You built an app with AI — maybe in a browser builder like Lovable or Bolt, maybe with an agent like Claude Code — and you're never going to read the code. This series is how to get serious anyway: take control of your files, then keep the app alive with the memory files, standing rules, and recurring check-ups that stand in for the review you can't do. It's not theory: Groundwork itself is built this way, by one founder with Claude Code, and every practice here comes from our own rules file and decision log.
Part 1
Lovable and Bolt got you a working app. The ceiling you've hit is called control — taking it means GitHub, a local AI agent, and one afternoon of basics. Moving-out prompt included.
Part 2
Every session starts from zero, and the app gets less consistent each time. Three plain text files fix it — a map, a decision log, a rules file — plus a parking lot for later.
Part 3
Groundwork is built solo with Claude Code, and its rules file has earned every line the hard way. The rules that transfer, translated — plus a paste-ready starter block.
Part 4
You can't review the code, but you can commission an inspection. Four copy-paste prompts to run every month — including the one that would have caught our $41 mistake.
Part 5
The day a stranger uses your app, mistakes stop being free. Backups you've actually tested, knowing how to go back, and hearing about crashes first — with the audit prompt.
The app is the half no prompt can run a company for you on. For the other half — incorporating, pricing, scope, going to market — see the founder guides →