AI agents in ITSM: Why are only 13% of IT leaders using them?

Every day, I speak with IT leaders who are asking the same question: “Are we using AI enough?” In my view, that is fundamentally the wrong question to be asking in 2026. The narrative around AI adoption has become too focused on the mere presence of the technology rather than the depth of its impact. When I look at the landscape today, I see a clear divide between those who are simply checking a box and those who are truly transforming their operations. The real question we should be asking is: “Is your AI resolving tickets, or is it just responding to them?”

The distinction is critical. I’ve seen that AI agents, the kind that execute tasks and workflows autonomously, are used by only 13% of IT organizations. While that number might seem small, this group is operating on a different plane of efficiency. They aren’t just using a different tool; they are leveraging a different kind of intelligence. While 87% of the industry is still caught in a cycle of manual intervention, this 13% has crossed the threshold into true agentic execution. This isn’t just a minor lead; it’s a growing gap that will define the winners of the next decade.

What sets AI agents apart from the AI most teams are using

For me, the difference is straightforward but profound. Most AI in service management today is purely advisory. It surfaces knowledge, suggests potential actions, and perhaps helps a human make a decision slightly faster. In other words, it “responds.” But responding is not the same as solving. AI agents go further because they act. They don’t just tell you how to fix a problem; they take the wheel and resolve it.

In practice, this means triaging and resolving tickets without any human involvement, running complex onboarding workflows from start to finish, and proactively catching issues before they even surface as incidents. I believe the human element should be reserved for high-judgment, complex work that requires empathy and nuanced strategy. The agent should handle everything else. This shift, from AI that supports human decisions to AI that executes them, is where I see the most significant productivity gains occurring.

That shift, from AI that supports decisions to AI that executes them, is where the productivity gains become significant.

The productivity gap between agent users and everyone else

My observations are backed by clear data. We’ve found that AI agent users are 50% more likely to report spending less time on repetitive work than the average AI user. I want to emphasize that point: this isn’t a comparison between agent users and those with no AI. This is a 50% advantage over teams that are already invested in traditional AI. That is a massive performance delta within a group that already considers itself tech-forward.

The difference shows up across every metric we measured. Agent users report more tickets resolved per team member, faster onboarding for new staff, and higher end-user satisfaction. When asked whether end users were satisfied with their AI implementation, 72% of AI agent users said yes. Among teams using AI without agents, that number dropped to 56%.

For any leader looking to demonstrate the tangible value of their technology investment, these numbers are impossible to ignore. They represent a fundamental shift in how we measure success in IT: not by how much technology we have, but by how much capacity we reclaim.

Nearly 90% of organizations are leaving the biggest efficiency gain on the table

Here is the context that I find most alarming: nearly 90% of organizations are still leaving their biggest efficiency gains on the table. According to our recent State of Service Management 2026 report, IT teams are losing an average of 35% of their working time to manual, repetitive tasks. AI agents are the most effective means we have for reclaiming that lost time, yet the vast majority of the market remains stuck in old work patterns.

One of the most important things I want to clarify is that this is not a company-size issue. Whether you have 500 or 5,000 employees, AI agent adoption consistently sits at 12-13%. The only slight outlier is mid-to-large enterprises with 2,000 to 5,000 employees, at 19%. This tells me that size alone does not create the conditions for agentic AI. Instead, it is a deliberate investment in maturity. You don’t simply “buy” an agentic future; you build the maturity required to sustain it.

Some industries are moving faster than others

Construction and telecommunications are currently leading in AI agent adoption, with 20-22% of organizations in those sectors using agents to automatically handle and resolve tickets. Finance and manufacturing are pushing hard on proactive execution, using agents to catch and address issues before they escalate.

Education and healthcare are moving more cautiously, largely due to the sensitivity of their data and the need to keep humans involved in certain decisions. The trade-off is that teams in those sectors still carry a higher share of manual work than their counterparts in faster-moving industries.

You don’t jump to AI agents, you work up to them

I’ve found that you don’t just jump to AI agents; you work up to them. My research shows a direct relationship between overall AI maturity and agent adoption. Organizations already using AI for automation are nearly twice as likely to have deployed agents compared to those using AI only for basic guidance. It is a progression, not a shortcut.

The teams getting the most value today are those that have spent the last 18 to 24 months building their capability: starting with knowledge support, moving into recommendations, and eventually reaching the level of autonomous execution. For every IT leader, the question is where you sit on that path today. The teams that commit to this journey now will be the ones setting the benchmarks everyone else will chase in 2027. I invite you to download the full State of Service Management 2026 report to see where the leaders are heading.