How it works
From webcam frame to posture score.
Plain-English, no marketing dust. Here’s exactly what happens every second you’re at your desk — all of it on your computer.
The pipeline
Four steps, all on-device.
Camera frame
Sitful pulls a frame from your webcam at ~15 fps — never uploaded, never stored.
On-device pose
MediaPipe extracts 33 body landmarks in WebAssembly. The frame is discarded immediately.
AirPods fusion
If supported AirPods are connected, head pitch fuses into the forward-head component for a true reading.
Score & store
Components combine into a 0–100 score — one row per second, written only to a local SQLite file.
See it live
Watch the score react in real time.
Lean in and it drops; ease back and sit tall, and it climbs. Real keypoints track your head and shoulders, and a calm nudge appears only when you genuinely need it.
Private by architecture
The only thing that leaves your computer is nothing.
No frames, no account, no telemetry. Pull the Wi-Fi and Sitful keeps scoring your posture — because there was never a server to send anything to.
Questions
How it works — FAQ.
Where does the pose detection run?
On your computer, in WebAssembly, using Google’s MediaPipe Tasks library. Frames are never sent anywhere.
Where does my data live?
In a SQLite file inside your app data directory. Export everything as JSON or clear it from Settings in one click.
What about the model file?
MediaPipe’s pose-landmarker model downloads once on first launch. After that, there are no network calls during tracking.
Sit better,
without thinking about it.
7-day money-back guarantee. Your data never leaves your computer.