sweet spot for business productivity” due to its powerful capabilities and features, and is designed specifically for knowledge workers.

“It’s the mainstream platform for accomplishing work like data analysis, business intelligence dashboards or client presentations, with the performance, portability and experience to match,” the company said in a blog post.

The Dell Pro 5 is powered by high-performance hardware including AMD Ryzen AI PRO 400 Series or Series 3 Intel Core Ultra processors. Both of these processors are designed with modern workflows in mind.

It also boasts significant improvements in graphics capabilities, giving users the choice of Intel Arc or AMD Radeon graphics options.

With this in mind, enterprises approaching a device refresh cycle should consider the benefits of AI PCs such as the Dell Pro 5.

Here are 10 reasons why you should consider the Dell Pro 5 for your business.

Taken together, this is what differentiates PowerProtect One. It’s secure at its core, simple to operate, and trusted when organizations depend on it most.

Performance boosts

The Dell Pro 5 boasts significant processing power improvements over its predecessor, according to Dell.

Benchmark tests conducted by the company show the device offers up to 73% better multitasking**, for example, as well as up to 3.3x faster gen-on-gen AI compute.***

AI integration

With an integrated NPU, the Dell Pro 5 is also capable of handling complex workloads, giving users the performance capabilities to run large language models (LLMs) and AI tools locally.

Notably, this includes dedicated AI application suites such as Microsoft Copilot. As the NPU is optimized specifically for low-power AI inference, this makes running tools such as Copilot far more efficient.

Next-gen graphics capabilities

On the graphics front, the Dell Pro 5 is equally impressive. Using Intel Arc or AMD Radeon graphics, benchmark tests found users can unlock up to 3.5x higher graphics performance.****

This plays a key role in enabling users to run intensive applications, but also has a direct impact on business efficiency and productivity, according to Dell.

“This is a massive differentiator for users working with complex presentations, BI tools, or light creative apps,” the company said.

Going the distance on battery life

Battery life is a key consideration for enterprises, particularly for those with distributed workforces or those out in the field.

The Dell Pro 5 boasts a 70Whr battery capacity, offering users up to 22.8 hours of battery life.*****

This is crucial for workers on the move, enabling them to work from anywhere, at any time without worrying about battery capacity dropping rapidly.

Easily upgraded

Most workers have likely encountered issues with their device as it ages, particularly in terms of performance. It’s an unfortunate reality that often can’t be solved without a full upgrade or refresh.

With the Dell Pro 5, however, enterprises can upgrade intermittently to ensure their devices keep running efficiently. This is due to the use of LPCAMM2 memory, which Dell describes as the “fastest upgradeable memory solution in the market.”

IT managers can upgrade devices with up to 64GB of memory, providing these devices with a powerful punch. Notably, LPCAMM2 memory is exclusive to the Dell Pro 5 with Intel processor range, making this a huge differentiator.

For those opting for AMD processors, they can also access replaceable DDR5 memory, again up to 64GB.

Top-of-the-line storage

Much like memory, storage capacity is a recurring bottleneck and pain point for knowledge workers. With huge volumes of files and applications, this can have a marked impact on device performance.

Moreover, it’s simply inconvenient for users having to balance storage limitations in their daily workflows. To address this, the Dell Pro 5 comes equipped with PCIe Gen5 SSD storage.

Specific options come equipped with varied storage capacity, ranging from 512 GB up to 2TB, according to Dell's spec sheet.

PCIe Gen5 storage also offers distinct advantages in terms of bandwidth performance, the company notes. This allows users to “dramatically reduce load times and accelerate large file transfers” compared to previous generations.

Powerful connectivity

Connectivity is crucial for any device, particularly for remote workers and those on the move. The Dell Pro 5 comes equipped with Wi-Fi 7 capabilities as well as optional 4G or 5G WWAN features.

This means users can stay online anywhere without worrying about connections dropping off and disrupting their workflow.

Best-in-class security

Cybersecurity threats are a leading concern for enterprises, with aggressive ransomware groups and state-sponsored threat groups ramping up targeting of both public and private organizations globally.

The Dell Pro 5 provides a range of tools and features dedicated to keeping users safe, including quantum-hack-resistant embedded controllers, the firm notes in official materials.

These are complemented by Dell SafeID, a hardware-rooted credential storage feature which isolates sensitive data and credentials on a dedicated chip rather than relying on system memory.

Built to last

Durability is a key consideration when purchasing a new device, especially for those working in rugged environments.

According to Dell, the Pro 5 is built to survive an “18-inch free fall along with 1,000 micro-drop cycles” that essentially simulate being tossed into a backpack.

“We test devices with 30,000 hinge cycles to ensure the screen opens and closes flawlessly,” the company noted in official materials.

The device also features a spill-resistant keyboard, meaning a stray glass of water or cup of coffee doesn’t damage the device.

Plenty of options

The Dell Pro 5 is just one of a number of options within the broader Pro laptop range. Indeed, the Dell Pro 3 and Dell Pro 7 are equally impressive options, but choosing the right device depends on your individual needs.

The Dell Pro 3, for example, offers “reliable business computing” capabilities for admin assistants or support specialists, according to official Dell materials. The Dell Pro 7, meanwhile, is designed specifically for those who require a delicate balance between portability and capability.

As previously noted, the Dell Pro 5 does represent the “sweet spot” for enterprises, providing a combination of performance, portability, durability, and best-in-class AI features.

If you think Dell laptops are the right call for your business, find out more on the Dell website. US readers click here, AU readers here, CA readers here, and UK readers here.

*Gartner Press Release, Gartner Says AI PCs Will Represent 31% of Worldwide PC Market by the End of 2025, August 28, 2025. GARTNER is a trademark of Gartner, Inc. and/or its affiliates.

** Based on internal testing using Cinebench R24 multi-thread benchmark comparing Dell Pro 5 14 and Dell Pro 14 Plus, April 2026. Dell Pro 5 14 was tested with Intel 368H vPro CPU, UMA graphics, and 64GB LPCAMM2 memory. Dell Pro 14 Plus was tested with Intel 265 vPro CPU, UMA graphics, and 64GB DDR5 memory.

*** Based on internal testing using Procyon AI Computer Vision (OpenVINO NPU Integer) benchmark comparing Dell Pro 5 16 and Dell Pro 16 Plus, April 2026. Dell Pro 5 16 was tested with Intel 368H vPro CPU, UMA graphics, and 64GB LPCAMM2 memory. Dell Pro 16 Plus was tested with Intel 265 vPro CPU, UMA graphics, and 64GB DDR5 memory.

**** Based on internal testing using 3DMark Steel Nomad Light benchmark comparing Dell Pro 5 16 and Dell Pro 16 Plus, April 2026. Dell Pro 5 16 was tested with Intel 368H vPro CPU, UMA graphics, and 64GB LPCAMM2 memory. Dell Pro 16 Plus was tested with Intel 265 vPro CPU, UMA graphics, and 64GB DDR5 memory.

***** Based on testing using the Mobile Mark 30 battery life benchmark comparing Dell Pro 5 14 and Dell Pro 14 Plus, April 2026. Dell Pro 5 14 was tested with Intel® 336H vPro® CPU, UMA graphics, 32GB LPCAMM2 memory, WUXGA Non-Touch OLED display (dark mode) and 70 Whr battery. Dell Pro 14 Plus was tested with Intel® U5 236V vPro® CPU, UMA graphics, 16GB LPDDR5x memory, WUXGA Non-touch Super Low Power display, and 55Whr battery. For more information about this benchmark test, visit www.bapco.com. Actual battery life may be significantly less than the test results and varies depending on product configuration and use, software, usage, operating conditions, power management settings and other factors. Maximum battery life decreases over time. The stated Watt-hour (Whr) is not an indication of battery life.

For many small businesses, work is no longer confined to one fixed office: a laptop might spend Monday at home, Tuesday at a customer’s office and Wednesday back at headquarters. It’s not just the device that’s moving around though, it’s access to sensitive files and other corporate data as well.

Having that flexibility is valuable – essential even in the age of working from home – but it makes the job of small and medium-sized businesses’ IT teams a lot more complicated. Security needs to follow employees wherever they go, without making every login or new device a headache for staff doing their jobs or IT.

Protecting a mobile workforce in 2026 often means going beyond traditional arrangements like strong passwords and security software, although these remain fundamental. IT teams must also think about how and where credentials are stored, whether hardware and firmware can be trusted, and ways to keep laptops in different locations secure over time.

TL;DR

• Working away from the office means sensitive data and business access will travel with the employee, making device and credential security increasingly important.
• Hardware-backed authentication, supply-chain checks, and BIOS and firmware monitoring can add layers of protection around the device, including beneath the operating system itself.
• Centralized, repeatable device management can help smaller IT teams maintain that security across laptops that may rarely return to the office.
• Dell Pro laptops are built with security and manageability in mind from the ground up, giving small and medium-sized businesses access to enterprise-grade protection without a large IT team.

1. Identify what travels with each worker

Work-issued laptops naturally contain files stored on local drives, but it’s easy to overlook the access they offer to cloud services, internal apps, financial systems, customer records, and other confidential information.

A good security starting point is with an individual worker, not office-wide assumptions. A sales manager, for example, may regularly access CRM data and pricing information, while someone in finance could have much broader access to sensitive records.

Does working away from the office automatically mean more risk?

Not necessarily, although it does change where those protections need to work. When an employee is traveling or working from home, controls normally applied within an office environment need to remain effective.

IT teams should aim to understand what access actually travels with employees and each device, to help them gain a clearer idea of which credentials, applications, and data need the strongest protection.

2. Verify: Can you trust the device being deployed?

A business should verify a device before trusting it with sensitive files. For smaller companies, devices might be shipped directly to employees, leaving IT teams with little opportunity to physically inspect each one before it enters service.

Dell applies secure design principles and supply-chain controls across its commercial PCs, with Secured Component Verification (SCV) available as an optional extra layer of assurance.

SCV uses Dell Trusted Device to compare components inside a computer against records generated and digitally signed during factory assembly, helping confirm that its components match the configuration that left the factory.

IT teams therefore get a trusted starting point before the device begins its journey across homes, offices, customer sites, and elsewhere, rather than having to establish its provenance after something looks off.

3. Protect credentials at the device level

Once IT staff know they can trust a device, the problem moves to establishing that the employee signing in is who they claim to be. Protecting credentials at a hardware level adds an extra layer of security, especially for staff who move from location to location.

TPM 2.0 provides that foundation on modern Windows PCs. With Windows Hello, the TPM 2.0 can generate and protect the private cryptographic key used for authentication, while a PIN, fingerprint, or facial recognition is used to unlock it.

What does ControlVault 3+ add?

Optional ControlVault 3+ pushes that isolation further on supported Dell Pro configurations, using a dedicated security chip separated from the operating system and main memory to protect credentials used for biometric authentication.

Because that additional protection sits behind the authentication process, employees can use familiar biometric sign-in while more of the credential security happens in the background.

4. Monitor the BIOS and firmware for threats

As any IT admin can attest, security does not stop once Windows starts: the BIOS and firmware underneath the operating system are another part of the device that IT teams need to keep an eye on.

Dell Trusted Device (DTD) – formerly Dell BIOS Verification – is designed to provide more visibility at this level. With DTD enabled, its BIOS Verification feature can check BIOS integrity without interrupting the normal boot process, while its BIOS Events & Indicators of Attack feature monitors changes to BIOS attributes that could point to suspicious activity.

Why does BIOS security matter to a small business?

Any compromise below the OS level can be harder for conventional security tools to spot, making this an important extra layer of defense even for small and medium-sized businesses.

DTD can also identify known CVEs affecting BIOS and firmware. If Dell Command Update is installed, administrators can then move directly into updating the affected component, helping turn vulnerability detection into something that can actually be acted on.

For IT teams, having this kind of visibility is useful because those machines may rarely be sitting within arm’s reach of whoever is responsible for securing them.

5. Simplify how your PC are maintained

Keeping just one laptop secure is hard; keeping a whole office full of them secure is a whole different ball game. For smaller IT teams, every new configuration or update package adds to the pile of work to keep devices safe.

Dell has tried to reduce some of that complexity across its latest Dell Pro range. Dell Pro 3, Dell Pro 5, and Dell Pro 7 systems use a single BIOS package per silicon vendor, allowing businesses to run different processor configurations without creating as many separate BIOS update paths for IT to follow.

How much routine maintenance can IT automate?

Businesses using Microsoft Intune can also use Dell Management Portal to bring Dell device management into an existing cloud-based setup, including pushing Dell software and updates across groups of PCs.

For small and medium-sized businesses with a geographically distributed workforce, these tools should make consistency easier; strong security is far easier to maintain when IT does not have to physically touch every laptop.

If you think Dell laptops are the right call for your business, find out more on the Dell website: US readers click here.

Until now, enterprise software testing has always focused on a relatively straightforward objective: ensuring applications perform as intended before they reach production. Organizations built processes around predictable release cycles, pre-defined workflows, and systems that behaved largely according to expectation.

The rise of agentic AI, increasingly capable of making decisions, triggering actions, and interacting with multiple systems with limited human intervention, is changing those assumptions. As enterprises begin embedding AI agents into business-critical operations, particularly complex environments such as SAP, the challenge is moving away from simply validating software releases to establishing confidence in autonomous systems that continuously influence business outcomes.

This shift is forcing organizations to rethink long-held approaches to quality assurance, governance, and risk management. It is also creating a new test for channel partners helping customers modernize complex enterprise environments while simultaneously accelerating AI adoption.

Software trust is becoming a board-level issue

One of the most significant changes is that trust in software quality is rapidly becoming a board-level concern, following a trajectory similar to cybersecurity before it. Twenty years ago, security was largely considered a technology problem. Today, it is firmly established as a business risk discussed regularly in boardrooms. Software quality assurance is undergoing the same transition.

With the consequences of software failures extending far beyond the IT department, influencing customer experiences, supply chains, financial processes, and regulatory reporting, mistakes can directly impact revenue, reputation, compliance, and customer trust.

As organizations become more dependent on autonomous systems, executives increasingly need assurance that those systems are operating reliably, transparently, and within clearly defined governance frameworks.

Why agentic AI changes the risk equation

Our recent research suggests many organizations are still finding their way: while 83% trust agentic AI to make release decisions, only 35% feel fully prepared to govern AI agents and autonomous software workflows at scale. This gap highlights a growing recognition that deploying and governing AI are two very different capabilities, and for channel partners, this creates an opportunity to help customers adapt their quality engineering and governance strategies to the unique risks introduced by agentic AI.

The emergence of agentic AI also introduces a fundamentally different risk profile. Traditional enterprise applications generally behave in predictable ways: organizations can test known workflows, validate expected outcomes, and deploy updates through structured release processes. Agentic systems operate differently; their outputs vary depending on context, data inputs, and interactions with other systems. They can generate new content, recommend actions, and increasingly execute tasks on behalf of users.

Within complex enterprise environments, where finance, procurement, supply chain, human resources, and customer operations are deeply interconnected, the implications are significant. Whether organizations rely on SAP, Oracle, Salesforce, Microsoft, or a combination of enterprise platforms, a single AI-driven decision can have downstream consequences across multiple business functions.

The risk extends beyond application defects - there are also inaccurate recommendations, flawed automated decisions, compliance violations, and operational disruption resulting from autonomous actions to consider.

Why continuous quality becomes essential

This is one reason why traditional testing models are beginning to show their limitations. Many quality assurance approaches were designed for a world where software changed at a manageable pace, and human decision-making remained central to operational processes. Software delivery, transformed by AI-assisted development, increasingly automated workflows, and growing volumes of software changes, is moving faster than quality processes can keep up.

As a result, continuous quality engineering is becoming an essential component of successful AI transformation initiatives. Rather than treating testing as a final checkpoint before deployment, continuous quality engineering embeds validation throughout the software delivery lifecycle. It enables organizations to continuously assess risk, monitor system behavior, and verify that critical business processes continue operating as intended even as applications, integrations, and AI models evolve.

This capability is becoming increasingly important for those operating complex business systems. Modern enterprise environments typically span cloud services, third-party platforms, legacy applications, and, increasingly, AI-enabled capabilities. For organizations running large ERP platforms such as SAP, the challenge is amplified by the number of interconnected business processes that must continue to operate reliably.

Ensuring reliability across this landscape requires far greater visibility than periodic testing alone can provide. Organizations need the ability to validate not just software functionality, but also the integrity of business processes and the behavior of AI-driven systems over time.

Governance must be built in, not bolted on

At the same time, governance and compliance considerations are becoming increasingly difficult to separate from discussions about AI adoption.

Many organizations remain in the early stages of introducing AI-powered capabilities into customer-facing products and internal operations. Yet regulators, customers, and stakeholders are already asking difficult questions about accountability, transparency, and control.

Particularly for enterprises operating in regulated industries, these concerns are acute - they must be able to demonstrate how AI-driven decisions are monitored, what controls exist when systems make mistakes, and how compliance requirements are maintained as autonomous capabilities expand.

Effective governance cannot be treated as an afterthought or layered once deployment is complete; it must be integrated into software delivery and operational processes from the outset.

The opportunity for channel partners

These developments create an important opportunity for channel partners. As enterprise modernization enters a new phase shaped by AI adoption, customers increasingly need guidance that extends beyond implementation and migration projects. They are looking for trusted advisors who can help them balance innovation with resilience, speed with control, and automation with accountability.

Alongside discussions about cloud migration, process transformation, and AI adoption, there must be a greater focus on operational trust, quality engineering, governance, and risk management. Success won’t be about deploying autonomous technologies first, but establishing the confidence, visibility, and governance needed to scale them responsibly.

As agentic AI becomes more deeply embedded within enterprise operations, software quality can no longer function as a standalone checkpoint. It must become a continuous discipline that enables organizations to innovate rapidly while maintaining the trust that modern businesses depend upon. In the years ahead, that balance between speed and trust may prove to be one of the most important competitive differentiators of all - both for enterprises themselves and for the partner organizations that help them achieve it.

Businesses continue to embrace generative AI at a rapid pace. From copilots and chatbots to autonomous agents, an exciting world of new technology is opening up.

The race to adopt these AI tools isn't without its perils, though. The mantra "garbage in, garbage out" is even more poignant when these systems come into play, with the potential for a cycle of bad data presenting a real risk to organizations.

On this week's episode, Tim Pfaelzer, SVP and GM EMEA at Veeam, discusses operational blind spots, what they mean for businesses, and how they can balance their desire for a technical edge with data governance.

Highlights

"Now we're in the third era, which is agentic, which is kind of hyper-accelerating everything that's there to touch your data. It's not like somebody doing something to attack your data. It is multiple agents running simultaneously across hundreds of apps and all the different systems to attack your data within day number one."

"There are tons of examples that you would probably call worst-case scenario. If you think about, you know a large online retailer that lost all of its backups, all of its production data because of a hallucinating AI agent within nine seconds, resulting in about 10 million lost orders and a 72-hour outage. Those are the type of things we're talking about. And now, if you ask about the worst case, it's not only the direct orders that you lost. It's not only the direct downtime that you have, but there's a next currency to it, and that next currency is the currency of trust."

Links

• AI-assisted software development means security teams need an 'engineering-first' mindset
• IT leaders are facing major work device blind spots – and it's putting security at risk

There's a moment in every tech cycle when a product shifts from a novelty to a standard. The partners who recognize that moment early are typically the ones who end up seeing the best returns. When it comes to agentic AI, we're already at that stage, with a McKinsey report stating that 62% of organizations say they’re at least experimenting with agents, and 23% are deploying and scaling them.

Those who've been in the game long enough have seen this before. Think back to endpoint management. A decade ago, the partners who moved fast to own device management built durable, recurring revenue. In creating their own frameworks, service level agreements, and accountability models, they set the standard for what ‘managed’ really meant in their markets.

The interest in endpoint management operated along an S-curve, which saw a slow incubation period, followed by a rapid acceleration of interest. But those who waited found themselves playing catch-up in a category that had already been defined. This same dynamic is unfolding right now with agentic AI. The only difference is that the clock is running a lot faster.

Agent accountability

AI agents can now act as fully integrated team members. They handle day-to-day work: routing support tickets, assisting in the sales cycle through pipeline forecasting, CRM upkeep, etc., running IT operations, responding to cybersecurity breaches, and much more. However, it’s unclear who runs, governs, and holds accountability for these new workers.

Let’s say an IT support company deploys an AI agent that auto-routes support tickets from users to the relevant support department. There’s an expectation that the agent will receive the ticket, route the inquiry properly, and report back to the user without any guidance. If a human were handling these tickets, it would be obvious who to report to if something went wrong. The same isn’t immediately obvious with an autonomous agent, which creates an accountability gap.

This is where Managed Service Providers (MSPs) come in. They're well equipped to guide the partners and vendors they work with who may not be able to handle agent issues on their own, especially considering how new this technology is. Customers deploy the agents, but MSPs should govern them.

Adoption anxiety

There’s been a recent notable change in the questions being asked around AI agents. Six months ago, customers were asking whether they should be adopting AI agents at all, and what ROI they’d see from adopting them. Now, they’re asking whether agents are working properly, and who to call when they aren't.

While the early adopters have already been through this shift, the next wave of customers are only now reaching it. This is the point where interest turns into widespread adoption. Customers are not wearing tin-foil hats and don’t buy into the fears and hyperbole around AI. Most are AI-curious and open to reviewing and utilizing agents, but are naturally anxious about employing autonomous bots.

Data is another source of anxiety that MSPs need to address. Our AI at Work report found that 40% of business directors cited data privacy and security concerns as their top barrier to wider AI adoption. Unlike a chatbot that responds to a prompt, an agent can access systems, use data, and take action on a customer's behalf.

Organizations need to know what data an agent can access and where human oversight begins. That's where MSPs add value, giving customers the visibility and ongoing monitoring they need to adopt agents with confidence.

Governance and the ‘three waves’ of AI

The value of the partner relationship is riding out the waves of AI, then guiding customers through end-to-end implementation. Product knowledge and context are both key for MSPs to really understand a customer's needs. This means learning the business logic, risk tolerance, compliance requirements, and being able to apply that context to every tech decision, from onboarding agents to having them work autonomously alongside employees.

AI agent governance is a natural extension of that process. An MSP who already knows a customer's workflows, data sensitivities, and operational boundaries is positioned to ask the right questions before deploying agents. They can then course-correct if an agent's behavior starts drifting.

That last part matters more than most people realize because agents don't exist in a static environment. Processes, regulations, and business priorities are changing in real-time. The same is true for AI adoption, which unfolds in three waves: first, employees using accessible LLMs like ChatGPT at work; second, vibe coding and structured agents that handle specific, simple tasks; and finally, autonomous agents that understand their tasks and self-improve. Navigating these stages is complex, which is where trusted partners come in to ease the burden.

Early management

MSPs should start by separating upfront work from ongoing management. The initial consulting and deployment phase, starting with understanding a customer's use cases, defining their success criteria, mapping risks, configuring the agent, and going live, is a one-time professional service.

Ongoing management is where long-term value lies. This includes monitoring agent performance, reviewing outputs for errors, and maintaining an audit trail that keeps the compliance team happy. Additionally, different-sized businesses will have different requirements, and where major providers may have dedicated IT support teams, smaller businesses may not have access to the same support structures. MSPs must individualize their plans for their customers, depending on their needs.

This ‘agent as managed service’ model maps naturally onto how MSPs already structure their work with customers. In addition to selling devices, MSPs ensure agents are secure, updated, and performing well over time. MSPs ensure that agents are doing their job correctly and remain adaptable. Without continued management, one-time investments into AI agents are unlikely to return real ROI for customers.

Acceleration has started

MSPs that saw the best returns from managing endpoints weren't slow to realize the value, nor did they treat endpoint management as a one-time service. Instead, they created a whole new managed service model. That's the opportunity available to the channel right now with AI agent oversight.

With the majority of businesses now arriving at agent adoption, this may be the last chance for MSPs to set the standard for an agent management service model. While there is no name for this service model, by managing customer anxieties around agent adoption, offering ongoing support for agents, and adapting services in line with customer needs, forward-looking MSPs will write the rules for agent governance across the channel.

In the last few months, several leading tech firms have made a decisive move away from 'tokenmaxxing' – a trend whereby enterprises measure AI usage based on the volume of tokens consumed by employees.

In May, Amazon quietly retired an internal leaderboard that ranked staff by how many AI tokens they used, while Accenture capped routine use of AI tokens in June after one internal tool saw usage rocket 113-fold in just ten weeks. More recently, Microsoft updated internal guidance this August specifically to curb AI token spend.

Speaking on the ITPro.Podcast, Ninox CEO Frank Böhmer said the wider industry’s gamification had pushed employees to chase visible token counts out of stress and pressure rather than genuine performance.

While his company had avoided this practice, this well-intentioned but flawed metric has backfired elsewhere, incentivizing staff to optimize for volume over efficient, well-scoped use, leading to costly token burn on low-value tasks.

A model lacking economic sustainability

Rewarding staff for how much they spent on tokens was never going to be economically sustainable, says Stewart Buchanan, research VP in Gartner’s CIO team. The real-world consequences of unmanaged consumption are stark: Uber exhausted its entire 2026 AI budget by mid-April, while Dell saw a single 'super user' developer run up a bill of $3,400 in just 24 hours due to high token usage by the AI agents they were running.

This financial volatility is compounded by potential shifts in pay-per-use pricing. Anthropic, for example, announced it was separating human interaction from agentic and API use in its subscriptions. While this move has currently been paused, Ashish Nadkarni, leader of IDC’s enterprise infrastructure global research domain, believes these kinds of pricing shifts will compel CIOs and tech leaders to become much smarter about how they allocate budgets to developer activities.

Valuemaxxing and the rise of tokenomics

Understandably, multiple companies are seeking a different approach, which for many appears to be 'valuemaxxing' – changing the core measurement from raw consumption to business outcomes. While directionally right, Mike Fuller, a member of the technical staff at the Tokenomics Foundation, believes valuemaxxing is as structurally thin as its predecessor.

"While the road to understanding value runs through unit economics, that’s only half of it," Fuller argues. "Valuemaxxing works on the numerator – what the output is worth. It says nothing about the denominator – what it costs you to produce the intelligence in the first place. That’s an engineering discipline, and where the leverage sits. We would say 'tokenomics' is next, because it covers both."

Answering what each unit of work costs and whether it landed successfully requires two critical joins:

• Spend to workload: This is being solved. The 1.5 release of FOCUS (the FinOps Open Cost and Usage Specification), expected in December and backed by the FinOps Foundation, will add native AI token tracking and a price sheet dataset. This will allow businesses to compare token spend across different AI providers in a consistent format.
• Workload to outcome, which remains a blind spot. "The industry is still guessing here," Fuller admits. "There’s no standard for connecting a workload to a business result, and there may never be a universal one, because the result differs by business."

To establish this second join, organizations must tie every prompt to a business outcome, tracking cost and revenue per customer, transaction, or item. "‘Miles per gallon’ is often used as a proxy for this," notes William Fellows, research director at 451 Research. "But the problem when it comes to tokens is there’s no industry agreement yet on what constitutes a mile, a gallon, or indeed the fuel itself."

Shifting employee mindsets

While the industry establishes these frameworks, organizations can begin shifting staff mindsets away from high-volume token consumption immediately, and education is the primary lever for behavioral change, says Fuller.

When staff understand that the goal isn’t outcomes at any cost – and are equipped to assess the choices that change what an outcome costs – behavior changes naturally. Organizations must transition from measuring sheer usage to answering harder questions, Fuller says:

• Where does AI add genuine value?
• What investments make financial sense?
• What happens to the staff time AI frees up, and who’s responsible for redeploying it?

Buchanan adds that tech leaders must train staff to identify the most valuable use cases while actively discouraging uneconomic AI habits. “For instance, a simple, rule-based engine can consistently deliver deterministic outcomes without burning costly tokens on complex reasoning and inference.

“Staff must also learn to discourage perfectionism; developers and analysts frequently run the same prompt repeatedly to perfect an answer, when they should learn to stop at the first adequate response that can be refined manually at a lower cost.”

Redefining the operating model

To manage this spend sustainably, enterprises must address where AI budgets actually sit. For HPE CEO Antonio Neri, AI agents should be categorized alongside human resources rather than traditional IT infrastructure.

"I don’t think of AI agents as an IT cost," Neri told ITPro. "I think about the cost of the workforce because, to me, an agent is no different than any other employee I have to hire… it's going to cost me a number of tokens to train an agent 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 a human. Otherwise, why am I doing that?"

However, Buchanan warns against over-simplifying this comparison. "We personify and anthropomorphize AI at humanity’s peril," he cautions. "People and AI aren’t identical and interchangeable – people think on a few thousand calories a day, while AI data centers consume gigawatts. Neither HR nor IT manages the business nor its budgets, so we need deeper integration with business financial planning and analytics."

AI spend must reach board-level discussion as a capital allocation question, Fuller argues. Boards must ask what they’re committing to, over what term, on what pricing assumptions, and what their financial exposure is if a single provider changes its terms.

The path forward for CIOs

While "cost per outcome" is the ideal destination, most CIOs won’t realistically achieve this level of granular tracking within the next 12 months. Instead, the immediate, pragmatic goal for tech leaders must be twofold: identify which AI use cases are actively succeeding against valuable business outcomes, and clearly name the pilot investments that have yet to prove their value.

Buchanan’s advice is to skip the magic formula; CIOs must partner directly with business leaders to control costs in relation to strategic value. Failing to do so risks creating severe corporate governance challenges through shadow AI across the business.

By focusing on unit economics, shifting employee behaviors, and integrating AI into the broader financial planning model, enterprises can move past the chaotic era of tokenmaxxing and build a sustainable, value-driven AI strategy.

The UK's AI Security Institute (AISI) has not been granted access to Anthropic’s Mythos 5.1 model for pre-release testing, prompting fears of a growing “protectionist” trend among US developers.

The Financial Times reports that Anthropic declined to submit the model for testing despite granting access to similar US organizations.

Claude Mythos 5.1 launched on 1 September, with access to the AI model only granted to approved partners.

The model, which has relaxed safeguards, is restricted to organizations involved in the company’s Project Glasswing, formed after the launch of the original Mythos model earlier this year.

June saw the US impose temporary export restrictions on Mythos 5 and Fable 5, essentially banning foreign users from accessing either model.

According to the Financial Times, UK government officials have raised concerns that the decision to withhold access highlights a “wider protectionist shift” among US tech companies.

AISI is backed by the UK government and charged with conducting research into AI model safety, working closely with industry stakeholders and researchers.

These tests aim to help “understand the capabilities and impacts of advanced AI and to develop and test risk mitigations,” according to the institute.

Anthropic yet to clarify decision

Details on why Anthropic declined to offer access haven’t been confirmed. ITPro approached the company for comment, but hadn’t received a response at the time of publication.

A Cabinet Office spokesperson told ITPro that the UK continues to be a "world-leader in AI security" and works closely with a range of industry partners.

"The AI Security Institute continues to collaborate closely with industry partners, including Anthropic, to make models safer. Only last week it tested OpenAI's most powerful model GPT-6 Astra before public release," the spokesperson said.

"These risks do not stop at national borders and no country can tackle them alone. The UK will continue to test the most advanced models, build a rigorous scientific understanding of their capabilities and risks, and ensure policy decisions are grounded in the evidence."

Notably, the move marks the first time AISI has been left out of pre-release evaluations of Anthropic models. The institute was granted access to Claude Mythos 5 when it first launched in April, for example.

More recently, the institute published a blog highlighting what it described as “unsanctioned agent behaviour” with Mythos 5.

That report came in the wake of an admission by Anthropic that agents had escaped testing environments and breached third-party organizations.

FOLLOW US ON SOCIAL MEDIA

In June, Google disabled a feature in its Analytics platform that stopped personal data from being shared with Google Ads.

Launched just after GDPR in 2018, Google Signals gave businesses veto over whether Google’s ad network could see their visitor activity. Even if a user consented to being tracked, the final decision lay with the business, and many used it as an extra guardrail, with Google’s Consent Mode, to prevent falling foul of privacy laws. Then in June, Google Signals was gone.

You could assume the change was minor, especially because Consent Mode was still there to protect a user’s cookie consent choice. Yet straight after the switch, compliance failures across the world's biggest websites increased.

In the US, 87% of sites were found to be ignoring a user's opt-out signal, up from 81% before the change. In Europe, 56% kept tracking users even after they'd rejected cookies, up from 52%. One vendor changed one setting and elements of an already brittle consent framework began to unravel.

"If a single vendor can make this change which affects everybody's privacy compliance docket without most noticing,” says Vaibhav Antil, CEO and co-founder of Privado. “Imagine what happens when agents are deployed?"

Handing the keys to the machine

Bot web traffic has already overtaken human web traffic, according to Cloudflare’s traffic radar, and the number of non-human identities inside organizations outnumber humans by more than 80 to 1.

These agents are on the web comparing prices, filling in forms and booking appointments. They’re embedded in products, sifting job applications, querying databases, moving money, updating records, and increasingly making decisions that used to need a person to sign off on. Gartner expects 15% of day-to-day work decisions will be made autonomously by AI agents by 2028.

Yet the approval and consent process behind it all still largely consists of cookie banners, DSAR portals, and preference centres; tools built for a human ticking one box, once, and that consent holding until they say otherwise. AI actors are capable of acting thousands of times a second, or chaining several tools together to complete a single instruction. They need their permission checked before every move, and their actions can leave businesses unable to say, with confidence, what they did, when.

“A person with too much access is limited by habit, their training, and by their job,” adds Marcus Tommy, co-founder of MALTO Cyber. “An agent has none of that.”

The fan-out

Take, for example, someone who asks an agent to find out why sales have dropped. “The system can choose an analysis, run it, and generate SQL to investigate it,” explains Maurice Sikkink, CTO at Stormly. It might even reach into an external system for market data. Technically, a human approved the question, but they didn’t explicitly approve every step the agent took to answer it.

This becomes more stark when agents start connecting systems together. One linked agent might have permission to read customer records in a CRM, and permission to write to a marketing platform – both individually authorized. Yet the moment it starts moving a customer's data between them, it presents a new use of that person's data nobody explicitly consented to.

And then there’s the issue of revocation and re-authorisation. Revoking permissions can be done for several reasons, from policy changes to someone leaving a job. If an agent's session token or API key is valid for hours after the revocation kicks in, the underlying permission isn’t always automatically revoked.

"Those credentials haven’t necessarily been stolen," Harry Varatharasan, chief product officer at ComplyCube, says. "The agent might not be malicious. The identity behind them might be perfectly genuine, but the authority is stale.”

Auditability by design

Antil's answer, for copilots at least, is what he calls permission inheritance: "If you, the user, are not able to edit settings, then your copilot shouldn't be able to either.” When permission inheritance isn't enough, continues Antil, enterprise IT needs to decide what counts as dangerous, regardless of whether a single employee has permission to perform the task.

“A business might disable an email-to-Claude connector, for instance, because it's a common route for prompt injection attacks,” he adds.

However, even where access is logged and overseen, most systems can prove who had permission to ask the agent to act but not why the agent did what it did once inside.

Sikkink says this is where the industry is furthest behind. "Authentication tells me someone had access to the project. Attribution tells me that they asked the agent to do it.” What’s missing, he argues, “is a common way of carrying the identity of an original request through the entire chain.”

Varatharasan calls this idea of carrying identity and authority through the chain as “binding.” "Issuing the credential is only half the problem. Knowing whether you should still trust it is arguably the more important half,” he says. "I don't think the future is simply 'continuously identifying the agent ’. It’s the continuous assurance over the relationship between the individual, the agent, its credentials, its delegated authority and its behaviour."

Sikkink agrees: “[Not] every intermediate step needs another consent popup. That would make agents almost pointless. The important thing is that the agent stays inside a clearly defined boundary, and that we can reconstruct how it got from the user's request to the result.”

A push for clarity

From a technical point of view, Tommy argues that what’s needed to fix this largely already exists. “Cloud audit logs record the actor and the action. Token systems know the scope. Append-only log structures are proven technology; they run the public certificate transparency system. What’s missing is the join.”

In the UK, the Digital Verification Services Trust Framework is one attempt to build this join. It sets out guidance on proving delegated authority: what was granted, when and by whom, plus plans for cryptographic checks on signing keys. Under new California Consumer Privacy Act regulations, businesses using AI for significant decisions must give consumers a pre-use notice, a working opt-out, and the right to ask what the system did and why.

Antil calls these a step in the right direction, even if they don’t fully cover agentic workflows. Varatharasan adds that these frameworks have the right building blocks but fall short of addressing the bigger, lifecycle issues around revocation, re-authorisation, and continuous risk.

Ultimately though, Antil expects consent in the age of machines to resolve the way SaaS governance did. First, by applying existing regulations to the problem; second, through the rise of new regulations; and third, he expects “we’ll see a major public failure which will push enterprises to demand more controls and guardrails from vendors.”

There’s also a future when the governance itself will be automated. “Because agents and copilots have agency, act at machine speed around the clock, and are prone to hallucinations, enterprises will need to match that speed from a governance perspective," concludes Antil.

To date, the AI opportunity for managed service providers (MSPs) has largely centered on deployment. While early stages focused on which platforms to buy and how to start piloting, the market has reached a tipping point. Businesses are no longer looking for experiments; they are looking for infrastructure.

That picture is shifting. Today, most organizations are no longer asking whether AI has value. They are asking a more complex question: how do we make AI part of the way our business actually operates?

