AI Coworkers, Agent Registries, and Platform Stack Wars

The "persistent AI coworker" framing is becoming the dominant mental model for how PMs should think about agents — and today's signals from Lenny's newsletter, Ethan Mollick, and AWS all point in the same direction: the interesting product question is no longer "can AI do this task?" but "how do humans and agents share context, authority, and accountability over time?"