How it works
The phone is the sensor. AI is the engine.
Presence is sensed at the edge, resolved by AI in a clean room, and returned as anonymized, deterministic insight — nothing identifiable ever leaves the room.
- Collect — The TV emits a rotating, anonymous Bluetooth token. Nearby opted-in phones detect it; Bluetooth proximity plus motion, angle, and second-screen behavior confirm an active human is present — no camera, microphone, location, or identity.
- Process (AI) — A per-TV machine-learning model calibrates detection to each room, sensor fusion separates a real co-viewer from a phone in the next room, and false positives are removed continuously. Device identity stays blind to all parties; every result sits behind a floor of 100+.
- Extrapolate — The deterministic signal projects and enriches panel-based demographics across the full detected footprint, delivered as anonymized reporting and segments via API or scheduled files. A sample becomes scale.