Terrain
Invite-only since October 2026.
Terrain is for women told their results are normal while their bodies say otherwise. The promise: understand your gut, try one thing, see if it works. Food comes first and supplements second, and every claim carries an evidence grade, with studies done mostly in women counted higher and studies a manufacturer paid for on its own product counted lower.
The story
I believe care should be designed around women, and that a plausible story is not proof. I wrote a 77-page product spec, locked the scope, and built the app and its research engine with AI, screen by screen, checking each one in the browser.
It went into production in October 2026, invite-only. My own product review on 26 Oct said the build was ahead of the evidence: no outside users yet. So the next step is not more features. It is talking to women and running one small test.
- Launched
- October 2026, invite-only
A few tips
- Lock the scope in writing before the AI writes a line of code.
- The rules decide, the AI writes: fixed rules pick the advice and the model only puts it in plain words.
- Log the cost of every model call. One full report cost about $1.82.
- Look at each screen before the next one gets built.
Related files on this desktop
Try it yourself
Copy this prompt and paste it into ChatGPT or Claude to try it yourself.
I'm looking at Shweta Kamble's Terrain on whatarewebuilding.ai: a women's gut-health app where you share your results or describe what's going on, get it explained in plain words, and get one thing to try. Fixed rules pick the advice, the AI only puts it in plain words, and every claim carries an evidence grade. Help me plan a starter version of an explainer app for a topic I know well. Ask me who it's for and what confuses them, then write a one-page scope, the rules that decide the advice, and the first screen. The gotcha Shweta hit: the build got ahead of the evidence, with no outside users yet. Plan who I'll show it to before planning features. Ask me for anything you need.
Tools I used for this project
Claude Code: Built both codebases, the app and the server, screen by screen.
Claude: Researches each question with web search, writes the plain-language copy, and answers questions in the app.
Gemini: Reads the uploaded PDF or photo of a lab report.
Expo: The app framework, tested in the browser first.
Render: Hosts the app and server.
