Voice everywhere

An idea. From the owner's voice memos of 13 November 2026, written up by the agent.

Never lose a recording

Losing a voice memo is the worst thing this app can do. A person who spoke for ten minutes while walking will not say it again the same way. Everything else can fail and be retried: the transcript, the debrief, the vault, the lane. The recording cannot.

This is why the screen shows the words and the level as you speak. It is not decoration. It is the evidence that the recording is being heard and kept, and it is what lets someone trust the app enough to stop checking it.

It is also why the note exists before the first word and why the recording is kept in a form that survives the app dying. See the design, where capture without loss is the first job.

Voice as the way in

Recording should not be a place you go to. It should be available from wherever you are: one control, on every screen, that starts a voice memo and returns you to what you were doing. The memo then goes where it should: the library, a vault, a lane to someone else.

The workflow that matters most: speaking while driving or walking, hands and eyes elsewhere, and trusting that it was captured; then, later, giving that memo to a particular person or agent.

More than one way to show it

There is no reason to pick one interface before trying several. One of them might look less like a recorder and more like an assistant that is listening: a single visual that shows it hears you, in the manner of the assistants in films. The lab is where these are compared, on the phone, the tablet and the laptop.

What is built

Recording with the words and the level as you speak, and a note that exists from the start. Not built: a control on every screen, starting from the Action button or Siri, the other interfaces. Recording with the screen locked is declared and untested.