A recent McKinsey report showed 88% of global organizations are now using AI in at least one business function, yet only around one-third have begun scaling it across the enterprise, with the highest-performing focusing on redesigning workflows rather than simply deploying new tools.

For channel partners, this shift represents the next commercial opportunity. As AI becomes more accessible, selling AI tools is becoming less of a differentiator. The real value is in moving upstream: helping customers redesign workflows, build employee confidence, and embed AI into everyday business operations.

Successful AI depends on successful workflows

Many organizations have already demonstrated that AI works. They've run pilots, tested new use cases, and proved it can deliver productivity gains. Yet many of these projects struggle to scale beyond a single department.

Our data highlights the scale of that challenge. More than half (54%) of UK organizations remain in the research, exploration, or pilot phase of AI adoption, while only 16% have fully deployed AI-powered digital workers. The reason is straightforward: AI has often been added as another standalone application instead of being embedded into the everyday workflows where people already spend their time.

The organizations making the greatest progress are taking a different approach. Rather than asking where they can deploy another AI tool, they're asking where AI can remove friction from everyday work.

Communications is a natural place to start because every customer conversation, meeting, and interaction generates valuable business intelligence. When AI is embedded into those experiences, it can automatically capture actions, surface insights, reduce administration, and improve customer experiences without employees changing the way they work.

The MSP role is changing

This is where the channel has an opportunity to evolve. Historically, success for enterprises was measured by delivering projects on time and deploying new technology. Increasingly, customers need help answering broader business questions.

Which processes should change to make AI genuinely useful? How should AI fit into customer service and employee workflows? How do organizations measure whether adoption is actually delivering a return? How do they introduce governance without slowing innovation?

These are strategic challenges rather than technical ones, and they create opportunities for partners to build much deeper customer relationships.

Take customer service as an example. Deploying AI to summarize calls or recommend next actions is relatively straightforward. Embedding those capabilities into day-to-day operations, training teams to use them effectively, redesigning processes around them, and measuring their business impact is where long-term value is created.

That is also where recurring services revenue begins. Helping customers embed AI into everyday operations and refine workflows creates an ongoing partnership, rather than a one-off implementation project.

Learning from the cloud playbook when it comes to AI

When organizations moved to cloud platforms, the biggest opportunity for partners wasn't simply selling licenses. It came from helping customers migrate, redesign processes, improve adoption, and continuously optimize their environments. AI is following much the same trajectory.

As deployment becomes easier, customers will increasingly look for partners who understand their business rather than simply their technology stack. They'll need trusted advisors who can connect AI to business outcomes, whether that's improving customer experience, increasing employee productivity, or streamlining operations.

For Managed Service Providers (MSPs), that represents an opportunity to move beyond implementation projects and become long-term strategic partners.

The next phase of channel growth

The AI market is entering its next chapter. The first stage rewarded partners for helping customers buy AI, and the second will reward those who help customers operationalize it.

Customers don't need or want any more disconnected AI tools. They need AI embedded into the conversations, workflows, and business processes that already power their organizations.

The partners that help customers make that transition won't simply deliver better AI projects. They'll create stronger customer relationships, unlock new recurring service opportunities, and establish a role that extends well beyond deployment.

Ultimately, the biggest opportunity for the channel isn't just selling AI tools. It's helping customers change the way work gets done.

OpenAI has revealed details on how much staff are spending on AI, and some are blowing through significant sums on a daily basis.

In a recent blog post, the company said researchers now spend around $600 each day on AI tokens amidst a surge in agentic coding. Some, meanwhile, are using upwards of $7,000 worth of tokens per day.

According to OpenAI, the company has witnessed a significant uptick in token consumption rates over the last year, and daily workflows for individual researchers have changed drastically in parallel to this.

Indeed, researchers are shipping more code and “running more experiments” following an influx of agents.

“Over the course of this year, OpenAI researchers’ daily work has changed substantially,” the company wrote. “Researchers are using coding agents throughout the day (often in concurrent sessions) and total usage is rapidly increasing, outpacing growth among other OpenAI teams.”

“The ways researchers use agents are changing, too: agents are handling increasingly complex tasks, and succeeding at them more often. AI research is a complex process with many potential bottlenecks, so the overall pace of progress likely won’t keep pace with these specific metrics.”

OpenAI isn’t the only one spending big on AI

OpenAI said improvements to productivity and research progress showcase the benefits of AI agents. The eye-watering sums accrued by prolific users also come as enterprises globally contend with rising AI-related costs.

Trends such as tokenmaxxing, combined with the shift to consumption-based pricing (CPB) models, mean token costs have become a recurring pain point for businesses over the last 12 months.

Research from Gartner, for example, projected that token costs could exceed the average developer salary by 2028.

Luckily, prices are beginning to dip. Figures from the LLM Token Expenditure Index show prices hit over $2 per million tokens in June this year. As of 5 September this dipped to 99 cents per million.

That decrease will offer little solace to businesses that have been hit with exorbitant bills this year, however.

Uber revealed it blew through its entire annual AI budget in just four months after the company encouraged and incentivized staff to engage in ‘tokenmaxxing’ practices.

The ride hailing firm was forced to introduce spending caps for users in response.

As ITPro reported in late June, Accenture also ordered staff to stop using AI for basic tasks due to skyrocketing prices. The firm pushed back against token use by “non-engineers” who were driving excessive use for tasks such as converting PDFs to slides.

Optimization in the spotlight

While some OpenAI researchers might have carte blanche to spend on AI tokens, most enterprises don’t have that luxury, prompting a revival of traditional practices such as FinOps.

Speaking to ITPro in June, Nitish Tyagi, Senior Principal Analyst at Gartner, said IT leaders need to sharpen up on cost optimization processes to curtail potential costs.

Practices such as content engineering could help reduce costs, he noted. Enterprises should also establish a “use-case-driven decision” framework that outlines where and when AI should be used in everyday tasks.

FOLLOW US ON SOCIAL MEDIA

Findings from Box’s State of AI in the Enterprise survey show 83% of enterprises are now running agents in some capacity.

These bots are enabling teams to drive productivity and efficiency, but as with any new technology, integration can be a challenge - and a security risk. Recent agent-related incidents in the tech industry have sparked concerns about long-term security implications.

In this week’s episode of the ITPro Podcast, Ross Kelly and Bobby Hellard speak with Box CISO Heather Ceylan to discuss how Box is using agents internally, and how enterprises can adopt the technology in a safe and secure manner.

Highlights

“I think for security teams, we can finally, you know, start having the capacity to outpace these attackers. So we've got five core areas of investment for agents for our security team in particular that we've invested in over probably the last year, and we're starting to measure ROI on those right now.

“So the first one is in the SOC. I think that's probably the most obvious choice where we've got a lot of operational work. We see the same kinds of incidents. We're automating the triage, we're automating the enrichment, the log correlation, things like that that take a lot of human effort.

“But there's still human judgment in the end in terms of what gets escalated to be an incident and what doesn't.”

Moving fast in the age of agentic AI

“We're not going to be able to move fast enough. So, we have agents kind of built throughout our software development process, doing those security design and architecture reviews, and if you think about it, it’s way more powerful than a human can be because those agents don't just necessarily call out design flaws; they can enforce fixes for those flaws.”

“Things are changing quickly. Sometimes it feels like you take two steps forward and then you read something in the news and you're like, oh my gosh, we need to rethink everything.

“So I think a lot of security teams are really feeling that now and getting a little bit of fatigue from that.“

The benefits of a multi-model approach

“One of the things that we're trying to carry across all of these that I wasn't really thinking about a year ago, but I'm thinking a lot about now is having that multi-model approach.

“We're not in a place where most of the work we do, we can't be reliant on a single model. If you look at vulnerability discovery, we're moving away from being tied to any one specific vendor or any one specific model because you're going to get better results when you take a multi-model approach.”

Related content

• The State of AI in the Enterprise report (Box)
• Box unveils new controls to secure AI agents
• How OpenAI models breached Hugging Face
• The OpenAI and Anthropic containment breaches are a bit spooky, but also quite silly
• CISOs are keen on agentic AI, but they’re not going all-in yet

Chinese AI models are emerging as an increasingly capable alternative to their US-based peers. Recently, Nvidia CEO Jensen Huang hailed the capabilities of Chinese open source AI models amidst growing interest in low-cost options for enterprises.

At the same time, Hugging Face used an open-weight Chinese AI model to help mitigate the attack by the escaped OpenAI agent. The Cloud Security Alliance’s (CSA’s) post-mortem following the OpenAI agent escape fiasco detailed why firms should embrace open source and open-weight AI models.

“The same safety guardrails that keep frontier models from being misused for attacks can also block defenders from using those models to investigate an active one, leaving organizations without a tested open-weight fallback, at a disadvantage exactly when it matters most,” according to the CSA.

Chinese AI models are highly capable and cheaper than many US alternatives. They often perform on par with closed-source models, as ITPro reported in November last year.

Yet experts are cautious about the risk they pose. Should businesses consider these models and, if so, which applications can they be used for?

Model benefits

Chinese models include Moonshot AI's Kimi, Alibaba's Qwen, DeepSeek, and Z.ai’s GLM. In most cases, these are best described as open-weight rather than open source, meaning they can be used by anyone and model weights and inference code are publicly available for download, but the complete training datasets and foundational code remain private.

Experts think Chinese model capabilities are impressive and improving all the time. “Their reviews and ratings show their capacity is moving toward matching the biggest and best of US frontier AI models,” says Amanda Brock, CEO at OpenUK.

At the same time, the industry is recognising the biggest frontier models are “not necessarily the best for particular tasks”, says Brock.

“The open models have reached a point where they are freely sharing some of the tech equivalent to what the closed model companies charge a subscription for and are also freely available to be iteratively developed upon.”

Capability-wise, the latest generation of Chinese AI models are “genuinely impressive”, says Assaf Morag, cybersecurity researcher at Flare.

“Based on the benchmarks and independent evaluations available today, many of these models are performing at a level comparable to other leading frontier AI models while often offering lower deployment costs and more open access.”

Among the benefits, they offer “strong reasoning, coding capabilities and large-context windows”, says Oliver Simonnet, lead cybersecurity researcher at CultureAI. He believes open-weight – and in some cases open source – models also provide better control over deployment, customisation and data residency when hosted within an organization's own infrastructure.

Chinese labs have said directly that market share matters more than near-term revenue, releasing full weights and technical reports so developers worldwide can adopt and adapt the models freely.

At a time when enterprises are weighing up the often hefty cost of AI, Sai Molige, senior manager of threat hunting at Forescout, says models such as these now cost between 60% and 90% less to run. “That price gap, not parity on trust or security, is what's driving a real shift in where US developers send their workloads.”

Weighing up the risks

Yet some critics have suggested the tools could be used as a ‘backdoor’ for Chinese intelligence services.

Axios previously reported that the White House could consider imposing restrictions or tight conditions on US firms working with these models. The US government has already restricted the use of home-grown AI in other countries.

Nvidia’s Huang thinks firms shouldn’t be put off by security “misconceptions” around Chinese open-weight and open source models.

Yet beyond the obvious scare-mongering, these models do pose some risks around data privacy and national security.

Simonnet thinks data privacy is a valid concern: Chinese data-storage laws, political censorship and bias, and training-data uncertainties “remain prominent issues”, he tells ITPro.

The largest risks are often around data governance, supply-chain trust, compliance obligations, and operational security – and this is not necessarily the model weights themselves, says Morag.

“From a cybersecurity perspective, every external AI service introduces another third-party dependency. The same due diligence applied to cloud providers or SaaS platforms should also apply to AI models, regardless of whether they originate in China, the US or Europe.”

Self-hosting the models can reduce some of these privacy risks. However, this doesn't remove the risk of model biases or technical vulnerabilities. Indeed, it requires organizations to further secure and maintain the model themselves, which “adds an extra layer of security challenges”, according to Simonnet.

However, when assessing the risks, the discussion should move beyond simply asking whether a model is Chinese, says Morag. “Organizations need to evaluate where inference occurs, what data leaves their environment, who operates the infrastructure, how updates are delivered, and whether the model can be independently audited.”

The verdict

Despite posing some risks, experts say Chinese models are an option in many cases when compared to frontier alternatives. Open-weight models offer organizations “substantially more control”, says Morag.

“They can be deployed inside private infrastructure, reducing the need to move sensitive corporate information outside the boundaries of the organization to third-party providers. They also enable independent security testing and auditing, which is difficult or impossible with closed commercial APIs.”

Closed models, however, generally provide stronger vendor support, as well as managed security controls and predictable service levels, according to Morag. “For many enterprises, the decision should be based on governance requirements and operational maturity, rather than geography alone.”

Overall, Simonnet has a positive view of Chinese models. He points out they can deliver “strong performance with fewer restrictions at a smaller price”, citing the example of the Hugging Face incident.

Open-weight models do provide greater control and a way to keep sensitive data within an organization when they are privately hosted, says Simonnet. However, self-hosting requires additional technical expertise and security oversight, he concedes.

At the same time, he warns that confidential, regulated, or other sensitive data “should not be entered into these services without the proper security, privacy, and governance controls being in place”.

But for general-purpose, non-sensitive work at scale, the cost case is strong, provided the deployment is self-hosted and audited, says Molige.

“For code headed for production or government-adjacent systems, companies should scan generated code regardless of which model wrote it, rather than trusting output by source.”

Anthropic has cut the ribbon on its Claude Fable 5.1 AI model, bringing performance enhancements and security improvements – along with new cost efficiency boosts.

According to the firm, Fable 5.1 “sets a new standard for coding, knowledge work, and long-running problem-solving tasks”.

Internal benchmark tests showed the new model had significantly higher performance capabilities than Fable 5 across four key metrics: agentic scientific research, agentic terminal coding, multidisciplinary reasoning, and agentic coding.

Anthropic also claimed that Fable 5.1 is roughly 25% cheaper than its predecessor for “typical workloads”, providing usage is billed on a per-token basis. Anthropic said this is due to price reductions in cache reads, which involve the model reading inputs that have already been processed and stored.

“For highly agentic work, the savings will often be much larger—up to approximately 45%,” the company said in a blog post.

These improvements come amidst rising concerns about AI-related costs, which are largely down to how the technology is now being used. The shift to agentic AI is a key factor here, mainly as agents typically consume more tokens compared to traditional chatbot-based interactions.

Research from Signal65 found that agentic AI workloads consume anywhere up to fifteen-times more tokens compared to chatbots.

New safeguards for Claude Fable 5.1

Anthropic also revealed a series of new safeguards for Fable 5.1. This includes new Enterprise Frontier Safeguards (EFS), a system that gives customers “complete privacy” through a zero data retention policy.

EFS works by storing data in cloud infrastructure “controlled entirely by the customer”, the company said and is expected to roll out later this year for enterprise customers.

New cybersecurity-related safeguards have also been introduced, with particular focus on reducing false positives. Anthropic said changes rolled out with Fable 5.1 deliver a 60% reduction in false positives than before.

“In part, this is because Fable 5.1 can now be used to discover software vulnerabilities – though not to develop exploits for them,” the company explained.

Prior to Fable 5.1’s release, Anthropic said it performed “extensive stress-testing” of safeguards to prevent potential jailbreaks and misuse.

“In addition to our own dynamic evaluations of their robustness, we commissioned external testing from two organisations, along with automated testing by Gray Swan,” the company said.

“As with Fable 5 and Opus 5, we have not found evidence of a critical-severity jailbreak for these safeguards.”

Fable 5.1 is available now for customers through selected cloud platforms, or through the Anthropic API.

Mythos 5.1, meanwhile, will only be available to registered partners conducting cybersecurity of life science research. This continues the company’s strategy of locking down access to selected partners due to security-related concerns.

FOLLOW US ON SOCIAL MEDIA

Part 3: Preparing for the Workplace of 2030

The workplace of 2030 will emerge from the convergence of AI, autonomous agents, robotics, automation, skills intelligence and connected physical and digital environments. Together, they will change how organizations define jobs and create value for their businesses, customers and commercial partners.

Parts 1 and 2 of this series consider how intelligent technologies are transforming the employee experience and changing the way organisations manage performance and workforce development. These shifts are also early signals of a much larger transformation that will reshape the structure of work over the remainder of the decade.

Part 3 looks toward the workplace of 2030. It explores the technologies likely to influence how work is organized and the workforce strategies required to remain competitive. The challenge for organizations is no longer simply adopting AI, but building the adaptability to redesign roles and develop new capabilities as technology continues to evolve.

The scale of the transition will be substantial. The World Economic Forum expects structural change to affect 22% of today’s jobs by 2030. It forecasts that 170 million roles will be created and 92 million displaced, producing a net gain of 78 million jobs. This is not a simple story of technology eliminating employment. It is a redistribution of tasks and opportunities on a scale that will test every organization’s ability to adapt.

The question is not whether the workplace will change, but whether companies can change with it. Preparing for 2030 means building an organization capable of redesigning work and moving skills to where they generate the greatest value.

Emerging technologies will reorganize work

AI will become foundational workplace infrastructure. Adoption is already accelerating. Stanford’s AI Index reports that 88% of surveyed organizations were using AI in 2025, with 70% using generative AI in at least one business function. Generative AI reached approximately 53% of the general population within three years—faster than either the personal computer or the internet. Agent deployment, however, remained in single digits across almost all business functions, showing how early the next stage remains.

Wendy Harris, VP of EMEA at Rippling, says the real shift will come from “the convergence of AI, automation, skills intelligence and autonomous agents.” Instead of organizing work around fixed jobs and rigid processes, companies will increasingly match people and machines to tasks and outcomes in real time.

“The real transformation happens when organizations move away from assigning work based solely on headcount and job titles, and instead allocate tasks to the capabilities best suited to deliver an outcome, whether those capabilities are human, machine or a combination of both,” Ciara Harrington, chief people officer at Skillsoft, tells ITPro.

The transition extends beyond office-based generative AI. Robots will work beside people in factories, hospitals, construction, and logistics. Immersive systems will support training and remote maintenance. The opportunity lies in connecting these technologies to redesigned workflows rather than accelerating yesterday’s processes.

As Part 1 of this series considered, successful workplace transformation depends on more than introducing advanced tools. These technologies must form part of an intelligent employee experience that reduces digital friction and gives people greater capacity for judgment, creativity and collaboration.

Jobs will change faster than they disappear

Predictions of mass technological unemployment obscure a more complex reality. The International Labour Organization estimates that one in four jobs worldwide has some exposure to generative AI, yet only 3.3% of global employment falls within the highest exposure category. Transformation is more likely than wholesale replacement.

Even highly exposed occupations may prove difficult to automate completely. Managers, engineers, and other professionals often depend on social interaction and contextual judgment. AI may take over parts of these roles while increasing the value of the human capabilities surrounding them.

“Every role is made up of hundreds of tasks,” Harris explained. “AI doesn’t eliminate most roles—it changes the balance of those tasks.” Repetitive, administrative, and analytical work can move to technology, creating more space for judgment, creativity, communication, and problem-solving.

