Now available
Video surveillance · AI vision
For camera networks that outgrew their NVR
A self-healing recorder and AI video platform for large sites. Cameras are declared in one version-controlled file rather than clicked into a UI. Recording is hardware-accelerated and connects to each camera exactly once. Storage and database maintenance run themselves. And AI is switched on per camera, so you pay for inference only where it earns its keep.
Proved on a 150-camera site recording continuously to a 100 TB array — including the unglamorous parts: a disk-fill safety net, a database watchdog, scheduled camera clock sync, and repair tooling for when hardware misbehaves.
- Per-camera detection, zones, masks and object filters
- Theft and shrinkage detection built from zones, paths and dwell rules
- Non-standard behaviour monitoring — after-hours activity, loitering, wrong routes
- Gesture, idle-space and motion rules, plus custom events over MQTT and webhooks
- Semantic search — find footage by describing it
- Face and licence plate recognition where it's lawful and useful
- Generated descriptions, so overnight events can be read not watched
- Inference on your own hardware — footage stays on site
- One CLI for provisioning, diagnostics, backup, repair and capacity
Object Detection
Zones & Masks
Semantic Search
Face Recognition
Plate Recognition
GPU Acceleration
Self-healing Storage
Multi-site