Half of agentic AI projects are still stuck at the pilot stage – but that’s not stopping enterprises from ramping up investment
Organizations are stymied by issues with security, privacy, and compliance, as well as the technical challenges of managing agents at scale
Agentic AI projects are failing to get past the pilot stage because enterprises can't govern, validate, or safely scale them.
Adoption of this latest iteration of the technology is still in its early stages, a survey of 919 senior global leaders from Dynatrace has revealed, but is growing rapidly, with 26% of organizations having 11 or more projects.
Organizations expect the biggest return on investment (ROI) from agentic AI in system monitoring (44%), cybersecurity (27%) and data processing (25%), and three-quarters are expecting an increase in their AI budget next year.
However, this may all be wishful thinking. Around half of these projects are stuck in the Proof-of-Concept (PoC) stages.
The main barriers to full implementation, respondents said, are concerns with security, privacy, or compliance, cited by 52%, followed by technical challenges to managing agents at scale, at 51%.
“Organizations are not slowing adoption because they question the value of AI, but because scaling autonomous systems safely requires confidence that those systems will behave reliably and as intended in real-world conditions,” said Alois Reitbauer, chief technology strategist at Dynatrace.
Seven-in-ten agentic AI–powered decisions are still verified by humans, and 87% of organizations are actively building or deploying agents that require human supervision.
Sign up today and you will receive a free copy of our Future Focus 2026 report - the leading resource for IT decision-maker insight on priorities and investment areas in AI, security and more.
Leaders said they expected a 50-50 human–AI collaboration for IT and routine customer support applications, and a 60-40 human–AI level of collaboration for business applications.
“With most enterprises now spending millions of dollars annually and planning further budget increases, agentic AI is becoming a core part of digital operations. At the same time, the data shows a clear shift underway," said Reitbauer.
"While human oversight remains essential today, organizations are increasingly preparing for more autonomous, AI-driven decision-making. The focus is now on building the trust and operational reliability needed to scale agentic AI responsibly.”
Observability is a problem with agentic AI
A recurring pain point for enterprises tinkering with agentic AI tools lies in observability, according to Dynatrace. Observability of these autonomous systems is needed across every stage of the life cycle, from development and implementation through to operationalization.
Observability is most used in implementation, at 69%, followed by operationalization at 57% and development at 54%.
“Observability is a vital component of a successful agentic AI strategy. As organizations push toward greater autonomy, they need real-time visibility into how AI agents behave, interact, and make decisions,” Reitbauer said.
“Observability not only helps teams understand performance and outcomes, but it provides the transparency and confidence required to scale agentic AI responsibly and with appropriate oversight.”
AI pilot purgatory is a recurring theme amongst executives. A study from Informatica last year revealed that fewer than two-fifths of AI projects have successfully made it into production, while two-thirds of firms have seen fewer than half of their pilot schemes make it into the real world.
More than four-in-ten cited a lack of data quality and readiness as the biggest obstacle to reaching production, while 43% also blamed a lack of technical maturity.
FOLLOW US ON SOCIAL MEDIA
Make sure to follow ITPro on Google News to keep tabs on all our latest news, analysis, and reviews.
You can also follow ITPro on LinkedIn, X, Facebook, and BlueSky.
Emma Woollacott is a freelance journalist writing for publications including the BBC, Private Eye, Forbes, Raconteur and specialist technology titles.
-
Samsung backs Mistral in record-breaking €3 billion funding roundNews The French AI company breaks European records, but still trails American rivals with a €21bn valuation
-
Advania targets fresh growth with new UK chief executiveNews The seasoned technology veteran will lead the IT services provider into its next phase of growth and transformation
-
“Enterprises are becoming much more rigorous about the economics of AI”: Snowflake wants to help you cut AI costs by choosing the right model for the right taskNews New dynamic model routing capabilities aim to help customers box clever when it comes to their AI model choice
-
Microsoft has joined the growing list of companies cracking down on ‘tokenmaxxing’News The company is updating internal guidance to reduce rising costs
-
‘AI cost management has the same problems that cloud had’: Enterprises are still facing huge AI bills thanks to ‘tokenmaxxing’ – that means FinOps practices are more important than everNews With firms facing surging AI bills, FinOps techniques are more important than ever
-
‘We are now seeing MAI models outperform general-purpose frontier models’: Microsoft CEO Satya Nadella touts in-house models to cut spiralling AI costs – and reduce growing reliance on frontier labsNews The Microsoft chief says pricey frontier models don't have to be used for every task, and its own in-house MAI models could be the key to reducing costs.
-
3 ways to get your data AI-readySponsored Data placement, preparation, and orchestration make or break an AI project – and the Dell AI Data Platform reduces the strain on enterprise IT teams across both storage and the data foundation
-
AI agents could make living off the land attacks ‘much more dangerous’, says CrowdStrike Field CTOAnalysis Living off the land attacks are evolving rapidly as threat actors target agents with deep access to enterprise networks and sensitive data
-
‘A company should be able to use a model without giving up the knowledge that makes it unique’: Microsoft CEO Satya Nadella says enterprises shouldn’t be sharing so much data with AI providersNews The Microsoft chief warned that corporate data could be at risk thanks to AI models
-
‘What we’re seeing right now is just rapid escalation in AI token spend’: Accenture tells staff to stop using AI for unnecessary tasks amid surging costsNews Accenture has told some staff to roll back the use of AI for basic tasks while the consultancy grapples with surging AI token costs.