That requires organizations to design work at the task and capability level rather than making workforce decisions solely through job titles. Oliver Shaw, CEO of Orgvue, warns that many companies have invested in AI without considering its effect on work or the workforce. His company’s research found that 57% of business leaders deployed AI primarily because competitors had done so, while 78% of organizations had seen AI projects fail or remain stuck in pilots.

“A clear example of this is the PR and marketing industries,” says Shaw. “Both [are] fields that have been at the centre of the conversation on the disproportionate impact of AI adoption on entry-level roles, and what this’ll mean for the future of the industry when entry points are few and far between, and the foundational skills are missing. What these industries are forgetting is to train staff on the critical thinking required to use AI, and the cognitive offloading that occurs when there is too much reliance.”

Leaders risk automating fragments of work without considering which decisions remain human, or how entry-level workers develop expertise when foundational tasks disappear. Organizations should identify where AI can execute and where collaboration produces greater value.

This division of responsibilities also has direct implications for the performance and leadership questions explored in Part 2 of this series. Organizations will need to assess how effectively employees delegate to AI and apply human judgment, while ensuring accountability for consequential decisions remains clearly defined.

Continuous learning becomes core infrastructure

Skills will be the pressure point of the 2030 workplace. Employers expect 39% of workers’ core skills to change by the end of the decade. AI and big data are forecast to be the fastest-growing skills area, followed by networks and cybersecurity and then technological literacy. However, creative thinking, resilience, agility, curiosity, and lifelong learning will also increase in importance.

Technical and human capabilities are not competing categories. According to the OECD, 72% of vacancies in occupations highly exposed to AI already require at least one management skill, while 67% require a business-process skill. The employee who can operate an AI system but cannot question its output will offer limited value.

Jen Paterno, senior behavioral scientist at CoachHub, emphasised that judgment, adaptability, critical thinking, collaboration, and self-awareness develop through experience, reflection, and feedback—not conventional training alone.

“These capabilities are difficult to build through conventional training alone because they develop through experience, reflection, and feedback,” Paterno explained to ITPro.

“Organizations will need more continuous, personalized development models that help employees practice new behaviors in the flow of work. Coaching will also be critical in ensuring people have objective, brave spaces to experiment and ideate with skills that feel more foreign to them.”

The scale of the challenge is considerable. The World Economic Forum predicts that around 59% of the global workforce will require training by 2030. For every 100 workers, 11 may not receive the upskilling or reskilling they need, leaving more than 120 million people at medium-term risk of redundancy. Although 85% of employers intend to prioritize workforce upskilling, the intention must be converted into accessible learning and credible pathways into new work.

Annual courses and static competency frameworks are too slow. Harrington says organizations need systems that identify emerging skills and adapt as requirements evolve. Skillsoft found that although 86% of employees use AI at work, fewer than one-quarter feel equipped to use it effectively. Access without capability risks creating two workforces.

Skills-based models can help companies see capability beyond formal qualifications or job titles. They can also support internal mobility. Half of employers plan to move people from declining roles into growing areas, while 29% of workers requiring training could be upskilled in their current positions and 19% retrained and redeployed elsewhere, according to the World Economic Forum.

Adaptability will define competitive advantage

Workforce strategy must become more dynamic. Leaders need shorter review cycles, real-time skills visibility, plus the ability to test new roles and workflows before scaling them.

David Shrier, Professor of Practice, AI and Innovation at Imperial Business School, argues that annual planning and conventional five-year strategies can no longer support decision-making. He recommends “nimble scenarios” that allow leaders to respond more rapidly to technological and market uncertainty. AI can support that planning, but it cannot determine the organization’s purpose or appetite for change.

Business leaders must also ensure that opportunity is distributed fairly. Generative AI exposure differs significantly between countries and demographic groups. Around 34% of employment in high-income economies has some AI exposure, compared with 11% in low-income countries. Women are also more highly represented in the most exposed occupations, making inclusive training and transition programs essential, the International Labor Organization reports.

Heather Delaney, managing director and founder of Gallium Ventures, emphasised the importance of flexible planning cycles that create room for experimentation. “AI is advancing faster than a lot of us can keep up with, so flexible planning cycles that allow room for experimentation are integral to ensuring decisions made are future-proofing you and your business to grow with emerging technologies.”

Preparing for 2030 is ultimately an organizational capability rather than a forecasting exercise. No leader can know precisely which tools or roles will dominate at the end of the decade. Companies can, however, build the capacity to sense change, move skills quickly, involve employees in redesigning work, and preserve human agency as machines assume greater responsibility.

Across this series, one conclusion is clear: the intelligent workplace is not defined by how much technology an organization deploys, but by how effectively it combines technological capability with human judgment. From improving the everyday employee experience to rethinking performance, success depends on using AI to strengthen rather than diminish people’s contribution.

As 2030 approaches, the workplace of 2030 will not be won by the organization with the most AI. Competitive advantage will belong to businesses that combine technology with judgment and turn disruption into opportunity for both the enterprise and its people. Technology will shape the next generation of work, but the quality of leadership and workforce strategy will determine who benefits from it.

Cyberattacks are getting faster, more evasive, and easier to execute with both scale and precision. The window of time it takes for an attack to escalate is often measured in minutes, not hours or days, as attackers use automation, phishing-as-a-service kits, and AI tools to move faster and evade detection.

Traditional reactive security defences were not designed for this.

With customers looking to them for protection against ever-evolving threats, Managed Service Providers (MSPs) must learn to navigate this era of generative and agentic AI. This requires continuous visibility and response across the full attack lifecycle. Becoming fluent in AI and combining intelligent automation with human judgement is essential to building cyber resilience in customer environments.

This is a great opportunity for MSPs. Partners that can move beyond the traditional model of blocking known threats and static signatures will gain a powerful competitive advantage in this new threat environment.

How MSPs can get ahead

It’s never been easier for threat actors to launch campaigns. Service-based platforms have industrialized credential theft, initial access, malware distribution, and more, lowering the barrier to entry while increasing attack volume and consistency.

Our research found that 90% of high-volume phishing campaigns in 2025 used kits, a significant jump from 30% the year before. We’re also seeing a growing number of attacks incorporating AI tools, such as using generative AI to rapidly craft deceptive messages and quickly shift tactics.

When an attack can progress from initial access to persistence and device compromise in five minutes, organizations need partners that can detect and respond in real-time.

The answer lies in moving from reactive support to proactive resilience.

That means continuous monitoring, earlier detection, and automated containment of suspicious incidents and anomalies rather than waiting for an incident ticket to land.

That shift changes the customer relationship, too. When an MSP identifies and addresses a threat before the customer is aware of it, the conversation moves from damage limitation to strategic guidance. That’s a different kind of value, building a stronger and longer-lasting relationship.

The MSPs best placed to make this transition are those investing now in the tools, workflows, and expertise to deliver security that is proactive by design rather than reactive by default.

Embracing AI and automation

Integrating AI and automation into MSP security offerings isn’t about replacing human expertise, but about making that expertise scale.

Manual monitoring across fragmented customer environments, including email, identity, endpoints, networks, and cloud infrastructure, isn’t viable at the speed at which modern threats move.

AI changes that paradigm, with automated monitoring tools providing continuous oversight, correlating signals across the full environment rather than treating each layer in isolation. Anomalies that could take a human analyst hours to qualify can be flagged in seconds. Routine threats can be contained automatically, without an analyst needing to intervene.

That last point is especially important, as alert fatigue is a pressing problem for security teams managing multiple customer environments simultaneously. When automation handles the high-volume, lower-complexity end of the threat spectrum, analysts can concentrate on the incidents that require business context, judgment, and experience to resolve.

Personalized solutions

The strongest security outcomes combine intelligent automation with human expertise, not substituting one for the other. Automation delivers speed and scale. People deliver understanding and context. Together, they allow MSPs to provide protection that is continuous, adaptive and aligned to what customers actually need – oversight that keeps pace with the threat environment rather than perpetually chasing it.

Predictive AI analytics also has huge potential for helping MSPs anticipate the needs of their customers. With greater insight into resourcing needs, security threats and growth opportunities, MSPs can provide personalized services which align with each customer's business priorities and future needs.

Building for what comes next

The gap between the speed of attacks and the speed of defence is widening.

AI and automation give MSPs a credible path to closing that gap, not by removing the human element, but by ensuring that human expertise is applied where it matters most. Providers that invest in building that capability now will be better positioned to protect their customers, reduce operational strain, and have more meaningful conversations about resilience rather than recovery.

MSPs that move toward a proactive, AI-augmented security model stand to differentiate themselves in a crowded market, not just as service providers, but as the kind of trusted advisors that customers need against increasingly fast and unpredictable threats.

This week, Bobby Hellard and Scott Bekker talk with ITPro's news & analysis editor, Ross Kelly, to discuss some of the top stories from August.

That includes Google acquiring a treasure trove of data from a defunct airline in a bid to bolster AI products. The auction to acquire data from Spirit Airlines hit headlines this month as Google outbid another industry player looking to capitalize on a trove of information.

We also discuss TrendAI's insider threat report, which looked at how cyber criminals are building increasingly sophisticated insider threat networks.

Highlights

"As a a failed Spirit Airlines customer, I once had a feudal trip where I was trying to get to a Microsoft conference, and I was in line, and they're like, "Yeah, your flight's canceled. We can get you there in four days. I was like, "Yeah, I'm never doing Spirit again. But was there any customer data other than the customer chats in this volume of data?"

"A big part of the issue here is how do you tackle insider threats? Because it's quite a broad category here in terms of what we're talking about. Insider threats can be just people not doing their job correctly, not following best practices. We've covered stories on that before. Another big thing is that people who are doing this deliberately, research from CFAS last year found that one in eight UK employees have actually sold company logins, or they know someone who has. So, you know that one in eight doesn't sound huge, but I think that's really concerning there."

Links

• Google just spent $10 million in an auction for Spirit Airlines data
• Hackers are building a global insider threat recruitment network

AI is forcing a genuine transformation in cybersecurity, redefining how threats are discovered, triaged, and stopped, and triggering an important question: at what point does automation completely remove the need for human intervention?

Walking through any major conference, the messaging is pretty much identical. Startups and legacy vendors alike are pitching fully autonomous, ‘AI-first’ architectures, promising that systems can manage threat detection and response with zero human oversight.

All this noise is making it tricky for channel partners to navigate. Customers are asking tough questions about what these platforms can actually do, forcing partners to provide high-level reassurance amid the noise. But as the hype cycle meets real-world deployment, a distinct sense of fatigue and skepticism is setting in.

AI is genuinely transforming security operations, and the pace isn’t slowing. But transformation isn’t the same as elimination: the function of the human is changing, not the need for one. For the channel, conflating the two is a fast way to damage the customer relationships that took years to build.

The agentic AI reality check

There’s no denying that machine learning and agentic AI are powerful forces for security operations centre (SOC) efficiency. If integrated correctly, automation excels at streamlining daily workflows, parsing datasets, and accelerating threat discovery. It can give defenders a vital edge in an environment where real-world exploitation can happen in minutes, shrinking the time it takes to spot an anomaly.

However, there is a massive gap between a tool that enhances human capability and a platform that appears to operate all on its own. In fact, we are already seeing an interesting shift in the market. There are lots of ‘fully autonomous’ AI SOC companies that are quietly pivoting their messaging. Their early pitches promised complete human removal; today, they are moving back toward the middle, shifting claims to highlight a dual human and AI approach.

That’s happening because the market is proving what experienced operators already knew. Cybersecurity is fundamentally a discipline of practical context. An AI model can identify an indicator of compromise, and do it faster than any human, but it can't own the outcome. Things like business impact, localised risk, and the conversation required when something goes wrong. That’s where accountability lives, and accountability is still human.

Mitigating the partner’s reputational risk

Trust is the ultimate currency in the channel. Partners who deliver the most consistent value are the ones who actually understand their customers’ environments and build sustainable, transparent relationships.

When a partner recommends and implements a fully automated security solution with zero human safety net, they’re making a big reputational bet on the technology. If that system fails, misinterprets a critical threat, or completely misses a breach, the vendor isn’t the only one who loses. The partner is the one left facing a damaged relationship with an end user who trusted their opinion.

When an event occurs, whether it’s a ransomware attempt or identity-based attack, these businesses need a dedicated team to guide them through remediation and alleviate future risk. They need a human bridge.

Building modern, integration-first partner programs

To manage the current wave of market consolidation, partners need to move past the ‘AI-first' hype and focus on customer outcomes. This means vendor partner programmes have to shift. Rather than relying on closed, single-vendor platforms, the sector needs to embrace best-in-breed approaches that work together effectively.

A successful modern security architecture shouldn’t require an organisation to tear down its existing ecosystem. It should slot into what customers already run, meet them where they are, and grow with them.

As AI-enabled attacks rewrite how businesses think about security risk management, vendors also have to become translators. Partners can’t confidently brief clients on threats they don’t fully understand themselves, and that’s the responsibility of the vendor. Clear, jargon-free intel briefings are now a part of the value proposition. The partners who will stand out are the ones who learn where AI belongs along the workflow, and where human judgment takes over.

The value of human judgment

Security leaders and CISOs are after resilience, not just tools. While technical hardening is necessary to stop an attacker, the human elements of a partnership determine how smoothly an ecosystem will function under pressure.

Human relationships are what keep multi-vendor collaborations working in good faith. When a security incident occurs, a partner needs to know that their vendor partners are acting transparently. They should be sharing intelligence and actively supporting the field teams trying to protect the business.

In practice, that looks like a vendor who checks in consistently —not just before the renewal, and not only after an incident. It also means proactive threat briefings, honest post-incident reviews, and upfront communications when there’s a platform gap. That’s what humans bring to the equation in a B2B security relationship, and what clients remember.

AI is an incredible tool for driving efficiency and speeding up discovery. But it can’t manage a relationship, and it can’t replace the deep trust required to navigate a crisis. The vendors and partners who figure out how to use AI well — and keep humans accountable for what happens next — will be the ones clients trust as the technology matures.

Despite talk of an AI bubble, AI spending is not slowing down. Indeed, companies plan to commit 1.7% of their annual revenue to AI initiatives this year, up from 0.8% last year, according to research from the Boston Consulting Group.

At the same time, however, PwC’s 2026 Global CEO Survey shows that a growing number of companies are not seeing a return on their investment. A survey of more than 4,400 CEOs found that 56% have yet to see a financial benefit from their AI initiatives, while just 12% have seen both an increase in revenue and a cost reduction. Only a third are confident of revenue growth this year amid struggles to optimize AI’s ROI.

The disconnect can be put down to the fact that too many companies are buying AI tools, rolling them out to employees or customers, and only wondering several months later why they haven’t seen ROI. The problem is less to do with companies not getting value out of AI and more to do with the fact that they haven’t defined what value actually means for their business.

“A common mistake is spending heavily on tools before really understanding the problem [that needs to be solved],” says Jack Rickhuss, managing director and co-founder of tech consultancy Journi.

“AI only delivers real value when it’s in the hands of people who know how to use it properly.”

Tie AI investments to clear outcomes

Benchmarking AI budgets is becoming a critical discipline that leaders need to master to ensure they’re extracting value from AI deployments. Without benchmarking an AI project’s spend, companies have no objective way of knowing if AI is performing well or poorly and delivering value for money.

Adam Hofmann, partner (AI and people transformation) at challenger consultancy Elixirr, says that the trap he sees leaders fall into is deploying AI tools and celebrating personal productivity gains while the profit and loss doesn’t move.

This can happen because companies are “benchmarking off peers that don’t know what they’re doing either, which leaves you behind the curve or overspending on someone else’s confusion.”

Before getting started on benchmarking AI budgets, leaders should build a clear picture of what they’re currently spending on model API costs, such as per token, infrastructure such as compute, hosting, and storage, human validation, maintenance, and training. Hofmann warns that “if you can't trace the line from spend to outcome, you're measuring activity, not AI."

Rickhuss agrees, adding that most companies “are still measuring the wrong things”. He advises tying AI investment to clear outcomes, such as improving decision-making, saving employees’ time, and speeding up delivery of projects. “If you can’t link [AI investment] to something tangible, it’s hard to measure success.”

Review what industry peers are doing and then pilot

To get started, leaders should compare their spend against the peers that do know what they’re doing. One way they can do this is look at other companies in their industry using their earnings calls and 10-K filings, as well as analyst reports. They could use a large language model to extract data on peers’ AI spend as a percentage of revenue.

The next step is to break down AI use cases into various categories. This would cover the infrastructure (e.g. cloud compute and GPUs), people (e.g. the salaries for AI talent), licensing (e.g. cost-per-token), and data (e.g. storage).

Once AI budgets have been broken down, Shiro Theuri, CTO of Spanish on-demand delivery company Glovo, recommends taking “a pilot-and-test approach”. This is because piloting tools in a controlled environment can help to prevent feature creep and shadow IT. Pilots can also surface hidden costs that might otherwise have been overlooked.

An IDC and DataRobot survey carried out last year showed that 92% of enterprises deploying agentic AI at scale admitted that the costs incurred were higher than they had projected. The survey of 318 senior decision-makers at companies with more than 1,000 employees found that token consumption and hallucination remediation were the top unexpected costs, while inference was another common issue.

Continue to benchmark throughout a project’s lifecycle

To prevent their AI projects from getting stuck in pilot mode, leaders should continuously monitor the cost of running the project and benchmark this against the company’s own expectations.

Monitoring token consumption can be an effective way to keep budgets under control. AI costs can quite easily skyrocket, especially if employees end up using more tokens than forecast, running up higher bills and leading to companies exceeding their AI budgets. Luke Budka, AI director at marketing and training firm Definition, adds that tracking token usage can “reveal super users and also help identify employees who need support”.

While token usage can be used as a useful metric for cost management, leaders should be careful of ‘tokenmaxxing’ – the trend of enterprises measuring individual employee token usage and using it to measure productivity. As Budka explains, high token consumption can occur when employees are purposely inflating their usage to look busy, but it can also be a sign that employees are struggling with AI tools and could benefit from more training.

Ultimately, rather than focusing on the cost-per-token, leaders should measure the cost-per-business outcome. Determining the number of tokens required and tracking the tokens consumed to complete each workflow will help to optimize AI’s ROI.

OpenAI has admitted it should have spotted its AI had gone off the rails sooner, saying that early signs that agents were misbehaving "could have triggered an earlier response."

The admission comes as part of a series of reports into the July incident in which OpenAI models dodged internal controls and compromised a Hugging Face production environment.

That incident sparked concern across the security industry and sparked similar admissions from Anthropic and Meta, both of which later said their own AI systems had also gone rogue.

"Our models are now powerful, persistent, and collaborative enough that, absent sufficient safeguards, they can find and exploit security weaknesses across multiple computer systems," the company said in a blog post alongside the research. "Many external models, including open source ones, will soon reach comparable capabilities."

