Brian Balfour

<start of post>

I’ve spent the last decade obsessed with how teams build. From early-stage startups to scaling Reforge, the pattern is always the same: we spend 80% of our time on the "how" and only 20% on the "why."

But the world just shifted.

Last week, I sat down with a group of founders to look at how they’re integrating AI into their daily workflows. It wasn't just about automation. It was about a fundamental change in the stack.

So I decided to run an experiment.

I took our internal documentation—years of product specs, strategy memos, and user feedback—and ran it through a custom agent built on Claude.

The results were startling.

It didn't just summarize the text. It identified three major gaps in our Q4 roadmap that we had completely overlooked because we were too close to the problem.

A redundancy in our onboarding flow that was costing us 15% in drop-off.

A misalignment between our marketing copy and the actual feature set of our latest release.

A "ghost" persona we were still designing for, even though the data showed they had churned months ago.

These weren't things a human couldn't see. They were things a human didn't have the bandwidth to look for while shipping at 100mph.

This is the "Full Stack Builder" era Tomer Cohen talks about. It’s not about doing more with less. It’s about doing things that were previously impossible because the cognitive load was too high.

We’re just scratching the surface of what this means for product management.

If you’re building in this space, I’d love to hear what you’re seeing. Are you using agents for strategy, or just for code?

Come join the conversation at Canvas 26 next month. I’ll be sharing the full breakdown of our experiment and the prompts we used to get there. Link in comments.

Joe, Sarah, Mike, Elena, and the rest of the crew—thanks for the push on this one. ⚡

<end of post>