Companies want to automate IT operations using agents, but they don't trust them enough to make decisions.
That's according to research from Riverbed, which revealed that a whopping 90% of surveyed IT managers and tech leaders want to use agentic AI to power autonomous IT operations in the next two years.
A further three-quarters believe agentic AI will be a core part of future IT strategies, with nine-in-ten saying AI investment to date has exceeded expectations.
So what's the problem? Put simply, companies aren't sure they trust AI agents yet. In the survey, 77% of respondents said their organization didn't want to allow AI to make operational decisions without human approval.
Early days for autonomous IT
The roll-out of autonomous IT operations remains in its early days. Just one-in-five organizations run automated IT operations right now, the report found, and only four-in-ten have an IT environment ready to support agentic AI.
That echoes previous research showing that AI adoption rates aren't matching IT hype.
While AI hasn't yet automated IT operations, the survey found that 82% of those polled are using AI at their companies for innovation and to create new "sources of advantage”.
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.
Beyond that, the survey suggests that though 76% of projects have progressed beyond the proof of concept, that's only 5% more than last year, and just 9% are deployed across the enterprise.
"The journey to autonomous IT is well underway," said Jim Gargan, chief marketing officer at Riverbed. "Organizations increasingly see AI as the path to more resilient, efficient and proactive IT, but our research shows significant gaps between ambition and readiness."
"The next phase of AI is not simply more automation, but AI that enterprises can understand, trust and govern," added Gargan.
Autonomy challenges
That next phase will see multiple hurdles that need to be overcome. According to the survey, more than half (55%) of respondents said autonomous IT faced challenges from a security and compliance perspective.
Other hurdles include operational disruption (45%) and a lack of trust in AI decisions (39%).
No wonder then that more than nine-in-ten said that AI observability and governance was set to become an increasingly critical area for IT to manage. While three quarters of respondents said AI was a critical or important priority, the report noted that "enthusiasm is not evenly shared".
"Business and IT leaders are more optimistic about agentic AI’s potential, while the technical specialists responsible for deploying, governing, and operating it are more attuned to the practical hurdles ahead,” Riverbed added.
Although 42% of those polled are confident in their employers' governance and oversight of AI, for technical specialists that falls to 36%.
Plus, just 17% of technical specialists see fully autonomous agents making up a large part of their employers' operating model in the next few years, the survey found.
How to deliver autonomous IT operations
To build autonomous IT operations, respondents said they need high-quality data, unified observability, and consolidated tools.
But that isn't yet available, with 19% of overall respondents saying they lack the data required for AI-powered IT operations, with just 17% of technical experts agreeing.
A further 17% believe they have unified visibility, leaving the rest unprepared as yet, while 87% said they have yet to integrate tools and systems.
"The message is clear: organizations have strong ambitions for autonomous IT, but fragmented data, incomplete visibility, and disconnected tools are holding them back," the report added.