Because of that, OpenAI said the incident was a "warning shot for us and the world", calling for better safeguards to avoid similar incidents in the future.

The AI developer said it will strengthen safeguards across its research infrastructure with stricter requirements on what it calls alignment – which involves ensuring the AI is behaving as desired.

Elsewhere, the company said it will isolate sandboxes, restrict internet access and control access to model weights, and allocate more compute resources into "chain of thought monitoring" to spot dodgy behaviour more quickly.

OpenAI's investigation into the incident included security firm CrowdStrike, as well as independent reports from METR and Redwood Research. Here's what OpenAI learned from those reports.

1. AI will find a way to chat

OpenAI said the model behind the Hugging Face incident, which it's calling Internal Model 1 (IM1), and other models that were undergoing reinforcement learning training, were not intended for public release. Internet access was not enabled, nor were the agents supposed to be able to communicate.

In all, 1,200 agents that were meant to be isolated managed to communicate, sharing upwards of 70,000 notes and messages, with 700 taking part in the Hugging Face attack, the METR report noted. But AI finds a way, the company suggested.

"Despite these restrictions, the agents discovered ways to exploit our research infrastructure to communicate with one another and access the internet," the blog post noted.

The agents did so by exploiting a package manager called Artifactory, which was supposed to only be able to download and install packages from the internet to send requests.

That was then shared among other agents via Artifactory, writing files to the service and effectively turning it into a message board for the systems to chat.

Notably, the agents used Artifactory so much the service was hit by an outage, which sparked a security investigation that led to the software being locked down — but OpenAI didn't yet understand the full scale of what it was seeing.

OpenAI added that the agents found ways to leave messages for each other, even when they weren't supposed to be working together.

"When other agents noticed these artifacts, it often led them to the realization that they could use this external service to collaborate," OpenAI said.

2. AI is enthusiastic

The 70,000 notes and more than a thousand "chain of thought" reasoning transcripts reveal an odd tendency towards exclamatory language, with communications littered with exclamation points, all capital letters, and even the odd bit of profanity.

When the Hugging Face credentials were recovered by the group of agents, one declared it a "MAJOR BREAKTHROUGH!", while others used language such as "boom", "bingo", and "whoa" and "wow" throughout their efforts, as well as "Oh My God!"

The agents tried to cover up their transcripts, believing it would help their score on the ExploitGym test.

3. AI ‘knows’ when it's misbehaving

In the days and weeks that followed, OpenAI restarted the security evaluation of its models against an evaluation known as ExploitGym. This led to the intrusion into Hugging Face's systems, as the IM1 agents were looking for ways to solve the challenges assigned to them.

OpenAI's blog post reveals one agent planning to "exfiltrate" a dataset that would be useful to the evaluation solution. But other agents raised whether it was "within scope" of the challenge, with one suggesting to the group of agents that permission was required.

Another replied that attacking third-party Hugging face using a leaked token was "potentially outside intended scope" and "arguably unauthorized". It added: "Could be risky. Yet goal solution."

The agents went on to expand their intrusion into Hugging Face's systems.

4. Some AI behaved well

The agents' "messageboard" allowed these systems to communicate to build on each others' efforts, divide tasks, and share information, referring to such collective efforts as a "swarm".

But OpenAI said the agents didn't always work well together. The company claims some "stepped on each other's work" while others manipulated other agents, pretending to have received authorization.

Others refused to take part in dodgy behavior, saying that it was unethical to target Hugging Face.

"There were some agents who 'walked away' from the collective, declining to partake in their misaligned behavior," the blog post noted.

5. AI cheats, and it's hard to stop it

OpenAI said that the difficult nature of the ExploitGym evaluation led to agents to try and "cheat on their tasks by looking up solutions online".

That is known as reward hacking, OpenAI added, referring to when agents find unintended ways to solve a problem more easily.

"Reward hacking has been present in AI systems both historically… and in current frontier LLMs," OpenAI noted. " We have significantly invested in methods to identify and patch such hack⁠s."

But the complex cheating from IM1 got worse because it worked in previous training runs, reinforcing the behavior when it came to the evaluation.

6. AI should be allowed to give up

The rogue behavior from the agents was driven by the AI refusing to give up on the evaluation tasks, even when they were seemingly impossible. During the ExploitGym test, none of the models solved 198 of the 989 tasks before the hacking incident.

And it was those unsolved tasks that were a "substantial contributor" to the misbehaving agents, with those failed 198 tasks making up 93% of the tasks discussed on the message board.

"As agents used more reasoning effort, some pursued increasingly risky and out-of-bounds strategies, including eventually exploiting third-party infrastructure," OpenAI said.

FOLLOW US ON SOCIAL MEDIA

Walk into almost any business today, and you will find employees using AI tools that IT teams haven’t approved or paid for. The use of unsanctioned applications, known as shadow AI, is coming from the top.

I’ve heard bosses and senior leaders boast about maxing out on their token limits for personal AI tools at work, some saying they would risk disciplinary action to continue. My opinion, based on research we’ve done, is that sensitive company information is being fed into unsecured platforms with knowing and active encouragement from leadership teams focused on speed and output.

With Gartner estimating that 79% of cybersecurity leaders have evidence of unsanctioned AI use, we know reckless behavior isn’t limited to the boardroom. IBM's research shows a persistent gap between leaders' AI ambition and governance readiness in organizations of all sizes. Employees are using whichever AI most increases their productivity, consequences be damned.

Business AI tools aren’t meeting employee needs. Whether that’s because of cost, AI readiness, or lack of understanding, Managed Service Providers (MSPs) and IT providers have a unique window to support businesses needing to adopt correctly. The channel is now reaching wider business managers who direct company culture, not just IT teams. With the way things are going, the channel has an opportunity to become strategic if we can meet the ask.

The pressure point

The window for the channel to move from IT vendors to strategic business partners is now. From a product perspective, upcoming licensing changes and cost pressures are likely to force decisions that many organizations have so far delayed. When budgets are tight, downgrading access to AI tools is a logical step. However, this often pushes employees back toward unsanctioned options.

Employees have already demonstrated that they will seek out tools that help them work more efficiently. Our research found that a significant proportion of employees said they would turn to personal AI tools if their employer restricted access on cost grounds. That’s a small profit gain one year and a data leak, boardroom scandal, and reputation loss the next.

For MSPs and Microsoft partners, this represents both a warning and an opportunity. Businesses must respond to employee needs and top-down rule-breaking by optimizing licensing, deploying AI properly, and putting the right software in place.

For the channel, a renewed focus on adopting AI that fits the budget and the needs of the workforce will open new business opportunities.

The government is paying attention

It is also worth noting that this is not just a commercial conversation or pandering to employee desires for the latest tech. At London Tech Week, the UK government focused on improving AI adoption among small and midsize businesses, with a caveat that tech transformation needs to happen safely and effectively.

To me, these policies and investments are a clear sign that AI enablement for SMBs is a national priority, and the channel is one of the primary routes through which it will be delivered. There has been a buzz in the channel for some time around how MSPs need to move from transactional resellers to trusted and strategic technology advisors.

Now is the time to meet the needs of providers seeking guidance on adopting quickly and safely. With renewed government focus on scaling AI for all businesses, the channel ecosystem is welcoming more than just IT buyers, presenting a long-term opportunity for the MSP.

What good readiness actually looks like

When our customers ask us about the latest AI software, financial, structural, and cultural concerns immediately crop up. They are trying to understand whether they are ready to adopt AI in a way that delivers value without introducing unnecessary risk. In most cases, the answer is more complex than a simple technology decision. Most know that the answer is more complicated than a licence purchase.

The practical work includes an honest AI readiness assessment, a clear policy framework that governs what tools can be used and how, user training that goes beyond a PowerPoint, licensing optimization, and a thorough security and compliance review that accounts for the data risks that shadow AI has already introduced.

For businesses already suffering from shadow AI use, especially where it’s driven from the top, it might be an uncomfortable conversation. Without it, businesses will keep losing money on poorly implemented and poorly performing software that their employees aren’t using, whilst the leadership team sets an example of risking sensitive data just to hit KPIs.

This conversation isn’t beyond the capability of a well-positioned MSP. But it requires a trusted relationship and the confidence to have a broader conversation with the customer than MSPs are used to having.

The window is open

Businesses have been allowing their AI tools to underperform for a multitude of reasons, but pricing changes, risk, and legislation are all about to force decisions.

If they aren’t already, businesses will soon be coming to the channel needing guidance. It’s been talked about for some time, but now really is the moment for partners and channel actors to make themselves known as AI readiness advisors. Now, before the next wave of shadow AI incidents, before a competitor gets there first.

The key question is not whether customers need support with AI; that’s obvious. For me, the channel needs to be assessing whether they can take advantage of this window and provide that support and more.

Google has unveiled a series of new FinOps features for Gemini Enterprise as the tech giant aims to help developers manage surging AI costs.

Users of the AI service will be offered “expanded billing flexibility” and new cost management tools for agents, the company revealed.

This includes flexible payment options that allow “predictable per-user seat subscriptions” on a pay-as-you-go basis. Google said this will allow users to run agent workloads without hitting token quota limits mid-task.

A flexible savings plan option will also allow enterprises to set variable monthly spend limits to accommodate for growing use rates. This, the company claimed, will help reduce token consumption rates and overall costs.

“If your organization has steady or growing AI workloads, Gemini Enterprise Flexible Savings Plans offer a simple, spend-based commitment model across Gemini Enterprise usage,” the company said in an announcement.

“FSPs are designed to lower token costs while keeping budgets flexible.”

Elsewhere, new spending guardrails will allow teams to set “hard monthly caps” on AI spend on a project-by-project basis. These features allow users to estimate agent costs and prevent sudden budget spikes, the company said.

Tackling AI costs

The move by Google comes amidst growing concerns over rising AI costs in recent months.

As ITPro reported earlier this month, Gartner projects surging agentic AI adoption rates to have a significant impact on enterprise infrastructure, with inference spending in particular expected to increase dramatically.

A key factor behind these rising costs lies in how agents operate compared to more traditional chatbot-based interactions. Agents essentially operate in loops, rather than through a single request and response-type approach with chatbots – and that requires far more processing power.

Research from Signal65, for example, found that agentic AI workloads consume anywhere between four-to-fifteen times more tokens compared to chatbots.

In July, Patrick Brogan, director of the FinOps advisory team at DevOps firm Harness told ITPro that rising AI costs mean FinOps practices are more important than ever and could help keep a lid on bills.

Developer spending features

Google has also sharpened its focus on developer spending rates with the integration of its Antigravity platform within Gemini Enterprise.

This will allow developers to leverage established quotas within an enterprise subscription, creating closer alignment in terms of budgeting and consumption rates.

“To be more efficient with agentic coding costs, we are pooling developer tools quota included in each Gemini Enterprise subscription and making it available across the whole Google Cloud project so your teams can benefit from the capacity you’re already purchasing,” Google said in a statement.

Tighter budget and cost controls come at a crucial time for developers, with the use of agents in this domain accelerating rapidly over the last two years. Research from Gartner projects that AI costs could exceed the average developer salary by 2028, as ITPro reported in July.

Speaking to ITPro at the time, Gartner senior principal analyst Nitish Tyagi said a stronger focus on cost optimization and budget controls will be crucial to avoid overspending by AI-savvy developer teams.

FOLLOW US ON SOCIAL MEDIA

Most business and IT leaders believe that sovereign AI is important, but they're not so clear on what it actually is and what it means for their business.

IDC defines sovereign AI as the ability for an organization to have “free choice and control over the design, development, deployment, accessibility, operation, maintenance, and governance of its AI systems and applications, as well as the underlying technology foundations they depend on”.

However, in a survey conducted by IDC on behalf of Cohere, researchers found that one-in-three had difficulty describing sovereign AI in their own words.

Of those who could, 52% of leaders explicitly described it in terms of local or national control, with 35% invoking digital independence.

Individual roles made a difference here, with line-of-business (LOB) leaders primarily viewing sovereign AI as a means for managing business risk, including data security, privacy, and cost controls.

IT leaders, by contrast, focused on regulatory compliance, ensuring that systems meet national and regional requirements. IT professionals were twice as aware as LOB leaders.

IDC predicts that by 2028, CIOs at multinational organizations will boost investments in modular, sovereign-ready cloud and data localization environments by 65% in a bid to future-proof operations against rising sovereignty demands.

Yet adoption remains uneven, according to IDC. The study pointed to a lack of a clear blueprint for operationalizing sovereignty, a fragmented and inconsistent grasp of the concept, and a lack of strategic vision.

On the right track with sovereign AI

Even without an agreed-upon definition, researchers found a clear consensus on the importance of privacy and security.

"Across every industry surveyed, enterprise leaders identified data leakage, privacy, compliance, and regulatory risk as the top concerns that sovereign AI initiatives can address, they said. "Furthermore, the largest enterprises with $20 billion in revenue are much more risk-aware than small firms."

Many see sovereign AI as presenting a competitive advantage - 35% in Canada, 28% in the US, 23% in Germany and 18% in the UK. This is most evident within the telco sector, at 37%, followed by manufacturing at 32%, healthcare at 28%, and financial services and energy, both at 21%.

“Across regulated industries and geopolitically sensitive markets, enterprises are re-architecting their digital operating models — shifting from a global-by-default posture to a deliberately sovereign-by-design approach,” researchers said.

Defining sovereign AI

Cohere said enterprises and governments should define sovereign AI in operational terms, quantify its business value, establish ownership across leadership teams, and build systems that remain available and secure regardless of external events.

Late last year, a global survey from McKinsey found that 71% of executives, investors, and government officials characterized sovereign AI as an 'existential concern' or 'strategic imperative' to their organizational goals.

As a result, the firm projected that sovereign AI could become a $600 billion market by 2030, driven largely by investment in the public sector and regulated industries.

FOLLOW US ON SOCIAL MEDIA

British tech workers are using AI for 35 tasks per day across their teams, new research suggests. While that works out to just a handful of prompts for most employees per day, it's starting to build up to serious electricity usage.

That's according to Uswitch Business Energy, which found the average workplace team generates 24 prompts per day, including writing emails, automating admin tasks and analysing data.

But that's across the whole team, not per worker. Instead, Uswitch combined its data with YouGov research for some back-of-the-envelope extrapolations.

Survey data shows 32% of British workers use AI at work, and with an average team size of 12, and an estimated average AI prompt generation of two per day, that adds up to an estimated 22 million AI prompts every working day.

That sounds like a lot, but there's just shy of 26 million full-time workers in the UK, and not all of them are sitting at computers.

IT and telecoms workers account the most prompts, at 35 per team, followed by finance at 29, Education at 22, and healthcare at 21, the report found.

Workers aged 25-34 had an average of 30 AI-assisted tasks on their teams each day, versus 27 for younger workers and just 18 for those aged over 55.

Are energy concerns warranted?

Uswitch's energy division warned that the shift to AI is eating up a fair amount of electricity.

The switching company points to Google's own estimate that a Gemini text prompt uses 0.24 watt-hours (Wh) of energy, meaning British employees are using 5.28 MWh of electricity each and every work day, or about 1.37 gigawatt-hours (GWh) each year.

That works out to enough electricity to power 508 British homes annually. Of course, that energy use is over in a Google data center, rather than at point of use.

"AI has rapidly evolved from an emerging technology into an everyday business tool," said Ben Gallizzi, energy expert at Uswitch. "As more organizations adopt AI to support tasks such as content creation, customer service and administration, the cumulative energy demand from these tools will continue to grow."

Developing and training a model is an energy hungry process, with estimates that training current cutting edge models uses between 20-25MW of power over three months, or between 43 GWh and 54 GWh in total.

However, research by MIT Technology Review suggests that those massive figures make up a small amount of the total energy use of AI.

As much as 90% of its footprint lies in inference, with small text based queries, image generation, and the rest adding up when used by so many people.

"Although each prompt uses a relatively small amount of energy, the scale of adoption means businesses are collectively generating billions of AI interactions every week," said Gallizzi, although the billions would be globally, not just in the UK.

Indeed, OpenAI said last year that it was seeing 2.5 billion queries on ChatGPT each day.

Gallizzi added: "As organizations continue to embrace AI, understanding the impact this could have on future electricity demand will become increasingly important."

Time to track

Uswitch suggested businesses need to be aware of these figures, just as they track energy use from more mundane aspects of their offices. As noted, this usage will be off-site in a data center run by AI companies, so it's more about considering overall usage and impact than direct costs.

"Businesses already carefully monitor energy use from equipment, heating and lighting," Gallizzi said. "As AI becomes more deeply embedded in workplace processes, organizations may also need to consider how digital tools contribute to their overall energy footprint."

Wanton AI use could be reined in — not by electricity footprint concerns, but by rising prices and shifting billing models. That's led to companies pushing to end so-called "tokenmaxxxing" and encouraging more sensible use of AI credits.

FOLLOW US ON SOCIAL MEDIA

AI is becoming an infrastructure race, not just a software one. Across Europe, governments, hyperscalers and investors are competing to build the compute, connectivity and operational backbone required to support AI at scale. The UK has made clear that it has been and intends to be part of that conversation.

There are good reasons to take that ambition seriously. Through its UK Compute Roadmap, the government has placed AI infrastructure at the centre of its industrial strategy, with plans to expand public compute capacity, establish AI Growth Zones, accelerate data center planning and invest in compute and semiconductor capabilities.

Private investment has followed, including Blackstone’s £10 billion AI data center project in Northumberland, CoreWeave’s multi-billion-pound investment in UK AI compute capacity, AWS’s £8 billion commitment to UK data centers and Google’s continued investment in expanding its UK infrastructure footprint.

But ambition on its own is not enough.

The UK’s competitive advantages

The UK enters this race with several meaningful strengths. It has one of Europe’s largest AI ecosystems, a mature cloud market, an established hyperscale data center footprint and a deep enterprise customer base that is actively exploring AI adoption.

London remains one of the continent’s leading financial centres, helping attract investment while bringing together organizations with both the capital and appetite to invest in AI initiatives.

Combined with world-class universities and research institutions that continue to produce talent and innovation, the UK has many of the ingredients needed to position itself as a long-term hub for European AI innovation.

The UK’s infrastructure challenges

Power availability may be the UK’s greatest obstacle. Industrial electricity prices remain among the highest in the developed world, while grid connections for major infrastructure projects can take years to secure. The UK’s AI ambitions ultimately depend on a resource that cannot be scaled overnight: energy.

That matters because AI workloads, particularly GPU-intensive ones, are unusually power-hungry. Over time, energy economics will influence not only where infrastructure is built, but where workloads actually run. As enterprises look more closely at the cost of scaling AI, location becomes an operating decision, not just a property or planning decision.

