← The Lab

Voice-to-Structured-Data

Not built yet · Phase 7

A voice note or a meeting recording becomes structured data: amounts, dates and categories pulled out into JSON and CSV, with the source phrase kept next to every extracted field.

What it will do

  • Amounts, currencies, dates and categories extracted per phrase
  • Confidence shown per field — you see what the model guessed
  • JSON and CSV output, transcript kept alongside

Anonymous

Up to 3 minutes

Vault

No duration cap, direct push to an MCP server

Nothing to try here yet. What already works: