Managing in the Age of AI
For most of my career, I understood management in a very simple way: deliver results through other people. That idea has been a core part of how leadership is taught and practiced for decades. Managers set direction, build teams, remove blockers, and help others do their best work. The expectation of management is slowly evolving. It’s no longer just about delivering through people. The real challenge is managing the output closely than ever before.
Engineers today walk into discussions with work that has been partially generated with the help of AI - code, documentation, and early implementation drafts. The first version of something that used to take hours or sometimes days to create can now be generated in minutes. With AI, teams can move faster and explore ideas more easily. But it also changes something fundamental about how work gets created and reviewed for managers. I remember a recent design discussion on my team where we were looking at a proposed implementation for a service change. Parts of the implementation had been drafted with the help of an AI coding assistant. The structure looked clean and the overall approach made sense, which made the discussion move quickly into deeper design questions. As we talked through the flow of the system together, we started exploring how the service would behave in different failure scenarios. The implementation handled the main path well, but some of the edge cases that matter in production systems needed more thought. Nothing unusual about that these kinds of conversations happen in almost every design review. What stood out to me afterward was how the thinking had shifted.
In the past, a lot of that reasoning would naturally happen while the engineer was building the implementation itself. Writing the code took time, and during that process people would think through edge cases, operational behavior, and system guarantees. With AI, the first version of the solution can appear almost instantly. The thinking still happens, but it often moves into the review and discussion phase instead of the creation phase. That shift is small, but it changes the nature of management in interesting ways. Managers used to focus on helping people do great work and ensuring the team delivered outcomes. Now there is another layer emerging: making sure the systems helping produce the work including AI tools are being used in ways that still preserve judgment and system reliability. When generating output becomes easy, verification becomes more important.
The manager’s job is no longer just delivering through people. Increasingly, it’s delivering through a system made up of people and machines, making sure that speed doesn’t outrun understanding. Personally, I’m still grappling with this shift, and I suspect many managers as well. The tools are evolving quickly, and the expectations around how work gets produced are changing with them. I’m still learning what it means to maintain the same standards of quality and accountability when part of the output now comes from systems rather than just people. My own thinking on this is still evolving, and I expect it will continue to as I gain more experience working alongside these tools.