The real challenge: integrating agents into a company

We have 9 AI agents on the team right now. We're almost where we wanted to be.

To be honest, it took me a while to really get the point of AI agents.

For me, it's no longer about the technology or the capabilities. The tech is already good enough. Better than that, even.

The real challenge is somewhere else: integrating agents into a company, figuring out how humans and agents can actually work together, beyond just having an AI that helps you get things done.

It's been (still is) an incredible journey. And it took changing course completely, more than once: rethinking the vision, correcting my own expectations several times about what agents could and should do.

A lot of what you see with agents is amazing to watch, but barely useful in the real life of a company at the end of the day. Flashy, more apparent than real.

The serious value only shows up once you get past that, when you stop chasing the spectacular and start asking what's actually useful.

That's the bar I've been working toward, with three goals in mind:

1) An unquestionable, objective value from having agents on board.

2) To complement humans, not replace them. (If they replace us, we settle into mediocrity — we'd only be as good as our AI, and therefore no better than anyone else with the same tools.)

3) To design a system as a company: optimizable, replicable, scalable.

This has changed how we operate as humans and as a team. It's changed the skills we need in our day-to-day work.

Even job titles are being called into question.

We've rethought every single process in the studio, designing a hybrid team of humans and agents.

The whole thing has been a rollercoaster… and I'd love to tell you more about it soon.