Prehension
Prehension provides a temporal gesture recognition SDK for Unity that runs entirely on-device on Meta Quest 3 and 3S headsets, recognizing both static poses and dynamic motion sequences like swipes, grabs, and throws. The SDK uses custom AI models trained to work across different hand types and movement styles, with confirmed performance at 72 Hz with GPU headroom for higher frame rates.
- Augmented & Virtual Reality
- Developer Tools
- Artificial Intelligence
- Software Only
- Gaming & eSports
Funding
Founders
Founder details are not available yet.
Product
Problem
Developers building for Meta Quest headsets typically rely on basic hand presence or static pose detection, which cannot distinguish complex gestures defined by motion. Adding custom gesture recognition from scratch requires significant computer vision expertise and can introduce latency or network dependencies that degrade the immersive experience.
Solution
Prehension offers a temporal gesture recognition SDK for Unity that recognizes both static poses and dynamic motion sequences in real time on Meta Quest 3 and 3S devices. The SDK runs entirely on-device without an internet connection, using custom AI models trained to recognize gestures across different people, hand types, and movement styles. This enables expressive gesture controls such as swipes, grabs, and throws that go beyond simple pinch or thumbs-up detection. Developers integrate the plugin directly into Unity projects, with testing confirming performance at 72 Hz with GPU headroom for higher frame rates. The platform also supports rapid iteration, allowing developers to rerecord gestures, train a new model, and experiment with new gestures in minutes rather than days.
Target Audience
Primary customers are XR studios building immersive games and interactive experiences, enterprise training providers requiring natural hands-free interaction, and AR/VR developers on Meta Quest seeking more expressive gesture control beyond basic hand presence.
Features
- Temporal gesture recognition tracking hand movement through sequences of poses over time, enabling swipes, grabs, throws, and motion-defined gestures
- Custom AI models that recognize gestures across different people, hand types, and movement styles
- Fully offline, on-device processing with no internet connection required
- Unity SDK optimized for Meta Quest 3 and Quest 3S with confirmed performance at 72 Hz
- Real-time recognition of static poses and dynamic motion sequences in a single package
- Rapid iteration workflow allowing developers to rerecord gestures, train new models, and experiment in minutes