Three small sites use the same AI-assisted build as their source. One shows the build, one turns it into guidance, and one turns it into critique.
The useful thing about these three sites is that they are recognisably related, but each has now been checked against its own brief: source record, collaboration guide, and critique. They show how the same AI work can be made to feel like a build record, a helpful lesson, or a warning.
This is the build-story version. It turns the AI-assisted app-building transcript into a sequence of cards about decisions, fixes, and product direction.
It is closest to the working process: what was asked for, what changed, and how the website took shape.
Open v02 →This is the teaching version. It keeps the same structure but makes the dialogue warmer, clearer, and more useful as a guide to working with AI.
It turns the build into practical advice: how a human gives direction and how AI helps shape it into something usable.
Open v03 →This is the critical version. It uses the same kind of card format to argue that AI can create false confidence, hidden supervision work, drift, and risk.
It is not just another polish pass. It reframes the same process as a case study in what can go wrong.
Open bad ai →The basic site idea stays stable: cards, sequence, story, and about sections. What changes is the meaning. The content moves from process notes, to helpful instruction, to criticism of AI-assisted work.
The comparison matters because it shows that an AI project is not only shaped by code or layout. It is shaped by framing. Change the labels, tone, and emphasis, and the same work can look like progress, guidance, or a warning.