Transmutify turns the files your customers actually send you — messy Excel exports, CSVs with weird headers, raw text, even photos of tables — into clean data that matches your schema.
How it works:
1. Define your target schema (or let it be generated from an example).
2. Send a file via the API or upload it in the playground.
3. AI proposes the column mapping to your schema.
4. Every run returns a hosted review URL: whoever knows the data checks the mapping, fixes values, and approves. No AI output enters your system unreviewed.
5. You receive clean, schema-conform records via webhook or download.
Built for B2B products where customer onboarding means "please send us your employee/user/product data" — and someone ends up massaging spreadsheets by hand.
What makes it different:
- Bring your own AI keys — it runs on your OpenAI/Anthropic account, so your data doesn't route through our AI contracts and you control model and cost.
- Bring your own storage — files can live in your S3 bucket.
- Human-in-the-loop by design — AI proposes, people approve.
- API-first — your app owns the upload and the results; Transmutify handles mapping, review, and delivery. Dry runs let you test the whole flow without provider calls.
- AI-agent-ready docs — Claude Code or Cursor can integrate it for you in an afternoon.
Free to use right now — try a transformation in the browser playground, no code or payment details needed.
