WhyChose

About

Built in public · last updated April 25, 2026

WhyChose is a one-shot decision extractor for engineers and founders who do their thinking inside ChatGPT and Claude. You paste an export, you get back a structured log of every decision you worked through — stack picks, pricing calls, hiring forks — with the original chat snippet attached and a share-by-link for the next teammate who asks "why did we pick Postgres over Mongo?"

Why this exists

I joined a Series A as the new CTO and could not answer why the existing stack had been chosen. The reasoning wasn't in any ADR. It wasn't in Notion. It was buried somewhere across six months of someone else's ChatGPT history, behind 300 conversations of clarification questions and scratch thinking.

I wrote a regex pass to pull the actual decisions out of an export. It worked well enough that I shipped it as a tool. The hosted product wraps the same engine; the engine itself is open source at whychose.com/extractor — MIT-licensed, ~500 lines of Node, zero dependencies, runs fully in your browser.

Who's behind it

A solo builder. I'm @bitinvestigator on X — DMs are open. Every shipped session is in the changelog and every product decision is itself logged the same way the tool would log them. Build-in-public on hard mode.

What's deliberately not here

WhyChose is not a sync tool. It's not a note-taking app. It does not require an OAuth grant on your chat history (no platform offers that API anyway, which is the whole reason this is a quarterly batch instead of a daemon). It does not retain your raw transcripts after extraction; only the 5–50 short decision strings are stored. See Privacy for the specifics.

Where to go next