About
About Accrete
Accrete started as tooling the founder built to see his own AI-coding sessions clearly. The story, the conviction, and what it became.
I'm Mike, the founder. Accrete exists because I wanted to see my own work clearly and couldn't.
The itch
I was going deep on AI-assisted engineering — long days inside Claude Code sessions, trying to get genuinely good at it. And I kept hitting the same wall: I had almost no visibility into my own practice. What was I actually doing in those sessions? Was I planning before coding or just telling myself I did? Which habits showed up in the sessions that went well and disappeared in the ones that didn't?
The transcripts had the answers, but they were sitting in files nobody reads, purged after about thirty days. So I built tooling for myself: parse my session history into a database, classify what happened in each one, keep it durably. Then look.
It changed how I worked. Not because the data flattered me — mostly it didn't. I'd believed things about my own habits that the sessions contradicted. That was the point.
The second itch
The other problem was teams. When I pushed for changes in how a team worked with AI, all I had was opinion — mine against whoever disagreed. Anecdotes don't move skeptical engineers, and they shouldn't. I wanted to be able to say "here's what the sessions show" instead of "trust me."
And I kept seeing the same waste everywhere: the most effective ways of working stay trapped in individual sessions. Someone on the team has quietly figured out something that works — a planning habit, a way of delegating to subagents, a saved workflow — and nobody else can discover it, because nobody can see inside their sessions. The knowledge exists. It just never travels.
What I believe
AI-coding skill is learnable. It's not a talent some people have and others don't — it's a set of concrete, observable behaviors that show up in session data, which means it can be measured, and what can be measured can be coached. That's the whole bet. How it works describes the mechanics.
I still run Accrete on my own sessions every day, and I still read transcripts — mine and, with permission, other people's. That habit is where the practice catalog comes from, and it's where the findings we publish in research will come from.
Where this is
Pre-launch. No customers to name, no benchmarks to cite yet — I'd rather say that plainly than imply otherwise. What exists is the product I built for myself, growing into one teams can use.
If the itch sounds familiar, join the early access list. And if you lead a team and want to shape what Accrete measures, become a design partner — that means working directly with me.