Edge-AI watch platform
AI that stays on your wrist.
The large language model runs on the watch itself, so transcription, translation and recognition work offline and are processed on-device. A platform you can launch under your own brand.
Illustrative AI-generated image
Where the data goes
On-device
Microphone & camera
→
On-device AI chip
→
On your wrist
Processed on-device. Works with no network.
Cloud AI
Microphone & camera
→
Network
→
Cloud server
→
Back to the watch
Needs a connection. Billed per API call.
Offline + private
Offline and private by design
Offline
Works with no network
Transcription, translation and recognition run on the watch, so they keep working in a basement meeting room, on a flight or at a remote site. Responses don’t wait for a round trip to a server, and there is no per-call cloud API cost.
Private
Processed on-device
Recordings, transcripts and images are processed on the watch rather than sent to a cloud server for processing. That suits business-sensitive conversations and teams that prefer to keep their data close.
On-device vs cloud
How on-device AI compares with cloud AI
The same features, a different place to run them.
On-device AI
Cloud AI
Where processing happens
On the watch
On a remote server
With no network
Keeps working
Stops working
Latency
No network round trip
Depends on the connection
Running cost
No per-call API fees
Billed per API call
Recordings and images
Processed on-device
Uploaded for processing
Architecture
Two chips, two jobs
A low-power main controller handles standby, connectivity and sensors. A separate on-device AI chip runs the large model.
Chip A
Low-power main controller
Always-on standby
Bluetooth, Wi-Fi and 4G connectivity
Audio capture
Sensor management
Chip B
On-device AI chip
Runs the large language model
Speech transcription and speaker separation
Translation
Text and chart recognition from images
Use cases
Three jobs it does without a connection
01
Meeting notes
Local transcription with speaker separation, then key points and to-dos, processed on-device.
02
Visual assistant
Capture a whiteboard or document with the camera. Text and charts are recognised locally and linked to the recording.
03
Offline translation
Translate conversations on the watch, without a network connection.
Key specs
The reference hardware
Architecture
Dual-chip: low-power main controller + on-device AI chip
Memory
4 GB + 64 GB
Display
1.75″ round AMOLED, 466 × 466
Camera
12 MP, rotating and detachable
Audio
Three-microphone array with low-power continuous recording
Connectivity
Bluetooth, Wi-Fi, 4G, NFC, multi-mode GPS
Sensors
Heart rate, SpO₂, skin temperature, 9-axis motion
Software
Developer platform / SDK available for partners
MOQ, NRE, lead time
Quoted per project
Health sensors are for wellness features and are not intended for medical diagnosis.
For your brand
Launch it under your own name
An ODM/OEM platform, not a retail product. You decide how it looks, what it runs and how it works.
01
Your design
Industrial design and CMF (case, colours, materials and finishes) with your branding.
02
Your AI apps
Developer platform / SDK available for partners, with access to audio, vision, sensors and connectivity for custom on-device AI apps.
03
Your model and features
The model and features can be customised, with no dependence on pay-per-call cloud AI APIs.
FAQ
Questions brands ask
Does it work without internet?
Yes. Transcription, translation and recognition run on the watch’s on-device AI chip, so they work offline. Bluetooth, Wi-Fi and 4G are there when you need a connection.
Where is my data processed?
On the watch. Recordings, transcripts and images are processed on-device by the AI chip rather than by a cloud server.
Can we put our own brand and apps on it?
Yes. We work on industrial design and CMF with your branding, and a developer platform / SDK is available for partners building their own on-device AI apps.
What are the MOQ, NRE and lead time?
Quoted per project.
Build your AI wearable
Tell us about your brand, your users and the on-device AI features you need.
Early Access
Join Early Access
Tell us about your brand and how you plan to use the platform.
We’ll contact early-access partners with samples and program details.