Fleet launches AI governance push for endpoint GitOps

Fleet launched an AI governance initiative at its First Fleet Virtual Summit, which focused on infrastructure as code for device management.

The summit brought together practitioners from Cisco, SandboxAQ, Treeline, Thumbtack and Somos to discuss how endpoint teams are moving device configuration from management consoles into Git-based workflows. Speakers linked that shift to faster remediation, clearer audit trails and a broader move to treat endpoint work as an engineering discipline.

Mike McNeil, Co-Founder and Chief Executive Officer of Fleet, opened by describing device management as one of the last IT functions still largely changed by hand in administrative consoles. He said teams that had moved device configuration into Git were working in a materially different way, with a standard place to record changes and recover from mistakes.

Several participants said the shift was not only about speed, but also about aligning endpoint operations with the way software and infrastructure teams already work. They described a model in which help desk tickets and support issues feed directly into engineering workflows rather than sitting in separate processes.

Fletcher Previn, Chief Information Officer of Cisco, said endpoint reliability directly affects employee experience and should be treated as part of core company infrastructure.

"The endpoint is something I have spent a lot of my career focusing on. If your most important asset is your people, the shortest path to engaging them is what is in their hand or on their desk. Yet many people do not devote much time or energy to it because they see it as a simpler, more pedestrian problem. But those quality-of-life issues answer the question: what is it like to work here?

"I have always tried to get desktop endpoint management teams and engineering teams working in an integrated workflow, so problems that appear in the help desk, phone calls or tickets flow naturally to the engineering team to resolve them. If IT can stop working like a help desk and start working like an engineering team, these become integrated workflows and problems do not need to wait hours or days to be resolved. They can happen in real time.

"There are other downstream benefits. When change management is no longer a morning meeting and becomes a pull request, time to resolve is much faster. The ability to apply AI and make dynamic changes in the environment improves. The need for multiple images for different types of employees and workflows goes away. Even in highly regulated spaces, you can have a more nimble, leaner, faster and more responsive environment where everybody is working like an engineer. It has long been the vision in the environments I have worked in to treat the endpoint as infrastructure as code," Previn said.

Operational shift

The discussion pointed to a wider shift in endpoint management, with teams increasingly using pull requests and version control to manage changes that would previously have been handled through manual review meetings or point-and-click administration. Participants said central change histories provided evidence for configuration decisions and reduced the risk of mistakes.

Previn also recalled how the limits of older endpoint management approaches could produce costly workarounds.

"I remember doing a divestiture of part of the business, and the proposal was to buy a second set of laptops for 50,000 people because we could not figure out how to manage those endpoints the way we wanted to. If we had the ability to manage that through Git, it would have been a totally different story," he said.

Betsy Kaiser, Staff IT Systems Engineer at SandboxAQ, said AI tools were helping endpoint teams respond to a faster pace of change, while Git-based workflows improved traceability.

"Things are moving at lightning speed, and AI has helped me keep up, even if it still feels like a fraction of the pace. It is also more secure, and I value the audit trail of changes because I can look up exactly what is happening to our configurations. We also have to keep up with other teams at the company, and they are all operating this way, so we have to work this way too," Kaiser said.

Scale pressures

The operational case was especially strong for service providers and teams managing multiple environments. Brock Walters, Platform Architect at Treeline, said GitOps workflows had become necessary for managing several Fleet instances while migrating customers from rival products.

That reflected one of the practical constraints discussed during the session: manual console work does not scale easily across large or fragmented estates. Somos, for example, manages Windows, Mac and Linux devices, underlining the breadth of the task facing endpoint teams with mixed fleets.

"I am now managing multiple instances of Fleet, and more every day, while migrating customers from other solutions to Fleet. There is absolutely no way I could do what I am doing unless I could make changes as quickly as I can using GitOps workflows. If I had to do all that clicking, I would spend the rest of my life clicking.

"So speed is not the only reason to do it, but for me it is enabling workflows that I do not think could happen any other way," Walters said.

Adam Quince, Manager of IT Systems Engineering at Thumbtack, linked the approach to better visibility into how engineering-led organisations use their devices. He said endpoint teams needed to understand user behaviour in environments where employees already work in code repositories and Git-based processes.

The summit agenda also covered AI-native endpoint engineering, open-source device management, AI agents and the longer-term direction of device management beyond 2027. Recordings of the sessions will be made available on demand, and the live event invited questions through the comments section.