Enterprise IT departments are witnessing a significant shift in how work gets done, with AI agents now executing roughly one-third of all IT actions. However, a comprehensive new study indicates that this automation is far from a complete takeover. Instead, it represents a collaborative model where human analysts remain essential for oversight, exception handling, and high-stakes decision-making.
According to data analyzed by IT automation platform Fixify, human analysts are rejecting about one-quarter of AI-proposed actions, though that rejection rate is steadily declining as systems improve. The research, which examined tens of thousands of human-AI interactions, suggests that the path to transforming IT work is not about replacing help desks, but about building a credible, hybrid workflow where AI handles routine tasks and humans manage complexity.
The Current State of Agentic IT
Fixify identified four distinct steps in agentic work: planning, proposing, approving or declining, and acting on approved steps. The study analyzed nearly 18,000 plans and over 147,000 actions executed by agents across 40 companies over a three-month period.
The findings reveal that agents are taking over one-third of IT actions, particularly in software, applications, security, and collaboration work. These areas are characterized by requests that are “repeatable and easy to reverse,” making them ideal candidates for automation.
Tasks that are well-understood and present low risk are best suited for the current generation of agents. Human analysts remain closely involved in higher-stakes areas like identity verification, setting up and removing IT access (onboarding and offboarding), and managing hardware environments.
However, the efficiency of AI is increasing as feedback loops improve. Over the three-month period, human approval of AI-proposed actions rose from 23% to 41%, and the rejection rate fell from 27% to 16%. This trend suggests that as agents learn from human corrections, they become more reliable and require less supervision.
The company identified six types of actions in AI automation. Running a skill, which involves actually doing something, accounted for 39.4% of all actions. Most of the rest were coordination tasks: sending a message to the human requester (27.7% of actions), leaving an initial comment (13.2%), giving instructions to a human analyst (9.8%), or waiting (8.8%). Running entire workflows accounted for just 1.1% of actions.
AI is building “scaffolding” that wraps around meaningful changes, often planning far more scenarios than the agent will execute. Typically, agents map out 15 possible actions but run only two. “The agent maps the paths a request could take, then walks down the path that makes the most sense as it meets reality,” the study noted.
Where Agents Succeed and Fail
While AI agents are making inroads into various IT functions, their success varies significantly depending on the task complexity and data quality. IT automation typically involves analyzing tickets and moving them along, which are low-risk tasks that agents handle well.
Agents do participate in areas like security, albeit only about 6% of the time. Their role in security is most notable in adding and removing people from groups or channels, unlocking accounts, resetting passwords, analyzing multi-factor authentication (MFA), provisioning or deprovisioning accounts, and assigning software licenses.
However, this identity-lifecycle work is where agents failed the most, particularly in onboarding and offboarding and identity-access management (IAM). The study found that “hardware and connectivity changes rarely fail; identity-lifecycle changes fail three-to-nine times as often.” This highlights a critical area where human oversight remains indispensable.
Matt Peters, Fixify’s co-founder and CEO, emphasized the importance of this hybrid approach. “That may sound less dramatic than replacing the help desk,” Peters wrote in a blog post. “It’s also a much more credible path to changing how IT work gets done.” He pointed to an example where an AI agent identified which team needed access to process a high-volume type of ticket. Rather than fully automating the process, the agent did the initial triage, asked questions, and then routed tickets to the team that had the information to act immediately.
“We didn’t need a world-ending hive mind,” Peters said. “We just needed to point a little conversational intelligence in the right direction.” This example illustrates how AI can enhance human efficiency without attempting to replicate human judgment in complex scenarios.
Why AI Breaks Down
Thanks to human-in-the-loop controls, Fixify was able to analyze scenarios where agent recommendations diverged from human judgment. This occurred about 23% of the time, providing valuable insights into the root causes of AI failures.
The largest failure category, accounting for nearly 50% of issues, was “target not found.” This means the agent couldn’t uncover what it needed, typically due to poor data. A user, group, account, or resource was not where the system expected it to be. When people change teams, groups are restructured, accounts are renamed, or work has already been done but not reflected in the system, this is more of an identity hygiene problem than an AI problem. The system needs cleaner and more current data.
Invalid inputs accounted for around 29% of failures, followed by unhandled errors, denied permissions, or invalid operations or configurations. The latter signal “real breakage” in integrations, according to Fixify. These failures indicate that while AI can handle routine tasks, it struggles when faced with incomplete or inaccurate information.
The study also noted that AI becomes more sophisticated over time, even if it might take a while. In hybrid systems, humans keep the most consequential changes under their own control, and iterative rejection and approval helps AI learn. Over time, agents’ plans get leaner and they start to re-plan when conditions change, rather than pre-planning all kinds of scenarios that may never occur. “That’s a sign of sophistication,” the study said. “Adapting in the moment is a more advanced behavior than trying to pre-script every contingency.”
In turn, humans second-guess the system less often and feel comfortable handing off more work. Instead, they control how agents behave, make high-impact decisions, and handle exceptions. “The hardest requests remain human-heavy, especially those that require repeated replanning or contextual judgment,” the study said.
What This Means for You
As agentic AI becomes embedded in more workflows, enterprises must evolve to accommodate this shift. This means investing in clean identity data and building strong playbooks, review workflows, and reliable integrations. Teams should judge agentic tools by their supervision loop and view rejections as a training process.
Analyst time, queues, and metrics should be built around reviewing proposals. Agent replanning can be seen as a routing signal: a single replan might indicate healthy adaptation, while repeated replanning means ambiguity, irrelevance, or unclear policies. “Make the review surface easy to understand so analysts can assess proposed actions and make quick decisions about how to proceed,” the study advised. “This is where the analyst’s attention belongs.”
For IT professionals, the key takeaway is that AI agents are powerful tools for handling routine tasks, but they are not a replacement for human expertise. By focusing on data quality, robust integrations, and effective human-in-the-loop processes, organizations can maximize the benefits of AI while minimizing risks. The future of IT ops lies in a collaborative model where AI and humans work together, each playing to their strengths.
Source: Computerworld
Over to you: Are you seeing your IT team adopt agentic AI tools yet, or is human oversight still the bottleneck?
You may also like
Microsoft Admits It Uses Windows 11 to Push Edge, Bing, and Copilot; Alliance Calls It Out
Inside Microsoft’s Windows 11 Vision: Marcus Ash on Design, Performance, and Legacy Compatibility
Microsoft Executive Modifies Doom to Play Inside MS Paint
Train Sim World 6 Drops Support for Xbox One and PS4: What You Need to Know
