Privacy by default: how your audio actually flows through Dipper
A plain-English description of where your recordings go after you save a note.
Voice is sensitive. Here's what happens to a recording after you stop and save a note in Dipper.
1. The note is created in your account
Dipper creates the note immediately so you can see it in the list while processing is still happening. That gives you an optimistic UI state instead of hiding the note until every backend step is done.
2. Audio is uploaded securely for processing
Audio is uploaded to Cloudflare R2 using presigned URLs. That lets the app send the file directly to storage without exposing storage credentials in the client.
3. Transcription and summarization run in the cloud
A Supabase Edge Function downloads the audio from R2, sends it to Groq Whisper for transcription, then generates the note summary through our LLM routing layer. The resulting note stores the transcript, summary, key points, action items, and related metadata in your account.
4. Audio stays available for playback
We keep the saved audio in storage so you can play the note back later from inside the app. That same stored audio is also what allows retrying failed processing or revisiting the original recording.
5. Deleting a note removes its stored audio
When you delete a note, Dipper deletes the note row and attempts to remove the associated audio objects from R2 as part of the same flow. Related embeddings used for chat are deleted too.
What we don't do
- We don't train AI models on your notes.
- We don't sell your note data.
- We don't expose your recordings publicly.
If any of this is unclear, email us. We'll answer.