What "local-first" actually means for a musician
Local-first means the camera and the microphone are read on your own device — the footage of you practising never leaves the room.
There is a particular kind of pause that happens when a camera is pointed at you. You are about to play, and some quiet part of you wonders where the picture is going. For most software the honest answer is: somewhere else — a server, a company, a place you can’t see. Local-first is the answer that closes that loop. The video and the sound are read, measured, and then let go, all on the machine in front of you.
Where the footage actually goes
When you practise with OuterPractice, the camera feed and the audio are processed on your device, in the moment, frame by frame. The software looks at the picture, notes what it finds — a shoulder lifting, a jaw tightening, a beat landing early — and keeps only those observations. The picture itself is not saved and not uploaded. Nothing is sent to a server, because there is no server in the path. What stays behind is a timeline, not a recording.
Why this matters at the instrument
Practice is unguarded by nature. It is where you sound worst, where the body braces and slips and tries again. That is exactly the material no one wants leaving the room. A tool that watches the body is only useful if you can forget it is there, and you can only forget it if you trust where the picture stops. Local-first is what makes the forgetting possible: the camera can be honest with you because it has nowhere else to be honest.
The recording you worried about was never made. Only the reading of it remains.
What local-first is not
It is not a privacy promise written in a policy you have to take on faith. The difference is architectural: the data has no route out, so there is nothing to leak, intercept, or quietly change the terms on later. It is also not a limitation. Processing on the device means the timeline is ready the moment you stop, with no upload to wait for and no connection to depend on. The room is the whole system.
None of this changes what you do at the instrument. It only settles the question underneath it — of where the picture goes — so that the question stops sitting between you and the playing. The camera watches, the device reads, and the moment you close the session, the only thing left is what you chose to keep.