PLATFORM OVERVIEW
The intelligence layer for the security operations center.
Sentinel turns the cameras you already own into real-time decisions. An on-site agent does the heavy AI work where the video lives, your data stays in your network, and one console ties it all together — cloud, on-premise, or fully air-gapped. No proprietary hardware. No cloud GPU.
Click to enlargeAny ONVIF camera, managed uniformly — from one to thousands.
How it fits together
Four layers, one platform.
Most video systems are a recorder with plugins bolted on. Sentinel is built as a platform: the AI runs where the video is, the heavy lifting can move to a server pool when you need it, and the console is the same product whether it runs in your cloud or inside an air-gapped room.
Cameras
Any ONVIF or RTSP camera you already own. No proprietary recorder, no vendor lock-in.
On-site Agent
A single binary at the site that does the AI, the streaming, and camera control — all locally, on CPU.
Cloud or On-Premise
Events, faces, evidence, and audit indexed and stored where you choose, with an optional worker pool.
Operator Console
One web console for the whole job: respond, investigate, govern.
The agent does the work where the video lives.
The on-site agent is a single binary that runs as a service at the customer’s site — it connects to the cameras, runs the AI on-device in ONNX, and hosts the live video. Video and inference stay local; only events, evidence, and overlays travel onward. That’s how Sentinel keeps footage in your network and skips the cloud-GPU bill entirely.
- On-device AI in ONNX — faces, plates, people, vehicles, behavior, threats, on CPU
- Live video locally — RTSP → embedded mediamtx → low-latency HLS
- ONVIF discovery & PTZ — finds cameras automatically, controls them over ONVIF
- Runs as a service everywhere — Windows, Linux, macOS — with auto-update + rollback
Click to enlargePoint the agent at the network — ONVIF discovery finds the cameras.
How it works
From cameras to decisions, then to evidence.
Connect
Install the agent and point Sentinel at your cameras. ONVIF auto-discovery finds them, or enter RTSP.
Understand at the edge
The agent runs on-device AI in ONNX — faces, plates, people, vehicles, behavior, threats — and converts RTSP to HLS, locally on CPU.
Index where you choose
Events, faces, and clips are indexed in your cloud or on-premise; a worker pool handles re-processing and backfills.
Act in the console
Alerts fire, the threat picture updates, and events land on the map, timeline, and Command Wall at once.
Prove it later
Investigate across cameras, export sealed evidence, and account for every action in a tamper-evident audit.
Go deeper
The platform, in detail.
AI Vision Pipeline
The models behind the detections: face, plate, label, person/vehicle, and behavior — on the edge and in the worker.
Learn moreBuilt for Scale
How Sentinel stays sub-50ms at millions of events: covering indexes, pgvector kNN, and soft-fail enrichment.
Learn moreDeployment & On-Prem
Cloud-managed, on-premise Docker, and fully air-gapped, plus the setup wizard and white-label.
Learn morePartners & White-Label
Deliver Sentinel under your own brand and extend it with new connectors and locales — without touching the core.
Learn moreSee the platform on live data.
Request demo access and we’ll send you a private, pre-loaded environment — discover cameras over ONVIF, watch the live AI overlay on a multi-camera wall, and follow a detection from the edge to a sealed evidence pack.