I rebuilt this hub in an afternoon with an agent — here's the exact loop
The channel is back, as a full learning hub. The first build log is the site itself: how it got built, the loop that built it, and why 'shipped' is the only metric that counts here.
Welcome back. learn to build.ai used to be a YouTube channel; it’s becoming the thing I always wanted it to be — a hub for actually learning to build with AI, where every lesson ends in something shipped.
Fitting, then, that the first build log is this site itself. It was built the same way everything here will teach you to build: an agent doing the typing, a human directing the loop, and a hard rule that done means live.
The thesis, in one line
Everyone can start a project with AI now. Almost nobody finishes one. The tools scaffold ~70% of an app in minutes and the tutorials leave you exactly there — right before auth, a real database, a deploy, and the debugging that happens when it breaks off the happy path.
That last stretch is the whole game. It’s what I do for a living as a forward deployed engineer, and it’s what this hub is built to teach. Not “learn to prompt.” Learn to finish.
How this site got built
The loop was the same one Track 0 describes:
plan → act → observe → repeat- Plan first. Before a line of code: the positioning, the three doors (free hub → paid community → consulting), the tracks, the project ladder. A page of decisions is worth an hour of the agent guessing.
- Give context, then let it act. A clear spec of the design system — the palette, the type, the components — so the agent wasn’t inventing a look every file.
- Review the diffs. Every change skimmed, not rubber-stamped. The two habits that separate shipping from stuck: read what changed, and keep version control so any mess is one
git resetaway. - Ship it. Built, run locally, verified, deployed. The site you’re reading is the artifact.
Total hands-on time: an afternoon. That’s not a flex about speed — it’s the point. When the loop is fed well, this is normal now. The skill isn’t typing faster; it’s directing well and knowing where “done” is.
What’s here now, and what’s coming
- Track 0 is live and written in full — the mental models everything else stands on.
- The project ladder — all twelve builds, each with a real “done” criterion.
- Resources — the curated library, no affiliate padding.
- Tracks 1–6 are scaffolded and being written now, in order.
The cadence from here: one build log a week, each one a real build broken down with the numbers — and each one feeding a new resource or track lesson. If you want it in your inbox (plus the reference agent I’m open-sourcing), the signup’s at the bottom of any page.
Let’s build.