Fixify Publishes New Report on How Agentic AI Is Changing IT
Analysis of nearly 18,000 agentic AI plans finds AI now handles 1 in 3 IT actions at organizations that have adopted AI automation
ARLINGTON, Va., Aug. 4, 2026 /PRNewswire/ -- Fixify, a provider of AI-native IT automation, today published How agentic AI is changing IT, a report analyzing how early adopters are using agentic AI to run IT. The report uses production data from more than 40 companies: 17,929 agentic plans, 147,351 individual plan actions, and 52,689 skill executions, captured from March through June 2026.
The full report, including an interactive version and downloadable PDF, is available at fixify.com/agentic-report.
Key findings
The report surfaces six major themes:
- AI agents now handle 1 in 3 IT actions, all under analyst supervision. Agent involvement was highest in software, applications, security, and collaboration work. Analysts stayed closely involved in higher-stakes areas like identity lifecycle, onboarding, offboarding, and hardware.
- Human oversight declined 27% to 16%, and it's training the AI. Analysts declined about a quarter of AI-proposed actions across the window, though that rate fell sharply over time. Each approval or rejection teaches the system what the team considers acceptable, turning supervision into training as well as oversight.
- 84% of system changes ran through the identity provider. Changing access is the most common thing IT teams use AI agents for, from group membership to credential resets and account provisioning. Most of that work runs through identity providers, such as Okta, making identity systems the central frontier for agentic IT automation.
- Roughly 3 in 4 automation failures are data problems, not AI problems. The largest failure category is target-not-found: the user, group, or account the agent tried to act on was not where the system expected it. Most failures trace back to stale identity data. Clean directories and timely deprovisioning improve reliability more than model tuning.
- Scale, not staffing pressure, predicts automation. The most stretched IT teams were not the most automated. Larger organizations were. The median AI-executed share ran from 26% in the smallest third of companies to 42% in the largest, a pattern that tracks with operating maturity and request volume rather than headcount pressure.
Supporting quotes
"The story here isn't 'AI replaces IT teams.' It's a new division of labor. In a well-designed system, agents handle the routine and not-so-routine work that you measure by volume, while analysts move up the stack to exceptions and improvements -- work that you measure by value," said Matt Peters, co-founder and CEO of Fixify.
"When agents fail in IT, the instinct is to blame the model. Our data says something different," said Peter Silberman, co-founder and CTO of Fixify. "Forty-eight percent of the time an agent failed, it was because the user, group, or account wasn't where it expected. That's a directory problem. One of the most impactful things you can do for automation reliability is to treat your identity data like the automation infrastructure it's become."
Upcoming Webinar
Fixify will host a live walkthrough of the report's findings on Thursday, August 27 at 1:00 PM ET. Registration is available at fixify.com/agentic-report-live.
About Fixify
Fixify helps IT leaders remove work from their team's queue by automating 75% of Tier 1 IT work across 40+ common IT requests. Our AI-native automation platform plugs into tools like ServiceNow, Jira, and Freshworks. It builds playbooks tailored to your organization, and then handles requests wherever they come in – Slack, Teams, email, or your ITSM – so your team can focus on higher-impact work. Visit fixify.com.
Media Contact
Michael Evans
Chief Marketing Officer, Fixify
[email protected]
202-599-8018
SOURCE Fixify
Share this article