Privacy without keeping the archive
You shouldn't have to choose between useful AI memory and a pile of raw recordings. We transcribe, extract, and then delete the audio.
There is a quiet assumption behind most meeting tools: to give you memory, they must keep your recordings forever. Every sensitive call, every offhand remark, sitting in someone's storage indefinitely. That is a liability you carry without choosing it.
It is also a false tradeoff. You do not need the raw audio to have great memory. You need what the audio meant.
Ephemeral by design
In qpNote, raw media lives only in a temporary buffer. After an entry is processed, the audio is kept for a short retention window (so you can re-transcribe if something goes wrong) and then automatically deleted. The deletion is logged, and the interface tells you plainly: audio is not stored.
Text notes have no media at all — the note body is the source, so nothing is buffered.
What actually persists
After processing, the durable record is the memory, not the microphone feed:
- Transcripts and text-note bodies (the source text).
- Derived memory — summaries, decisions, tasks, risks, and entities.
- The links that connect them back to their source.
That is everything you need to search, ask, and revisit — and none of the thing you would least want leaked.
Trust is more than deletion
Privacy and trust go together. Every answer in qpNote cites its source and declines when the evidence is thin, so you are never quietly misled. Every object is scoped to its workspace, queries filter by membership, and secrets stay out of source code. Deleting the audio is the headline; grounded, isolated, honest answers are the rest of the story.
Keep the memory. Drop the archive.
Want to try it?
Join the alpha