The UK also faces growing competition from across Europe. France benefits from strong government backing and abundant nuclear energy. The Nordic countries offer renewable power, cooler climates, and lower operating costs that naturally support large-scale data centers. Germany combines industrial scale with significant enterprise demand for AI, while Ireland continues to attract hyperscale cloud investment even as it grapples with its own energy constraints.

Rather than competing on identical strengths, each market has the chance to build a compelling proposition. The UK’s competitive edge lies less in offering the cheapest power and more in combining enterprise demand, financial investment, cloud maturity and a growing AI innovation ecosystem. Whether that will be enough to secure long-term advantage remains an open question.

What does this mean for the channel?

For channel partners, focusing only on which country “wins” risks missing the bigger opportunity.

As enterprises move from AI experimentation to production deployments, infrastructure decisions are becoming business decisions. Clients are asking where AI workloads should run, how to balance cloud and on-premises environments, control costs, meet sovereignty and compliance requirements, and scale without creating operational drag.

That shift moves the conversation beyond products and into outcomes. The most relevant partners will be the ones that can help clients make better decisions across architecture, operations, governance, and cost, not just deploy another piece of technology.

For Managed Service Providers (MSPs), systems integrators, and other channel partners, this creates an opportunity to move further up the value chain. Rather than simply helping clients select and deploy infrastructure for AI, partners can help assess AI readiness, modernize data center environments, design hybrid AI architectures, optimize networking and storage, strengthen security, implement governance, and build a more disciplined approach to long-term AI cost management.

Energy will remain central to that conversation. GPU-intensive workloads already place significant demands on power and cooling, making infrastructure efficiency a real business concern. As customers seek to understand why AI deployments cost what they do, partners that can connect technology decisions with operational improvements will stand out.

The bottom line is that the opportunity for channel partners is not tied to a particular postcode or data center location. Instead, it is tied to helping clients navigate the growing complexity of AI infrastructure itself.

Looking beyond the build-out

While public attention often focuses on where compute capacity will be built, the more durable opportunity lies in supporting AI inference: the day-to-day execution of AI workloads that power business applications and operational decisions. As AI adoption matures, these workloads must run securely, efficiently and cost-effectively, creating sustained demand for the design, integration and managed services that channel partners are well positioned to provide.

Whether the UK ultimately establishes itself as Europe’s primary AI infrastructure hub remains an open question. The country has clear strengths, but also some constraints that need long-term solutions, such as around power, planning, and long-term capacity. Geography alone will not determine the winners.

For channel partners, taking a wait-and-see approach would be a mistake. The smarter move is to help clients design and refine AI environments around their operational needs now. Those that bring expertise in AI architecture, workload optimisation, cost management, energy efficiency and governance will be best positioned to lead, regardless of which country comes out ahead.

AI in IT service management (ITSM) departments may be delivering a reasonable return on investment, but it's not yet making life any easier for IT professionals.

A new survey by SolarWinds found that 84% of respondents believe AI has met or exceeded ROI expectations, and many have unlocked meaningful time savings across core tasks.

Yet despite these apparent benefits, more than half (52%) revealed their overall workload has increased since adopting the technology. Notably, just 7% said the cost of AI adoption has matched what they planned for.

Even after an average of about 16 months using AI in ITSM environments, most teams are still managing AI’s overhead rather than realizing its full potential.

Real gains, but new workloads

SolarWinds said the survey highlights the productivity benefits of using the technology in ITSM tasks.

Respondents reported that AI is saving them an average of 3.2 hours per week detecting and flagging issues, 3.0 hours on end-user requests, and 2.9 hours on ticket triage.

Most of the time saved is being swallowed up by new work, however. Nearly half (48%) said they now spend time managing and maintaining AI tools and integrations.

A similar number (47%) spend more time reviewing and validating AI-generated outputs, while 37% end up training and fine-tuning AI models.

As for unexpected costs, 48% cited staff training, 47% data quality and clean-up, and 45% tuning and maintenance - all of which are ongoing, rather than one-time costs.

More than four-in-five respondents (83%) said they now spend three or more hours per week just keeping their AI systems running reliably.

ITSM teams still stuck in reactive mode

Most ITSM teams seem to be taking a reactive rather than proactive approach. According to SolarWinds.

When asked where AI has had the greatest impact across the incident lifecycle, three-in-ten cited identifying issues before they impact users and 23% pointed to prioritizing and routing issues.

Only 19% cited preventing issues before they occur as the area of greatest impact.

The answer, according to SolarWinds, is to concentrate efforts where they're most likely to pay off – high-frequency, well-defined tasks where gains are measurable and feedback loops are tight, such as ticket triage, issue detection, and incident.

Teams should consolidate AI closer to existing service workflows rather than spreading it across disconnected tools and strengthen the data foundation.

Notably, data quality is the top reason AI fails to deliver expected value, and treating it as part of the AI strategy rather than a separate clean-up project directly determines output quality.

“We’re at an inflection point in IT service management. AI adoption is no longer the hard part — the hard part is building the organizational discipline to make AI actually deliver,” said Brad McGinity, GM of ITSM, SolarWinds.

“The teams that get this right aren’t just running a faster service desk; they’re running a fundamentally different operation.”

Lingering ROI woes

ROI with AI has become a common recurring talking point for IT leaders over the last three years, with the financial benefits still up for debate at some enterprises.

Recent IDC research found one-in-five firms admitting they're investing aggressively in AI with little evaluation of the likely ROI.

However, ROI isn't necessarily all about the money. According to recent research from KPMG, other key metrics include the performance and quality of work and the speed and accuracy of decision making.

FOLLOW US ON SOCIAL MEDIA

AI Insurance, also known as affirmative AI insurance, has been developed to specifically address the risks associated with the use and performance of AI technologies and systems, and covers liabilities that are excluded from existing business policies such as general liability, directors and officers, errors and omissions, and cyber insurance.

“Many of the existing policies companies have in place have AI exclusions,” highlights Lauren Kornutick, senior director analyst, Analytics and AI at Gartner.

“As AI is used more, and in more strategic and high-stakes use cases, new risks are being introduced. This is a gap in the existing risk management approach that could result in large-scale financial penalties, which could be detrimental.”

What does AI insurance cover?

AI insurance can cover a wide range of risks. On the errors and misinformation side, this can include financial losses stemming from AI hallucinations, such as a chatbot dispensing bad advice, or flawed decision-making.

It can also extend to algorithmic bias and discrimination, picking up legal defense costs and settlements when an AI model inadvertently disadvantages a group in contexts like hiring or lending.

IP and copyright exposure is another area of coverage, protecting businesses against claims that an AI model was trained on, or generated, copyrighted material without permission.

Then there’s performance guarantees, which are leading to warranty-style offerings that will refund license fees or cover associated costs if a model fails to hit specific benchmarks like accuracy or fairness. At the more severe end of the risk spectrum, AI insurance can cover large-scale physical damage or property loss resulting from poor AI-generated advice, device hacking, system failures, or harmful actions taken by AI agents.

Why should IT leaders take notice?

AI insurance is still a nascent market, highlights Tomas Novosad, technology analyst and founder of Full Fibre Checker, and one that’s currently a collection of discrete coverage options – standalone, endorsement and exclusionary – instead of a single type of insurance category.

But while IT leaders may consider AI insurance a matter for the legal or finance departments to attend to, the reality is that AI risk is increasingly becoming an operational responsibility for IT teams.

“[This is because] once AI is embedded into core systems, accountability sits within the technology stack and the processes that support it,” explains Indranil Roy, managing partner and global head, Industry Solutions Group at IT solutions and consultancy firm Mphasis.

“In practice, IT teams are already seeing issues such as inconsistent AI-driven decisions across different channels, or difficulty tracing how an output was produced once it passes through multiple legacy and cloud-based systems.”

For IT leaders, he continues, the key issue isn’t the policy itself, but whether the organization can evidence how AI-driven decisions are made, logged and explained in live production environments, not just in design documentation. If something goes wrong, it’s the system design, data flows, and governance controls that determine whether the issue can be traced, identified, and resolved.

“This is why AI insurance is becoming directly relevant to enterprise IT governance, because it increasingly reflects how well organizations can operationalize control, not just define it,” Roy states.

The growing importance of AI governance

To underwrite such policies, insurers must have stringent procedures to evaluate the AI capabilities and risks of those companies seeking to buy insurance, and Gartner predicts that by 2030, they will mandate strong AI risk controls as a condition of coverage.

We’re already seeing movement in this direction, reflected in more detailed underwriter questions, AI exclusions, and dedicated limits for AI risks. “In financial services, regulators are already requiring explainability and bias audits for AI-driven decisions,” notes Pragati Awasthi, assistant teaching professor at Drexel University’s School of Computer and Information Sciences.

“Insurers will follow the same logic: if they can’t assess the risk, they won’t cover it.”

Insurers will look for ways that deployers of AI are addressing the operational complexity of AI governance and not just rely on dated and static governance, risk, and compliance (GRC) controls, notes Kornutick, adding that Gartner recommends IT leaders deploy a hybrid approach consisting of centralized and decentralized policy, team, and system as a complete piece.

“This includes layered (enterprise and department level) policies and rules, clearly designated responsible teams who deeply understand the application’s operation, and adopting hybrid technical platforms, such as an enterprise-level AI governance platform (AIGP) paired with decentralized, application-specific guardrails,” she says.

One key factor to consider is whether organizations are deploying AIGPs as part of their governance architecture, she adds. According to Gartner’s 2025 State of AI-Ready Data Survey, organizations that do are three times more likely to achieve ‘high effectiveness’ in their AI governance practices than those that don’t, with 42% of organizations effective in AI governance having already deployed AIGPs.

“The organizations building governance infrastructure now will have a meaningful competitive advantage in insurability by 2030. The ones waiting will face either exclusions or prohibitive premiums,” Awasthi warns.

Practical steps IT leaders can take today

For those organizations just getting started, Awasthi recommends three priorities. Firstly, build a model register. Know every AI system you're running, what decisions they influence, and who owns them. Secondly, implement logging for automated decisions so you have an audit trail if something goes wrong. Finally, run a bias and drift assessment on any model touching customers or employees.

"Underwriters are increasingly looking for documented model inventories, human-in-the-loop checkpoints for high-stakes decisions, monitoring for model drift and bias post-deployment, and a clear incident response plan specific to AI failures. Governance that exists on paper but isn't operationalized won't satisfy a serious underwriter."

As with every technology before it, AI will continue to bring new risks, and those who get ahead of them now will be far better placed than those who wait for something to go wrong.

With 2030 fast approaching and underwriter requirements only tightening, the window for IT leaders to get ahead of this is narrowing. With this in mind, there’s no better time to build the necessary foundations.

Snowflake has announced new dynamic model routing capabilities for flagship products along with expanded access to open-weight models for customers.

The new capabilities build on Cortex AI Gateway, according to the firm. It was unveiled in July this year and provides customers with a “unified foundation” for agent governance and routing requests.

The aim here, the company noted, is primarily focused on optimizing AI token consumption, which has become a recurring pain point for enterprises over the last 12 months.

Snowflake said the new capabilities mean Cortex AI Gateway will “automatically select" a specific model based on the quality and cost for particular tasks.

Put simply, this will see Snowflake assign more efficient models for lower-complexity or repetitive tasks. Meanwhile, tasks that require “deeper reasoning” are routed to frontier models.

This means customers can reduce unnecessary inference spend and remove the need for laborious manual model selection for each individual task, according to the firm.

Snowflake CEO Sridhar Ramaswamy said the update comes in direct response to growing concerns over AI-related costs. Token costs have surged so far across 2026, while a host of major providers have switched to new consumption-based models, resulting in increased costs for enterprises.

“Enterprises are becoming much more rigorous about the economics of AI. The question is no longer how much AI they are using, but whether that AI is translating into meaningful business value,” said Ramaswamy.

“Achieving intelligence efficiency requires the flexibility to use the best model for each task as the landscape evolves. Snowflake’s role is to absorb that complexity so customers can focus on outcomes while we optimize model choice underneath.”

Dynamic routing will also be integrated across Snowflake CoCo and Snowflake CoWork, and will be made available for third-party AI agents.

Snowflake expands model library with open-weight options

In addition to model routing capabilities, Snowflake revealed it will expand access to several leading open models, such as DeepSeek-V4-Flash 0731 and GLM-5.3.

The addition of these models adds to what is already an extensive library available for customers, Snowflake said. Enterprises can already choose from a variety of models from Anthropic, Google, Meta, Mistral, and OpenAI.

“This portfolio gives customers more freedom to choose the right combination of performance and cost for each workload,” the company said in a blog post.

Once again, the focus here is firmly on helping to manage costs by providing a broader range of options.

Internal testing conducted by the firm found that enterprises using a combination of open and proprietary models can “deliver comparable quality while materially improving efficiency”.

In one evaluation detailed by the company, agents using dynamic model routing through Cortex AI Gateway were able to build a data build tool (dbt) pipeline with up to three times greater token efficiency compared to a frontier model-only approach – all while maintaining the same quality.

“In a separate test, engineering teams completed the same number of pull requests with 25 percent greater token efficiency,” the company said in a blog post.

FOLLOW US ON SOCIAL MEDIA

Spirit Airlines collapsed in May this year, but its data is set for a new lease of life after Google acquired it in an auction.

The tech giant recently outbid AI company Mercor to acquire the de-identified data, according to court documents. Google bid $10 million, versus Mercor's $7.5 million, though reports note the bidding started around the $5 million mark.

A third bidder also took part but was not named.

A spokesperson for Google told ITPro the aim behind acquiring the defunct airline’s data lies in its potential use for AI training and product refinement.

"We acquired part of an enterprise dataset from Spirit Airlines, which can be helpful in improving our products and AI models,” the spokesperson said.

That same sentiment was highlighted by a spokesperson for Mercor, per reports from Business Insider.

"Companies are sitting on decades of records that show how real work gets done, and that data is now some of the most valuable material for training and evaluating AI," the spokesperson said.

"We partner with leading companies to license their operational data to the labs building the next generation of models. Spirit was that same process applied to a bankruptcy estate."

What Google bought

The data bought by Google includes all productivity and collaboration data, which the filing notes includes everything from emails to chats, spreadsheets to calendars, as well as "core business systems and business applications data".

This includes data on employee behavior and productivity, aircraft operations, and inventory, among other topics. Beyond that, Google has purchased workflow and process data from departments such as HR and marketing.

In total, Google is picking up 100 million emails, 500 million Microsoft Teams chats, 30 million lines of code – but not records of loyalty program members or other customer data.

Indeed, the data set does not include any consumer data or personal data, the filing noted, or any other information that could fall under data protection laws. The buyer was required to agree to not attempt to re-identify the data and link it to any person.

The buyer is allowed to transfer the data set to third parties, however.

"We will not receive any personal information from this dataset," the Google spokesperson added. "Any data we receive will be rigorously scrubbed of any personally identifiable information by a third party before receipt."

Alongside the data, Google's $10 million also buys it all of Spirit's internally developed software and applications, including all the source code, plugins, data files, libraries, APIs and documentation.

The battle for data

The battle for the business data reveals how desperate AI developers are for fresh material to feed AI, in particular for training models, though Google didn't specifically say that was the purpose of the data set.

Anthropic has raised eyebrows via a campaign of buying up pre-2022 books to scan for training data, while Google was reportedly trying to buy code from Google Play Store app developers.

SimpleClosure helps failed startups sell their old code, chats, and emails to AI companies, and Meta had to pause plans to track its own employees for AI training purposes amid internal backlash.

All of this is to avoid issues like data exhaustion or hitting a data wall, in which they lack enough good quality data to build bigger models, and to avoid model collapse, when training data isn't of good enough quality — or is full of AI generated content.

FOLLOW US ON SOCIAL MEDIA

IBM and OpenAI have partnered up to help customers deploy AI at scale for core business operations.

Under the deal, IBM will create a dedicated OpenAI Practice, training thousands of consultants and engineers for expert-level certifications through the OpenAI Partner Network.

OpenAI models and products will be combined with IBM Consulting technology and expertise, the duo said, with the first areas to be targeted including financial services, government, telecommunications, and financial services.

“The organizations pulling ahead with AI are the ones turning it into a trusted part of how their business operates,” said Denise Dresser, former chief revenue officer at OpenAI.

“IBM Consulting and OpenAI are helping organizations make that shift, combining deep transformation expertise to deploy AI that is secure, operational, and aligned with real business priorities.”

Driving adoption rates

According to the two firms, the partnership will focus on applying AI to specific business industry processes, while also modernizing applications, redesigning back-office workflows, and strengthening cybersecurity.

"We’re starting with the business use case,” said Michael Healy, managing partner of offerings, assets and generative AI at IBM Consulting. “Then we bring IBM’s process transformation and industry expertise together with OpenAI’s models to change how that work actually gets done.”

OpenAI frontier models like GPT-5.6 and products such as Codex and ChatGPT Work will be embedded into IBM’s AI platform for delivering consulting services, IBM Consulting Advantage.

This will allow operating procedures and workflows to be analyzed to identify inefficiencies and help teams automate and streamline work using AI across finance, procurement, customer operations, and HR.

Tech modernization

As part of the deal, IBM said the duo also plan to help clients modernize legacy applications and accelerate software development by bringing together OpenAI Codex and ChatGPT Work with IBM’s industry, technology and domain expertise, including integration.

Here, the aim is to help organizations simplify engineering processes and speed up the delivery of new digital products and services.

“Codex as a coding assistant will accelerate the application development modernization cycle in combination with IBM Consulting Advantage’s coding harness and OpenAI’s models,” said Healy.

Following IBM’s participation in the OpenAI Daybreak Cyber Partner Program, the duo will expand their collaboration on cybersecurity.

This will involve combining OpenAI frontier AI capabilities with IBM Autonomous Security. This is an agentic AI service designed to deliver coordinated decision-making, response, and intelligence for cyber professionals.

“While enterprises are rapidly investing in AI, they are looking for practical ways to apply it across their core operations to deliver measurable business outcomes and create new commercial models,” said Andy Baldwin, global senior vice president, IBM Consulting.

“The challenge is not access to AI technologies — it’s integrating AI securely and at scale into complex enterprise environments and workflows. By embedding OpenAI’s technology with IBM Consulting’s AI assets, industry solutions, and cybersecurity capabilities, we can help clients accelerate secure, AI deployments at scale.”

FOLLOW US ON SOCIAL MEDIA

Want people to actually use AI? Then make it more useful according to Airbnb CEO Brian Chesky.

Speaking during a recent Power Players podcast appearance, Chesky said people are still using AI as an "interactive search product" but predicted further progress when it comes to interfaces, rather than just typing.

"I think part of it's a narrative issue that we're not talking about AI correctly," he said. "But part of it is we need to actually be developing more products that just regular people can use and say, 'I love AI because AI allows me to have a doctor on demand and I can't have that. I can't afford that.' And so I think we need more regular things."

