Fleet has published research showing that enterprise Mac administrators are using AI-generated code and configurations in production environments. It also found that when changes go wrong, rollback times often stretch from hours to days.
The findings are based on a survey of more than 250 enterprise IT practitioners who manage Apple devices.
According to the research, 86% of respondents allow AI-written output to reach production devices after some form of review. In the previous 30 days, 61% had used AI to write a mobile device management profile or configuration, 62% had used it to write scripts and code, and 68% had used it for troubleshooting or log analysis.
Those figures coincide with slower recovery times when changes fail. Nearly 70% of respondents said they could not roll back a bad configuration change within an hour, while 43% said it takes more than a day. Another 26% said they could recover on the same day, but only through manual intervention.
The results point to a mismatch between how quickly teams can create and deploy changes with AI tools and how slowly many organisations can reverse them across managed Apple estates.
"AI can help push changes out in seconds, while a mistake can drain hours of valuable IT time and resources," said Allen Houchins, Chief Information Officer, Fleet.
"Automating work with AI isn't inherently bad, but human accountability and judgment is still critical. Teams need a clear record of what changed and a fast way to undo it when something goes wrong. Speed is only an advantage if the pace of recovery matches," said Houchins.
Admin to engineer
The report presents Mac administration as a role moving closer to software engineering. It found that 57% of respondents regularly use large language model assistants, 46% use a code editor, and 42% work with CI/CD runners. AI assistants are now used more often than mobile device management consoles, cited by 53% of respondents.
That shift appears to be influencing how practitioners view career progression. Nearly two-thirds, or 64%, said they already see a defined path from administrator to senior and staff engineer.
Some teams have adopted workflows more commonly associated with software development. The survey found that 28% combine human review of AI-generated changes with version control, giving them a record of what changed and a route to rollback.
"Mac admins are basically engineers now, whether their job description calls them that or not," said Mike McNeil, Chief Executive Officer and Co-Founder, Fleet.
"They're writing code, using AI and shipping changes to production, but they usually don't have the safety net engineers take for granted. Software teams have Git. They can see what changed, review it and revert it. Device management should work the same way. Then IT can actually take advantage of AI without relinquishing control," said McNeil.
Apple complexity
The survey also examined Apple's management framework. It found that 78% of respondents are using or piloting declarative device management, suggesting broad uptake of newer approaches to Apple fleet management.
Even so, administrators said keeping pace with Apple remains difficult. Nearly a third, or 31%, identified Apple's release cadence as the hardest part of Mac administration.
Most respondents also work beyond Apple-only estates. Just 8% said they manage Apple devices exclusively, indicating that most teams operate across mixed-device environments that may include other desktop and mobile platforms.
That broader remit adds to the complexity of device management, especially when AI-written scripts and settings are introduced into production. In mixed estates, a slow rollback process can complicate incident response and operational continuity because teams may need to trace issues across different operating systems and management tools.
Fleet argues that managing configuration as code provides a more structured way to handle these risks. It says version control and review processes give IT teams better visibility over changes and make mistakes easier to reverse.
The data suggests such practices are not yet standard across the wider market. While many Mac administrators are already using AI in day-to-day work, fewer appear to have paired that use with the engineering controls commonly seen in software development teams.
The report highlights two overlapping shifts in enterprise IT: the growing use of AI to generate operational changes, and the gradual redefinition of endpoint administration as a coding and automation role. According to the findings, the challenge is that management systems and rollback processes have not advanced at the same rate as the tools now used to create and deploy change.
Among respondents, that gap is most visible in the contrast between rapid AI-assisted change creation and the fact that 43% need more than a day to reverse a bad configuration change.