1. Check authentication

2. Upload audio

Generate a UUID for the note, then submit the audio file:
The API returns 202 Accepted while processing runs asynchronously.

3. Poll transcription

  • 200 means transcription is complete.
  • 202 means processing is still active.
  • 422 means transcription failed; read the response detail.

4. Generate Insights

After transcription succeeds, send the transcription object to the Insights endpoint:
See Audio to Insights for a complete Python example that performs this flow automatically.