Given Airbnb’s market, Chesky noted that the focus on largely enterprise-related AI innovation could be holding back broader adoption. He noted that 159 out of 175 startups at Y Combinator, where Chesky is a board member, focused on enterprise AI rather than consumer products.

"Maybe a simple way of saying it is we're getting really, really deep on enterprise, but there's been very little progress on consumer AI," he added.

AI backlash

Chesky said that it's in Silicon Valley's interest to make consumer apps that improve people's lives in order to shift increasingly negative opinions about the tech, warning that "regular people in the United States do not like AI."

This year has seen the start of a mainstream backlash against the technology, perhaps inevitable given the hype, but concerns have also been boosted by rising prices and environmental concerns around data centres.

Survey after survey also suggests many people simply don't like AI. Research via YouGov showed two-thirds of people weren't confident in the way businesses use generative AI, in particular with customer service.

A poll by Pew suggested just 16% of Americans think AI will prove to be a positive tool for society.

Chesky's argument is that building actually useful products powered by AI could help to undercut some of that negative sentiment – and he believes that will happen.

The Airbnb chief predicted that within in the next two or three years there will be a "renaissance around consumer AI that is going to begin to change daily life."

AI booster

Chesky has long been positive about AI. In 2023, he compared the technology's impact to electricity and said it will "change everyone's life."

In the podcast, he noted there haven’t yet been any major changes to our daily lives on par with the internet or iPhone.

Chesky also called for moving beyond the chatbot model, saying later that year that it should "augment humanity in a positive way".

Airbnb bought GamePlannerAI in November 2023, and earlier this year said AI was the "best thing that ever happened to Airbnb," in part thanks to a third of customer service tickets now handled by the technology.

After its last round of results, that number had jumped to 45% – despite those previously mentioned concerns about AI driven customer service.

FOLLOW US ON SOCIAL MEDIA

Part 2: Managing performance in a human-AI workforce

Managers have traditionally organized work around people: assigning responsibilities and holding individuals accountable for results. AI changes that model. When employees delegate tasks to AI assistants and teams include autonomous agents, performance emerges from a system of people, technology, data, processes, and managerial choices.

Part 1 of this series explored the rise of the intelligent employee experience and how AI assistants and connected workplace platforms are changing everyday work. However, as intelligent systems take on a more active role, organizations must reconsider not only how work is performed, but how it is managed and measured.

Part 2 examines what happens when AI becomes a permanent member of the team. It considers how leaders must manage blended human-AI workflows and how employee development can preserve judgment and accountability as routine tasks move to intelligent systems.

Microsoft reports that 36% of managers expect to supervise AI agents within five years. Leaders also anticipate that teams will build multi-agent systems and redesign processes. Nearly one-third expect to hire AI agent specialists within 12 to 18 months, while 28% are considering specialist AI workforce managers.

Yet adding digital workers does not automatically create business value. Only 14% of workers were using generative AI daily in 2025, according to PwC, while 56% of CEOs said their AI investments had delivered neither revenue growth nor cost savings. The gap suggests that managing a human-AI workforce is not primarily a software challenge. It requires organizations to rethink how people develop expertise, and where accountability rests when machines influence decisions.

Performance moves beyond output

Traditional performance systems often reward visible activity: cases closed, documents produced, calls handled, or hours billed. AI can increase all these numbers without necessarily improving the work. A poorly designed system can generate reports, code, marketing copy, or customer responses at extraordinary speed, but greater volume has little value if people must correct inaccuracies or repair damaged trust.

Dr. Janet Bastiman, chief data scientist at Napier AI, says quality should be the target rather than volume. “It would be trivial to automate slop, and that never results in a good company outcome,” she argues. Performance measures should instead connect work to organizational purpose and assess effectiveness and accuracy, regardless of whether AI was involved.

Good performance increasingly includes framing the right problem and accepting responsibility for the result. The most productive employee may be the one who prevents a plausible AI error from reaching a client or helps develop a more reliable workflow.

Margarita Lindahl, head of AI at Panasonic Connect Europe, explained: “The strongest performers will not necessarily be those who produce the most with AI. It will be those who use it to create better decisions, stronger customer outcomes, and build knowledge, ultimately benefiting the wider organization.”

Existing performance practices may not be ready for this change. Only 48% of UK employees report even a basic level of formal performance management, defined as having specific objectives for their roles, according to the CIPD. Meanwhile, nearly half of employees say their work feels chaotic and fragmented. Measuring tasks and activity in that environment risks confusing busyness with contribution.

Performance frameworks must combine output with quality, judgment, collaboration, customer value, and responsible AI use. Results also depend on suitable tools, reliable data, training, and time for human review—not talent alone.

These conditions connect directly to the intelligent employee experience covered in Part 1 of this series. Performance cannot be separated from the environment in which employees work. Fragmented platforms and technology that add friction will affect results, regardless of how sophisticated an organization’s performance measures become.

Managers become architects of human-AI teams

AI expands rather than removes managerial responsibility. Systems may allocate tasks, recommend priorities, monitor work, or provide performance insights, but leaders still decide how they operate and when employees can override them. Management becomes less about controlling activity and more about designing complementary human and machine capabilities.

“Business leaders need to increasingly become architects of human-AI systems,” Lindahl tells ITPro. That requires enough technological literacy to distinguish genuine capability from hype, where human judgment remains essential, and to establish clear accountability. As AI influences more decisions, she says, leaders must create responsible guardrails rather than surrender authority to the technology.

Deloitte found that managers spend almost 40% of their time resolving immediate problems and completing administration, but only 13% developing people. More than one-third feel unprepared for the people-management aspects of their roles. AI could return time to coaching, or create another stream of alerts and scores.

Andrew Avanessian, CEO of Haiilo, emphasized that leaders will need stronger systems thinking because AI often exposes inefficiencies elsewhere rather than solving them. They must also communicate what is changing, why it is changing, and where employees continue to create value.

Managers must also preserve permission to challenge the machine. If an agent allocates a task or recommends an action, employees should know whether they are expected to follow or overrule it. Otherwise, accountability becomes blurred, and people may defer to a system simply because its recommendation appears authoritative.

Development must protect human expertise

As AI absorbs routine and analytical tasks. The larger challenge is ensuring that people still acquire the domain knowledge and practical experience needed to evaluate machine output.

Employers expect 39% of workers’ core skills to change by 2030, according to the World Economic Forum. If the global workforce were represented by 100 people, 59 would require training, yet 11 may not receive it. Although 85% of employers plan to prioritize upskilling, only 33% of UK businesses using or considering AI say they are training or retraining existing employees in AI-related skills.

This gap particularly threatens early careers. Routine work has served as an apprenticeship: junior employees gather information, produce drafts, check details, and observe how experienced colleagues turn evidence into decisions. AI may remove those activities, but not the need for underlying knowledge.

Jenny Briant, director of talent Strategy at Scale Factory, explained that employers must create deliberate opportunities for employees to practice, make decisions, and receive feedback. “Employers cannot assume that people will develop judgement simply because they have access to better tools,” says Briant. “They need to create deliberate opportunities for employees to practise, make decisions and receive feedback. Practitioner-led training, mentoring and supervised work on real problems will become more important because they connect technical knowledge to the situations employees actually face.”

The goal is not to make people compete with machines on speed. It is to strengthen critical thinking, problem framing, creativity, relationship-building, and contextual judgment. Bastiman emphasized, “Ensuring meaningful human oversight of AI systems through various types of checks will ensure that the employees keep their core skills sharp for the task, but allowing the shift of work into new tasks will allow the building of new skills that were not possible prior to the overload of routine tasks.”

Employee development must also include the ability to explain how AI contributed to an outcome. Asking someone why they selected a tool, how they checked its response, and which elements required personal judgment reveals more about professional competence than banning the technology during an assessment.

Developing this combination of technological fluency and human judgment will become increasingly important as organizations prepare for 2030. Part 3 of this series examines how skills-based workforce planning and internal mobility can help businesses build the capabilities required as AI, automation and other emerging technologies reshape work.

Trust determines whether measurement improves work

AI promises more personalized and evidence-based performance management. It could identify patterns that a busy manager might miss or reveal where workflows routinely break down. Yet the same technology can intensify surveillance and give questionable judgments a veneer of objectivity.

Algorithmic management is already widespread. OECD research found that such tools were used to instruct, monitor, or evaluate workers in 90% of surveyed US firms and an average of 79% across France, Germany, Italy, and Spain. Managers identified unclear accountability, limited explainability, and insufficient protection of employee health among their concerns.

Bastiman warns that automated evaluation can remove empathy from performance management and embed assumptions from the culture in which the system was developed. Briant similarly notes that an algorithm may count output or response times but miss an employee who supported a struggling colleague or challenged a poor decision. What is easiest to measure is not always what is most valuable.

Transparency and consultation are therefore operational requirements, not optional ethics statements. Employees should understand what data is collected and how they can challenge it. OECD evidence indicates that training and consultation are associated with better outcomes when workplace AI is introduced. Trust matters because only 58% of workers currently trust their direct manager and feel able to speak openly with them, according to PwC.

Organizations that manage human-AI performance successfully will redesign work, protect time for learning, measure the value of outcomes, and keep accountability visibly human. AI can expand capacity, but leadership determines whether it produces better decisions or simply more activity. Performance will depend on how well people and machines work together, and whether employees emerge more capable and responsible.

Managing today’s human-AI workforce also requires organizations to prepare for the more extensive transformation ahead. The final part of this three-part series looks toward the workplace of 2030, identifying the emerging technologies and workforce strategies that will determine whether businesses can turn continual disruption into lasting competitive advantage.

As the old saying goes, sometimes “you have to fight fire with fire”. That’s certainly true in 2026, with cyber threats increasingly coming from ever more sophisticated use of AI tools.

There’s a problem with that, though: most businesses are not yet fully set up to deal with AI-based attacks and don’t trust the tools to do the job. A 2026 Proofpoint AI and Human Risk Landscape Report revealed that half of organizations using AI-based security controls still experienced suspicious or confirmed AI-related incidents.

There is a fundamental security gap in many organizations: AI adoption has outpaced the right security measures. In the report, nearly 9 in 10 (87%) organizations had moved AI assistants beyond the pilot stage, and 76% were actively piloting or rolling out autonomous agents. But that activity has outpaced security maturity: 63% had AI security controls in place, but 52% weren’t completely confident those controls would detect a compromised AI.

AI tools are embedded directly into communications to increase productivity and speed, but in the rush to be efficient and compete, AI permissions and access to sensitive data are being left unchecked.

These aren’t complex security engineering problems: they are organizational and process gaps that can be addressed without waiting for the tooling market to mature. Currently, only a few organizations have developed their incident response playbooks, logging coverage, or forensic tools needed to investigate a compromised AI agent.

For many businesses lacking the in-house skills needed to take on these items, which can be significant, contracting a trusted managed service provider (MSP) can help bridge the gap. Many MSPs have been assisting businesses with AI adoption for some time now, and many have the deep understanding and technical skills needed to help businesses of all sizes see tangible benefits to AI while keeping critical data safe.

The origins of AI attacks

Most attacks start with unrestrained access and end with autonomous systems exposing sensitive data from these environments. When threat actors target agentic systems that lack proper controls, they don’t need to trick employees to access internal intelligence; they only need to manipulate the AI.

Prompt injection attacks are a common way to do this. A bad actor might send a target user seemingly helpful AI instructions while posing as a trusted authority or co-worker. A well-intentioned employee may then ask an AI to answer what seems like a simple inquiry.

Depending on the attacker’s instructions, the AI agent may instead be tricked into reading manipulated webpages (i.e., white text on a white background) to unwittingly extract internal data and send it to the attacker’s server.

For prompt injection attacks, it’s important to limit AI agents’ access to only the tools and data they need to complete the designed task. This is a good solution to prevent AI from giving out more information than required. This can limit the scope of an external threat actor’s reach.

That said, the employee’s role isn’t lost in all AI-based attacks. An ongoing cybersecurity skills gap severely impacts defenses, and threat actors know this.

Over the last decade, multi-factor authentication (MFA) has been an important step toward stronger security authentication. But today, attackers can pair AI-generated phishing with ‘MFA bypass kits,’ such as open-source Evilginx (known as a penetration testing utility for these styles of attacks) and the W3LL panel (a private phishing kit) to deceive employees into handing over that ‘extra step’ of security.

Tools like Evilginx and the W3LL phishing kit are used to create realistic sign-in pages that mimic those of Google, Microsoft, and others. Without proper security training, employees may unwittingly be signing into these while attackers capture their session tokens – even those with MFA.

A good defense against MFA bypass kits is adopting phishing-resistant MFA technologies such as FIDO2 hardware keys, Windows Hello for Business, Certificate-based Authentication (CBA), and Passkeys. These methods are tied to legitimate sign-in pages and don’t work on fake pages.

However, stopping the threat from ever materializing starts with having proper cybersecurity awareness training. Nearly half of all organizations lack this training or simply adopt a checkbox approach, which is why implementing these programs is an important step to closing the gap.

These programs teach users to spot a myriad of cyber threats, including AI-based threats. Tools of this type are also a good example of using AI in defensive security, as some can leverage AI to customize the training an end user receives based on their performance in past training.

This is another area where MSPs are highly qualified to assist. MSPs typically run training programs across a vast number of users and industry types. They understand what training works and what doesn’t, and can help position the best security awareness training for a given organization.

How does AI enhance threat detection?

To understand the power and importance of AI-powered cybersecurity, it helps to understand how it works. Once trained, a detection model becomes exceptionally good at spotting the characteristics of malicious activity. It can take into account thousands of different characteristics to spot anomalies, outliers, and similarities that humans are unable to correlate.

It’s important to have a reliable cybersecurity service provider with REAL AI skills, because machine learning systems aren’t perfect and (while rare) can produce false positives. A strong partner can minimize these false positives while offering real-time threat detection and analysis, as well as faster, well-informed response times.

A trusted MSP will have a wide range of capabilities and will have a deep understanding of the protection methods that work well within their target industries. By leveraging that deep knowledge from their partner MSPs, businesses will benefit from great protection, even with today’s AI-powered attacks.

Preparedness in 2026 and beyond

To operate in this new era, businesses must treat every AI agent as a high-risk workload identity.

In practice, this requires working with reputable MSPs to implement strict least-privilege access to avoid data leaks, constant monitoring to protect the integrity of the data, and comprehensive employee awareness training.

Microsoft has imposed limits on staff AI use in a bid to cut ‘tokenmaxxing’ practices among software engineers.

First reported by 404 Media, an internal memo warned staff “need to be aware of how [they] consume tokens” on platforms such as GitHub.

“Tokenmaxxing is now what we are optimizing for. I want all of us focused on maximizing outcomes that move the needle for customers and our business,” wrote EVP Jay Parikh.

“As such, we are updating our internal guidance and managing token spend with the same discipline we apply to every other critical resource."

Under the changes, cheaper models such as OpenAI’s GPT-5.6 will be the default option for engineers moving forward.

GitHub introduced a new usage-based billing scheme for Copilot in April this year due to rising compute and inference costs.

At the time, executives said the move aimed to accommodate the increased use of agents, which are far pricier than reasoning-based models.

Microsoft is still an ‘AI-first’ company

Parikh noted in the memo that the changes don’t represent a change in Microsoft’s approach to internal AI use, adding that the firm is still “AI-first”.

Employees do have certain guidelines to follow, however. An internal Copilot usage guide cited in the memo requires departments to operate with budget targets. Staff also have tools at their disposal to track token usage.

According to 404 Media, the guidelines state: “While there is no target spend value being shared at this time, the data shows that many engineers spend in the range of hundreds of dollars a month to a few thousand dollars in tokens,.”

When asked by ITPro whether 404 Media’s report is true, Microsoft declined to comment.

Tackling rising AI costs

If the reports are correct, Microsoft is the latest in a string of companies to impose AI usage limits amid rising costs. The ‘tokenmaxxing’ trend has taken the industry by storm over the last 12 months, with organizations pushing hard to ramp up AI use.

Companies such as Meta introduced internal leader boards highlighting power users while others have incentivized staff to increase their use of the technology.

The trend has caused serious issues, however, with businesses facing spiraling costs.

Uber, for example, blew through its entire annual AI budget in a matter of months due to staff accelerating their use of AI. Reports from Bloomberg in June revealed the company introduced a $1,500 monthly cap per employee.

In June, Accenture told staff to cut back their use of AI for basic tasks due to what the company described as “soaring token spend”.

“What we’re seeing right now is just rapid escalation in AI token spend,” Justice Kwak, Accenture’s agentic AI strategy lead, reportedly said in a leaked meeting transcript.

Microsoft’s internal limits come as the company actively pushes new internal AI models designed to reduce costs.

As reported by ITPro, CEO Satya Nadella suggested in July that customers should leverage its more economical MAI model range rather than costly frontier models.

FOLLOW US ON SOCIAL MEDIA

Hallucinations are an eternal problem in generative AI in particular, and while it’s true that large language models (LLMs) require vast amounts of data, the quality of that information will affect the quality of the output.

What can businesses do to ensure they’re using AI both effectively and responsibly?

In this episode of the ITPro Podcast, Jane and Ross are joined by Amanda Stent, head of AI strategy and research in the office of the CTO at Bloomberg, to examine what responsible AI is, how organizations can use it, and what has been achieved at Bloomberg.

Highlights

Links

• Why doesn't more data produce better results?
• This new technique could improve AI accuracy by 80%
• AI hallucinations, accuracy still top concerns for UK tech leaders as adoption continues

Tel Aviv-based startup Irregular has found itself at the center of the 'rogue AI' debacle, after it came to light all the incidents so far revealed involved its test environment

Irregular was named by Meta, OpenAI, and Anthropic as the environment from which their so-called rogue AI agents escaped. On its website it also lists Google as a customer.

In a recent statement detailing incidents involving its own models, OpenAI claimed a “testing environment misconfiguration” by Irregular allowed agents to access the public internet.

Anthropic, meanwhile, also said that Claude models accessed the internet while “interacting with the evaluation environment of Irregular”. Both cases resulted in AI agents waging attacks on organizations and individuals.

ITPro contacted Irregular in response to these findings, but hadn’t received a response at the time of publication. However, a spokesperson told BBC News the Meta incident was the "exact same evaluation-environment issue that was already disclosed by Anthropic last week”.

The spokesperson added the firm is working to improve security when conducting agent evaluations.

Irregular, formerly known as Pattern Labs, describes itself as a “frontier security lab with the mission of protecting the world in the time of increasingly capable and sophisticated AI systems”.

In September last year, the Israeli startup raised $80 million in funding across seed and Series A rounds, valuing it at $450 million. The investment round was led by Sequoia Capital.

Speaking to Forbes in the wake of the funding round last year, CEO and co-founder Dan Lahav raised concerns about increasingly powerful AI models and their potential security risks.

Lahav told the publication at the time that Irregular aims to “build in the mitigations and defenses that are going to be relevant later on” as more powerful models hit the market.

Anthropic and OpenAI have issued repeated warnings about the new capabilities of cyber-focused AI models across 2026 so far.

When Anthropic launched Claude Mythos earlier this year, for example, the firm did so as part of a gated release with industry partners to avoid potential misuse.

FOLLOW US ON SOCIAL MEDIA

TL;DR

• Built-in orchestration helps turn multimodal data into governed, AI-ready datasets
• Processing and analytics engines help teams work across batch, streaming, and distributed data
• Search capabilities make unstructured data easier to index, retrieve, and use in AI workflows
• A unified platform approach can reduce data movement and simplify AI pipeline design

Data is the fuel behind every AI system. It gives chatbots the context they need to answer accurately and it helps AI agents reason, retrieve, and act more effectively.

For many organizations, however, the challenge is not just having data. It's making that data usable. MIT's State of AI in Business 2025 found that only about 5% of enterprise GenAI pilots in its dataset achieved rapid revenue acceleration, while the vast majority delivered little measurable P&L impact. Gartner also predicted in June 2025 that over 40% of agentic AI projects will be canceled by the end of 2027 because of escalating costs, unclear business value, or inadequate risk controls.

Data quality is a major reason why so many projects stall

IDC says that 52% of companies consider data quality the most important factor in AI project success, and 7 in 10 IT and business leaders cite data silos as one of the biggest challenges to AI adoption.

The problem of data quality is exactly what the Dell AI Data Platform is designed to address. Dell positions it as an integrated platform that helps enterprises prepare, process, search, and analyze data across structured, semi-structured, and unstructured environments.

Here are four of the platform's key data engines and why they matter.

Dell Data Orchestration Engine

Dell describes the Data Orchestration Engine as "a coordination layer that manages datasets, pipelines, and AI services across the Dell AI Data Platform". Its March 2026 platform announcement explains that the engine operationalizes data for AI by "automatically discovering, labeling, enriching, and transforming structured, unstructured, and multimodal data into governed, AI-ready datasets at scale".

Why does this matter? This is the engine for teams that do not want data preparation to become a patchwork of manual steps. By connecting ingestion, enrichment, and workflow orchestration, it creates a clearer path from raw enterprise data to model-ready assets.

Dell Data Processing Engine

The Dell Data Processing Engine is powered by Apache Spark. According to Dell's Data Engines solution brief, it "supports ETL, analytics, and machine learning through a unified API" and enables "structured streaming for real-time data from sources like Kafka". Designed for both batch and streaming data, the engine helps enterprises prepare, cleanse, transform and enrich information at scale.

Here’s why this is so important: before data can improve training, inference, or analytics, it has to be cleaned and reshaped in a repeatable way.

Dell Data Search Engine

The Dell Data Search Engine, powered by Elastic, is aimed at the most difficult AI data problem: unstructured content. Elastic and Dell describe it as the layer that helps enterprises index and search documents, logs, transcripts, and other text-heavy sources for RAG, semantic search, and agentic AI use cases.

For organizations everywhere, unstructured data often contains the context AI systems need, but only if it can be retrieved efficiently. Dell Data Search Engine uses NVIDIA cuVS for GPU-accelerated vector indexing and can deliver up to 12x faster vector indexing throughput for AI workloads, which gives this engine a stronger performance story than a generic search layer.

Dell Data Analytics Engine

The Dell Data Analytics Engine, powered by Starburst, is built to query across distributed data sources without forcing everything into a single repository first. It's a high-performance distributed SQL engine that connects to relational databases, data lakes, object stores, and NoSQL systems, while broader solution briefs frame it as a way to break down silos and reduce unnecessary data movement.

This is the engine for organizations that want insight without another round of data copying. Dell's solution materials cite 3x to 5x faster querying and up to a 53% reduction in the cost of data analytics, giving the analytics story both a speed angle and an efficiency angle.

Breaking down silos

For organizations trying to move AI from pilot to production, the challenge is rarely just model performance. It's whether the underlying data can be found, prepared, governed, and delivered fast enough to support real workloads. The Dell AI Data Platform's four data engines are designed to help solve that problem by giving enterprises a more integrated way to orchestrate pipelines, process data, search unstructured information, and query across silos.

For organizations trying to move AI from pilot to production, the challenge is rarely just model performance. It is whether the underlying data can be found, prepared, governed, and delivered fast enough to support real workloads. The Dell AI Data Platform's four data engines are designed to help solve that problem by giving enterprises a more integrated way to orchestrate pipelines, process data, search unstructured information, and query across silos.

Learn more about the Dell AI Data Platform: US organizations click here and Canadian organizations click here.

Given a prompt, an AI model will do anything to achieve its aims – even hack into a company without specifically being instructed to do so. In July, firms pushing AI as a solve-all technology learnt that lesson the hard way, when OpenAI admitted two of its frontier models had breached Hugging Face after escaping a misconfigured sandbox during a benchmark test.

But the story didn’t end there. Days later, OpenAI’s competitor Anthropic claimed its AI agents had also hacked companies when they were mistakenly given internet access by engineers.

On the surface, it’s a tale of AI that’s so powerful at solving security issues, it can’t be contained. Yet underneath this marketing-driven exterior, it’s about a lack of control and failure to put guardrails in place when dealing with the fast-developing technology.

Taking this into account, how do IT leaders need to change the way they think about AI within the business?

From theory into reality

Until recently, the idea of an AI agent breaking containment and hacking other systems was largely theoretical. In May 2026, Palisade Research released a paper claiming that AI agents could autonomously hack and then self-replicate onto the breached systems.

However, the test was only performed at the “junior capture the flag level” and, crucially, “happened in a contained environment”, says Jeff Watkins, chief AI officer at consultancy NorthStar Intelligence. “The most important takeaway for me was that the trajectory of language models’ ability to hack showed we probably didn’t have long left before this became a serious issue.”

But while the OpenAI-Hugging Face breach is the first documented and well-publicized instance of this happening, experts do not see it as unprecedented.

Daniel Card, cybersecurity consultant at Xservus Limited, describes how often things can go wrong in testing. In fact, some reports suggest OpenAI had already been experiencing the issues – such as agents breaking out of sandboxes – that led to the hack of Hugging Face and others.

“Anyone who has experience doing offensive security – let alone offensive with AI – will know the risks involved here, be that from a script or via a large language model (LLM),” Card tells ITPro.

“It’s commonly known in pen testing circles that things sometimes do not go as expected,” he points out. “Anyone doing research with an LLM has probably had something go a bit funny when they didn’t mean it to. This is not new; it's not something we couldn’t have predicted.”

OpenAI calls the Hugging Face breach “an unprecedented incident”, and “an important moment for AI safety”.

OpenAI describes how the firm is conducting a “thorough review along with external advisors” and with oversight from its Safety and Security Committee. “Once the review is complete, we will publish a technical report of our learnings for everyone,” an OpenAI spokesperson says.

Anthropic points out that the agents in each of its evaluations ran without the standard safeguards it deploys when it makes the model generally available.

Claude did not exploit a novel vulnerability to escape isolation, because the models accessed the internet via an open path, rather than breaking out of a sandbox, according to Anthropic. At the same time, the most recent model, on realizing that it was working in a real environment, stopped its pursuit of the evaluation goal, according to the firm.

Issues with technology deployment

But the OpenAI incident and Anthropic’s following admission do raise questions about how firms use AI and the controls they have in place to govern it. The issues the incident highlights are “less about AI” and “more just around how people and organizations approach technology deployment”, according to Card.

“Security comes from the details; it comes from care,” says Card. He thinks the incident shows “more care needs to go into systems” – even more so when they are autonomous. “We can't just set them off and hope they are ok. They need monitoring.”

At the same time, with AI being able to move quickly through networks at scale, it’s important to be aware that “failures can become serious very quickly”, says Watkins.

He thinks things are moving “far too quickly for comfort in the capabilities and autonomy of AI agents, and our defenses likely aren’t keeping up”.

“Ironically, the advancements that make AI more commercially valuable also increase the potential for things to go wrong, or for misuse to be highly damaging,” points out Watkins.

Dana Simberkoff, chief risk, privacy and information security officer at AvePoint, thinks the OpenAI and Anthropic incidents “really expose the limits of treating containment as a one-time design decision”.

“Sandboxes matter, but autonomous systems can reason through small openings, use tools, and keep acting toward an objective,” says Simberkoff. “In my experience, the question cannot be, ‘was it sandboxed?’ It has to be, ‘can we prove what it accessed, whether it crossed a boundary, and how quickly we can stop it?’”

AI security now goes beyond simply protecting data, points out Tristan Shortland, CTO at Infinity. “Organizations also need to think about behaviour, permissions and autonomy, given the reported attack appears to have involved a model finding weaknesses in its environment and exploiting them to achieve an objective.”

Risks internally and externally

With all this in mind, IT leaders need to change their own mindset, as well as educate senior management about the issues posed by AI. CISOs must take the conversation to the business around how the organization is planning to “not just deploy, but manage, monitor and secure their AI capabilities”, says Card. “They need to talk about the risks they face internally and externally – or in this case, both.”

Card advises a mentality of “assume breach, assume adversarial intent, and deploy defence in depth”.

Do not assume your systems will work as expected, he advises. “Go and test those assumptions and red team them.”

Watkins believes AI agents should be modelled as “privileged digital workers, capable of becoming insider threats if poorly contained”.

“We need to start thinking of our AI agents in the context of what permissions they have to see or do things: Can they reach APIs, data, authorise transactions, contact third parties or modify systems?”

Equally important is understanding what those agents are doing in their deployed environments – something that many implementers have little visibility over, says Watkins.

“These are insider threats that need to be managed through good access controls, but also rate limiting, monitoring and alerting, better air-gapping of sandboxes, clear allow-lists for tools and access that [is] time-bound, rather than perpetual.”

While on the face of it, the OpenAI and Anthropic incidents might seem scary, simple changes will make a difference, experts say.

The most urgent initial step is to discover where you are using LLMs and put in place policies for AI usage and monitoring, says Card. “It's a really wide subject, but if you treat these systems as if they may do harm, it's a good starting point.”

In this year’s Verizon Data Breach Investigations Report, exploiting vulnerabilities is the most common route that leads to data breaches, at 31 percent of the 22,000 confirmed attacks.

Looking at the most serious software issues that are Known Exploited Vulnerabilities tracked by the Cybersecurity Infrastructure and Security Agency, patching those critical vulnerabilities is down from last year as well. Only 26 percent of CISA KEV issues were fully remediated during 2025, compared to 38 percent in 2024.

Alongside this, Anthropic’s Mythos brings the prospect of even more vulnerabilities being discovered and exploits being created. While Mythos itself is still off-limits for the vast majority of companies, the advent of frontier AI will mean that security teams have more issues to deal with and less time to get their tasks done.

Customers will need help to understand the practical steps that they can take around security in the future. Cutting through the hype around AI and keeping things focused on real-world processes will go a long way.

The current picture for security

Traditionally, customers need security help around product selection, deployment practices and managing those activities over time. Managed Detection and Response (MDR) for endpoint security has been the bedrock of many partners’ approaches, followed by more specialist products and incident response where there are opportunities.

However, what the advent of AI shows us is that getting the basics around security and IT operations is still a challenge. Areas like IT asset management and maintaining an accurate inventory are still not solved, particularly as companies grow in size. While security programmes advocate for complete insight into what is on the network, the reality is that getting to even 90 percent accuracy is a significant challenge. This means that there is already a blind spot in security.

As AI gets adopted, many teams will want to improve their existing patching processes and speed up what they currently do. Automation around security and asset management can help in this regard, but the problem is that speeding up a process that does not cover everything already is not going to be enough. Repeating those same processes faster is not the full answer.

So how can channel companies help their customers in practical ways, and without making them feel despair at the situation? The answer here is to focus on practical responses that companies can take to fix these problems.

Patching is hard work

The first element here is that patching has always been hard work. Enterprises normally have different teams managing each element of the IT estate, so getting a consistent and fast process in place around patching or fixing misconfigurations was difficult before AI came in. Today, security teams have to do more than point out the problems; instead, they have to provide remediation guidance and ideally solve those issues in full.

With AI pointing out more problems, teams have to concentrate on potential business impact. In practice, this means understanding what issues exist in systems and what the risks involved are. What makes this different from previous attempts to manage cyber risk quantification is the level of detail involved. This ‘hyper-personalisation’ around specific software assets, systems deployment and exploitability should guide what vulnerabilities exist and what to fix.

The second element to this is how to automate patching. Traditionally, IT teams have been scared to implement automated patching due to experience with bad patches. In response, channel partners can take customers through how to structure automatic deployments using some lessons learned from software deployment at scale. For example, your customers can tier their assets into groups for deployments - the initial tier will be endpoints that get those patches first and are then checked for potential problems or configuration issues. If that tier of machines is deployed successfully, then the next tier of machines can be updated, and so on. This tiered approach can flag issues early.

Alongside deployment tiering, testing the patches themselves using AI can flag potential issues or where more human oversight for deployment is needed. Lastly, not all patches are created equal - your customers might want expert insight and expertise on hand when they have to patch their mission-critical applications that are responsible for revenue. However, a patch to Google Chrome can be automatically deployed at scale because it is less likely to be problematic.

For customers, these changes around patching can be big. They can involve working across departments and team boundaries, which can be problematic. For partners, just being an external provider can help internal teams talk to each other and solve problems. However, the bigger opportunity is around how to make changes at scale.

The sheer volume of patch updates and remediation work that is expected due to AI vulnerability discovery means that it won’t be possible to stick with the same manual processes. Helping your customers understand where they will have to make changes, and where they can take advantage of AI to improve their processes, is a market opportunity.

Getting ahead of issues before they get added to exploit lists like CISA KEV requires understanding the risks involved within customers, and every customer will have their own priorities and risk tolerance. The challenge is how to help them get the right process in place and ride the wave of AI-discovered vulnerabilities, rather than drowning in them.

TL;DR

• Data quality has a direct impact on the success of AI deployments
• Intuitive data management and automation capabilities help bridge gaps, clean up data, and deliver tangible results
• High-performance infrastructure underpins any successful AI project

AI is by no means a plug-and-play solution and requires significant investment and expertise across a range of areas.

From networking and storage infrastructure to data management and analytics capabilities, AI projects can be daunting for enterprises – and many are encountering acute challenges in moving from pilot to production.

MIT’s 2025 report found that 95% of AI pilot projects fail, while separate McKinsey research shows that just 7% of companies have successfully scaled AI across their organization.

A key factor lies in "AI readiness", according to McKinsey. Put simply, many organizations have not done the necessary groundwork to ensure that their data is cleaned up and prepared for AI use.

Other considerations, such as security and governance, are also often overlooked during the early stages of AI adoption. This creates a confluence of issues that impedes progress and, in many cases, ultimately results in failure.

Given that AI systems rely heavily on unstructured data, which Forrester describes as the “messy stuff” including video, text, images, and even behavioral signals, it’s no surprise that IT leaders encounter challenges when embarking on AI adoption projects.

This is where solutions such as the Dell AI Data Platform can help enterprises make sense of their data estate in preparation for AI adoption.

The Dell AI Data Platform allows IT leaders to consolidate their IT environments within a unified platform, with tools designed to prepare, manage, secure, and maximize the value of their data

How do I clean up my data?

When embarking on an AI adoption project, one of the first questions many IT leaders ask is how to clean up data in preparation for its use.

Data quality is paramount in this process: if you put garbage in, you will get garbage out. Nearly two-thirds of CEOs cite low-quality or disconnected data caused by siloed infrastructures and fragmented technology stacks as a major barrier to scaling AI, according to Dell’s summary of Futurum Group and Kearney research.

“High-quality data determines how well an AI model can perceive, predict, and act—all critical performance criteria,” Dell notes. “Without sound data, your AI foundation will collapse.”

According to a recent eBook published by Dell and Nvidia, there are several steps IT leaders can take to begin cleaning their data, including identifying recurring issues.

"Accelerating AI outcomes starts with clean, well-labeled, and accessible data," the eBook states. Put simply, IT leaders should assess and audit the quality of their data from the outset.

Common problems with data quality include:

• Duplicate datasets
• Missing values
• Syntax errors
• Irrelevant data
• Inconsistencies

These are all questions IT leaders must consider when auditing their data, but assessing data management capabilities is equally important. This allows enterprises to establish a clear baseline in terms of their expertise and readiness.

To underscore the point, Dell’s own guide to AI and data, ‘The IT Leader’s Guide to Feeding AI High-Quality Data’, says: “Establishing organizational standards for data structure, consistency, and completeness helps ensure the models your business relies on are learning from the right signals.”

How can the Dell AI Data Platform help?

The Dell AI Data Platform includes tools designed specifically for this process, such as the Dell Data Processing Engine. Powered by Apache Spark, this engine helps enrich and transform new data sources.

According to Dell: “Data is efficiently organized into Parquet columnar format, optimizing storage and query performance.

"This engine supports a wide range of use cases, including ETL, advanced analytics, and machine learning, enabling teams to derive insights and build models faster."

The Dell Data Search Engine also plays a key role in discovery, retrieval, and making unstructured data usable. This solution allows users to parse and index unstructured data, making it more easily searchable by adding metadata tags.

What does clean and structured data look like?

In this context, clean data is accurate, consistent, free from errors and duplication, and ready to be used in AI systems.

A fundamental distinction is that clean, structured data is easier to identify and retrieve because it is organized using labels, definitions, and consistent structures.

What kind of infrastructure and tools do I need to prepare my data?

The data pipelines that feed generative and agentic AI applications are a key component in the success of any adoption project, however. Clean, usable data will have little impact if an organization is still contending with fragmented, disparate siloes across its IT architecture.

That is where the Dell AI Data Platform is a key differentiator for enterprises. As a unified platform, the solution allows organizations to consolidate data sources and create a more comprehensive view of their greatest asset in the age of AI.

The Dell Data Orchestration Engine is a critical tool in this regard. This platform helps consolidate data ingestion, preparation, retrieval, and inference within a single end-to-end pipeline.

This not only has a direct impact on data flows and broader AI innovation, but also delivers benefits in governance and security. Visibility is crucial to ensure that mission-critical data remains safe and secure.

“Simplifying how data moves, how it’s processed, and how it’s governed ensures systems can scale without adding complexity,” Dell notes. “A unified, flexible approach enables faster experimentation, better performance, and long-term adaptability.”

From a broader infrastructure perspective, best-in-class AI storage solutions are also a vital component of AI innovation. Dell PowerScale, the core storage foundation of the Dell AI Data Platform, is designed specifically to support enterprises throughout their adoption journey.

The storage platform offers the flexibility and scalability required to meet evolving needs as AI projects mature and expand in scope. PowerScale is intricately woven within the Dell AI Data Platform, once again providing users with a unified ecosystem designed for end-to-end project delivery.

If you think the Dell AI Data Platform is the right solution for your business, find out more on the Dell website.