I’ve been spending a lot of time thinking about, talking about, and building AI agents at Google - and one question I get asked a lot is:

How do you think about agent design as a PM?

I’ve been working on this post for a while because I wanted to give it the depth and care it deserved. It's a distillation of some of the core product frameworks that help me navigate this fast-evolving space. Frameworks that help shape how I think, and that I use to move from vague concepts to concrete product designs.

Here are 5 product frameworks I keep coming back to 👇