A computer-vision layer that snaps frames from any camera stream, processes them asynchronously, and runs four ready-to-go analysis modules — counting people, recognising staff, tracking inventory on shelves, and reading the room's emotional temperature.
Real-time headcounts, dwell time, queue length, and hourly/daily trends. Anonymised — no identities stored.
Face matching against your team roster. Attendance, presence at posts, late arrivals, shift handovers.
Watches your shelves and bins. Detects low stock, mis-shelving, and replenishment events — by object class, not SKU.
Aggregated, anonymous facial-expression signal across the visit. Trends by hour, by zone, by shift.
A continuous, anonymous headcount per zone. The counter handles overlapping detections, re-entry, and queue dynamics — and feeds charts you can act on without a data team.
Define zones in the camera frame (entrance, queue, fitting room). The counter tracks live occupancy and queue length, exposing both as API and live chart.
How long people stay, and which way they move. Heat-map ready out of the box.
The counter never persists identities. Faces are reduced to detection tokens that expire when the visitor leaves the frame.
Generated charts and digests — emailed, posted to Slack, or exposed via the AiGAP analytics package.
Match faces against a roster you maintain. Get attendance, presence at posts, late arrivals, and shift handovers — with the strict guarantee that nobody outside your roster is identified.
The model only matches enrolled team members. Anyone outside the roster stays anonymous — the system never builds a face database of strangers.
First seen / last seen, time at assigned post, lunch break detection, late arrivals. Daily and weekly logs.
Detects when the outgoing shift hasn't been replaced by the incoming one — alerts before the gap costs you a sale.
Face embeddings are computed on-device. The raw images leave the camera as anonymous tokens; only matches against your roster surface to dashboards.
Object-class inventory tracking. The camera watches what's where, how full it is, and what's missing — without barcode scans, weight sensors, or SKU mapping.
Trained on your product categories (or trained live by you, on-screen, by drawing boxes). The shelf shows up as a tiled grid with per-cell fullness.
Configurable thresholds. When a cell drops below the line, an alert goes to the right phone, the back room screen, or the right Slack channel.
If a class shows up where it shouldn't, the camera flags it. Cuts the daily walk-the-floor audit to a glance.
Every restock event is timestamped. See how long shelves stay full, who restocks what, and when patterns drift.
Aggregate, anonymous facial-expression signal — not as a per-person judgement, but as a zone-level mood index over time. Pair with hours and staffing to see what actually moves the dial.
The chart is a rolling zone-level mean. The system can't show you "person X looked happy" — that's a deliberate boundary, not a missing feature.
Slice the mood signal by hour-of-day, by camera zone, or by the shift on duty. Spot where service tilts the room one way or the other.
Cross-reference with your POS, your staffing roster, your queue data. The emotion signal isn't a verdict — it's a clue.
Expressions are encoded into anonymous mood vectors on-device. Raw frames are discarded after the snapshot pipeline finishes. No face library, no recording.
Each camera feeds an asynchronous snapshot processor; the snapshots are dispatched to whichever modules you've enabled. Results stream back to dashboards and the AiGAP core in seconds — without ever blocking the camera.
RTSP, MJPEG, ONVIF, IP cams, USB webcams, or phone cameras. AiGAP adapts to the source.
The processor pulls frames at a rate the analysis can keep up with — no dropped frames, no blocked camera.
The same snapshot is handed to every enabled module in parallel — count, staff, inventory, mood.
Faces and objects become anonymous vectors at the edge. Raw frames are discarded after the snapshot expires.
Live dashboards, hourly digests, low-stock alerts, late-arrival pings — wherever your team works.
A few sectors where camera-driven AiGAP is already paying its way.
Live queue length on the till tablet. Coffee bean shelf alerts. Mood by hour cross-referenced with the espresso machine's repair log.
Lobby presence tracking for staffing. Restaurant mood as a service-quality signal. Late-arrival nudges for night-shift.
Waiting-room queue analytics, no-show detection. Anonymised mood signal for waiting experience — never per patient.
Workbench inventory tracking. Staff presence at safety-critical posts. Mood as a fatigue signal at end-of-shift.
We onboard pilot customers in a single day — bring your camera stream and your roster, we'll ship the dashboard and the alerts. Pricing is per-camera, paid monthly.