In my latest article on Substack I've written about how I have become a manager of AI agents. My home AI setup now has a hierarchy. There is Number One - the agent I mostly speak to. That agent runs a fleet of agents with specific roles. Advisers. Red-teamers. And there is one agent whose entire job is giving the rest of them blunt feedback. But the strange part is that it feels striking familiar. It feels like running a clinical teaching unit. On Substack: why physicians will become managers of AI, and why we should start teaching those skills overtly: samirgrovermd.substack.com/p…