From camera feed to evidence, in four steps.

Step 1: Connect the cameras you have

Eye AI works with standard network cameras and recorders. If your cameras provide a video stream to your network, we can read it. No replacement, no re-cabling in most cases. We check each camera view in a free camera-readiness check first: angle, light, distance and what is realistic.

Step 2: Mark what matters

On the camera view, you or our team draw lines (a gate, a door, a dock edge) and zones (restricted areas, parking bays, exits, aisles). You decide what counts as an event and how sensitive each one is.

Step 3: An on-site computer does the watching

A computer at your site analyses the video as it arrives. Your video does not leave your network to be analysed. We size the computer to your number of cameras and chosen analytics (see Deployment).

Step 4: You get proof, not noise

• A live dashboard on your network (password protected)
• Events with time, zone, object and a snapshot
• Counts and dwell times by hour, day and gate
• Reports you can export (CSV) for managers, auditors and insurers
• Alerts to your team's phones (WhatsApp/email) Launching

Built so nothing slips through

• Nothing-is-silent safety net: if something moves and the AI could not name it, you are shown the moment and can label it. Every label makes the system better for your site.
• "Were we blind?" stream health: a record of every minute each camera was healthy, with the gaps listed.
• Self-healing: if a camera feed drops, Eye AI reconnects and records the gap.

What a project looks like (our real process)

1. You send a request. One form: your site, cameras and what you want to see or solve. We reply within 24–48 hours (working days, Monday to Friday).
2. Conversation and site survey. We talk it through, then review your existing cameras, recorder and the areas to watch, on site where possible.
3. Solution proposal. A tailored solution mapped to your site and your answers.
4. Your approval. You sign off on the scope before anything is built.
5. Demo on your own cameras. We build a working demo on your footage so you see it before committing.
6. Final commercials. A clear one-time and recurring quote, scoped to what you saw.
7. Build, deployment and accuracy ramp-up. We deploy on site, then spend 1–3 months tuning against your real conditions to reach the best accuracy.