Import from ChatGPT & Claude
Native support for both export formats. Upload the JSON you already have — no API, no OAuth, no scraping.
WhyChose
WhyChose
Paste your ChatGPT or Claude export. Get back every decision you worked through with AI — structured, searchable, shareable with your team.
Pre-launch. Building in public at @bitinvestigator — every session logged to /changelog.
The problem
You use ChatGPT or Claude 20+ times a week to think through architecture calls, pricing, hiring, trade-offs. The ideas live in your chats — but the decisions, the actual forks in the road, are buried across 300+ conversations. A new engineer asks why you picked Postgres over Mongo. A board member asks how you landed on $49 pricing. You remember the shape of the argument, not the reasons — and CMD+F across three AI tools doesn't surface them either.
/decisions/0042.md. It went stale in two months because writing them by hand is the first thing that falls off the backlog.How it works
One-time click in ChatGPT or Claude: Settings → Data Controls → Export. You get a JSON file of your conversations — same file you'd get for GDPR anyway.
Drop the JSON into WhyChose. Extraction runs in your browser — we never store the transcript, only the 5–50 decision records we surface.
A searchable log of decisions with the original chat snippet, date, and trade-offs considered. Send a private link to a teammate in one click.
What you get
Native support for both export formats. Upload the JSON you already have — no API, no OAuth, no scraping.
We don't index every message. We extract the 5–50 durable calls from a quarter of chats — with trade-offs listed and the original snippet attached.
Filter by date, category, or keyword. Find "why we picked Postgres" in two seconds — with the reasoning thread still attached.
One click creates a private link the next engineer or investor can browse. No account required to read. Great for onboarding.
Pricing
$0/mo
Solo — just trying it out.
Most popular
$9/mo
Engineers & founders with a quarterly habit.
$29/mo
5–50-person SaaS teams, up to 20 seats.
Questions
No. Extraction runs in your browser. We persist only the 5–50 decision records we surface — short strings like "chose Postgres over Mongo, Jan 14" — not the raw transcript. The JSON you upload is never written to our disk as durable storage.
Those are empty notebooks — you have to write into them, and you won't. WhyChose starts from the thinking you already did in ChatGPT or Claude and extracts the decisions automatically. Output still exports to Notion, Obsidian, or Linear on Pro.
None of the major AI chat platforms expose a conversation-history API today, so real-time sync isn't possible. A quarterly batch matches how teams actually review decisions — end of quarter, end of sprint, before a board update — without asking you to grant OAuth to your entire chat history.
A fork in the road with real trade-offs: stack choices (Postgres vs Mongo), pricing calls ($49 vs $29), hiring judgements, architectural migrations, vendor selection. Not clarification questions or scratch thinking — just the durable calls that shaped what you built.
A solo builder who got burned as the new CTO at a Series A, couldn't answer why the stack was picked, and realised the reasoning lived in someone's ChatGPT history — six months of scroll. Building in public at @bitinvestigator, with every ship logged at /changelog.
Early access opens this quarter. Free tier at launch, no credit card. One email when it's live.
Get early access