PeopleSense AI
Talk to us
On-device vision, not just cloud uploads

Cameras that understand
your people, not just record them.

PeopleSense AI turns ordinary CCTV into a live intelligence layer — presence, attendance, dwell, and behaviour, computed at the edge in real time and surfaced in a console your team actually opens.

<150msface-to-identity, on-device
24/7continuous inference, no cloud round-trip
2product lines on one platform
OTAfleet-wide model & firmware updates
One platform, two products

Built for the workforce floor and the storefront.

The same on-device vision engine, tuned into two purpose-built products — because a factory floor and a sweet shop counter ask completely different questions of a camera.

WorkSight — Shift

Workforce intelligence

Attendance, presence and zone activity — computed from the cameras you already have, with no badge, no app, no manual sign-in.

  • Real-time who's-on-site, per zone and department
  • Arrival, departure and dwell — reconciled automatically
  • Unauthorised-absence and after-hours alerting
  • Live camera wall with capped, audited playback
WorkSight — Retail

Storefront analytics

Understand footfall, dwell and service moments across a store or a chain, without touching a POS integration.

  • Footfall, repeat-visitor and peak-hour patterns
  • Dwell-by-area — counters, aisles, seating
  • Queue and counter-wait signals
  • Security and after-hours event review
Architecture

Inference happens at the camera. Decisions happen in the console.

Nothing waits on a round-trip to the cloud to know who just walked in. The edge device does the seeing; the console does the deciding.

01

Edge device

An on-site AI unit runs detection and recognition directly on its NPU against your existing RTSP cameras — no re-cabling, no cloud dependency for the core loop.

camera → edge NPU → local decision
02

Sync & retain

Only the compact, meaningful signal — presence, identity, zone, timestamp — is synced to the cloud. Raw video stays local; nothing streams continuously off-site by default.

edge → compact event → cloud
03

Console & alerts

A live, role-based console turns that stream into dashboards, alerts and reports — reachable from a browser, tuned for the person who has to act on it.

cloud → dashboard → decision
Under the hood

The parts that make it hold up in a real building.

On-device face recognition

Detection and recognition run on the edge NPU against a live camera wall, with pose and quality checks before a match is ever recorded.

Zone & dwell analytics

Every space is mapped into zones so presence and time-spent are attributed to an area, not just a timestamp.

Fleet-wide OTA updates

Camera configuration, detection models and agent firmware update over the air across every site, with staged rollout and rollback.

Live camera console

A capped-playback, audited live view — HLS delivery through a global edge network, not an open, unmonitored stream.

One console, many sites

Tenant, client, site and camera are first-class — a new location is configuration, not a new deployment.

Privacy by construction

Face data stays scoped to the site it was captured at; retention and access are configurable per customer, not shared across tenants.

Bring your own cameras. We'll handle the seeing.

Most sites are live within a couple of weeks of the first walkthrough — no re-cabling, no ripping out your existing NVR.

Get in touch

Tell us about your site.

Number of cameras, the building type, and what you're hoping to see — that's enough for us to tell you if it's a fit.

We reply from a real person, not a ticket queue. If you'd rather write directly:

hello@peoplesenseai.com

Staging build note: this form does not submit anywhere yet — it's marked disabled on purpose until the real intake endpoint is approved. Use the email link above for now.