One pipeline, from lens to responder. Every stage below runs on the cameras you already have, and every threshold in it is set per site.
One Trinocular AI pipeline, from lens to responder.
Continuous capture
Supported protocols
Decode and schedule
People, faces, objects
Rules set per site
Sent to named staff
Trinocular AI adds intelligence on top of compatible existing CCTV.
Input Layer
Trinocular AI Engine
Compatibility depends on stream type and quality, protocol support, placement and lighting. Not every camera suits every model.
Output
A camera tells you what happened. Trinocular AI helps tell you what is happening.
Traditional CCTV
Trinocular AI
Trinocular AI routes each alert. Recipients and thresholds are set per organisation.
Routed to
In most cases no. Trinocular AI adds an understanding layer on top of compatible existing CCTV over ONVIF and RTSP. Whether a specific camera can support a specific detection depends on its angle, resolution and lighting, which we assess before committing.
Edge, on-premise, cloud or hybrid. Sites with footage that cannot leave the premises run inference locally; sites that want minimal on-site hardware run it in the cloud. The deployment model is a per site decision.
The connector discovers cameras on the subnet and indexing begins the same day. The work that takes real time is tuning thresholds and rules against your own footage, which is what stops the system producing alerts your team cannot action.
Edge deployments continue detecting locally and backfill when the link returns. Alerts queue and fire on reconnect with their original timestamps intact.
Bring your existing CCTV. We will assess what your angles can actually support and show you the detections running on your own footage.
