Edge AI · Wearables
Gestures, activity and motion classified by the wearable's own microcontroller. No phone, no cloud.
Swipe, tap, rotate, flick from accelerometer and gyroscope.
Walk, run, cycle, stairs, rest. No phone attached.
Sleep stages and movement from the wrist. Data stays on the chip.
The model runs on a Cortex-M0: no bigger MCU just for the AI.
The binary is locked to device UIDs: it can't be extracted or cloned.
CSV, train, download. No data scientist.
Accelerometer and gyroscope for each class, as CSV.
AutoML picks the engine and quantises for the MCU.
It runs only on the devices you register.
Link the C library into your firmware. Zero dependencies.