Loading projects…
A running gallery of things I've built — small apps, trackers, and essays, each shipped from my phone and live right now. Tap any to open it.
Almost everything above was built by an AI agent working from a todo list, then reviewed and shipped from my phone. Here's the path a change actually takes.
It reads a todo list, does the work, and logs what it did. I review from my phone and grade the runs, which is how it learns what's worth building.
A structured store of context — projects in flight, decisions already made, preferences — so each run doesn't start from zero.
Small serverless endpoints backed by Redis. No machines to patch, no migrations, no ops. Git push is the whole deploy pipeline.
I'm a product manager. I can read code and tell when something's wrong, but the agent writes it. The interesting part is the review loop, not the typing.