How it works
From guessing to live data.
Every install follows the same five steps, whether it is one store or the first ten of a network rollout.
Live in about two weeks.
- Site survey
Entrances and street coverage mapped
- Install, after hours
Our team, our cameras. Stores never close for us
- Calibration
Counting checked door by door
- Staff exclusion set up
Your team stops inflating your traffic
- Accuracy audit
We measure, you verify
- Live
Dashboards, app, weekly digest
Edge AI
The AI lives inside the camera.
Video is processed the moment it is captured, on the camera itself or on an AI unit inside your store. Footage never leaves the building. Only encrypted numbers reach your dashboard.
AI on the camera
Visitor counting, age and gender profiles and group detection run on a chip inside the camera itself. Powers every Standard plan store.
More AI, still in-store
Pro stores add recognition modules and higher accuracy on dedicated AI hardware inside your store. Nothing moves to a cloud.
Video never leaves
Only numbers reach the dashboard, encrypted, on EU servers. Footage stays in the building, under your control.
Accuracy
Numbers we stand behind.
We publish field-measured ranges from real stores, not lab numbers. After install, we audit accuracy against a manual count and calibrate until it is right.
| Metric | Accuracy |
|---|---|
| Visitor counting | up to 99% |
| Staff exclusion | ~90% |
| Gender detection | ~90% |
| Age estimation | ~85% |
Your agents, our numbers.
Every count, conversion and benchmark over a public API. MCP support, so your own AI agents ask Vitrac directly.
api.tassvision.ai/v1/docs →Return
