Every team will have agents. Who leads them?
Within a few years, most teams will work alongside AI agents the way they work alongside spreadsheets today: quietly, and everywhere. The interesting question is no longer whether. It's who leads them, and how.
From tools to teammates
Agents draft, sort, check and route. They are fast and capable, and they make confident mistakes. That isn't a reason to keep them out. It's a reason to manage them the way you would any new member of the team: with a clear role, a reviewer and room to grow.
What changes for leaders
Ownership. Every agent needs a named owner: the person who reviews its work and answers for the result. An agent without an owner is a liability, however good it is.
Judgment becomes the job. As routine work moves to agents, taste, decisions and relationships matter more, not less. The people who hold them become more valuable.
Waiting is the target, not people. The best uses of AI remove handoffs, re-keying and queues. Cutting people cuts the knowledge that made the work good in the first place.
What production taught me
I spent my career building pipelines and tools for creative teams, first at Lucasfilm and Pixar, then for twelve years at Riot Games. The tools that lasted were the ones the team asked for and owned. The ones pushed onto people were quietly routed around. Agents will follow the same rule: change works when a team pulls toward it.
Where to start
- Pick one workflow where work waits.
- Map it with the people who do it.
- Give the agent a named owner.
- Prove it in weeks, not quarters.
- Keep the keys: your accounts, your data, your knowledge.
The future of work is still human. The teams that do best will be the ones that learn to lead agents well, and keep their people at the center.