Advising

Two areas. Agent systems that act on someone’s behalf: trust, control, what a person still has to approve. And engineering orgs trying to become AI-native without losing whatever made them good — standards, reuse, developer enablement, the practices that decide whether AI-assisted code is still an asset in six months. What I’m offering in both cases is what a team builds differently because I was there, which is easier to show you than to promise, so ask about your case and I’ll tell you if I’m any use.

It arrives as “help us adopt AI”, or as “the agent works and we can’t ship it”, and either way it becomes time with your team — design reviews, working sessions, the argument about what “done well” means when a model wrote half of it. Whether it was worth anything shows up in one place: is any of it still happening six months after I’ve gone.