The Edge AI platform, end to end

DataFactZ Edge AI is built in three layers that map directly to where the work happens: detection on the device, reasoning in the cloud, and escalation through the channels your team already lives in.

1 — Control dashboard, written by AI

Every morning Claude reads the overnight telemetry and writes the headline: what happened, where, and what needs you first. Below it, the KPIs that matter — open violations, response time, camera-hours observed. The page is persona-aware: a safety lead, an operations director, and an executive each see the surface that matters to them.

2 — Live monitoring

Every connected camera, streaming in one grid. Detections render as bounding boxes the moment the edge device sees them — no round-trip to the cloud, no lag. The platform speaks ONVIF, RTSP, and generic IP cameras out of the box.

3 — Alert triage

Each alert opens to the frame that triggered it — bounding boxes, confidence, model, site, and zone. Acknowledge or resolve in one click. Claude flags when a burst is one detection loop, not twenty separate problems, so the team isn’t buried in duplicates.

4 — Plain-English rule engine

Describe the situation you care about in a sentence. Claude compiles it into a precise, structured detection rule — condition, severity, scope, and notification routing — ready to fire. No query language to learn. Every rule is reviewable before it goes live.

5 — Effectiveness telemetry

Every rule carries its own track record — how often it fires, how often those fires get resolved. Claude reviews the last 24 hours and tells you when a rule is noisy, quiet, or worth tightening. Engine status is honest: fires, partial, not yet.

6 — Sites & floor plans

Upload a floor plan and drop each camera where it lives, with its field-of-view cone. The same map your install team built becomes the operator’s live device inventory — online, alerting, or offline at a glance.

7 — Connectors

An honest catalog of every way to get cameras into the platform: our edge stack, ONVIF (Axis, Hikvision, Dahua), generic RTSP, and Azure IoT Edge (coming soon). When something is still in development we say “coming soon,” not “available.”

8 — Azure-native architecture

The cloud half runs on Azure: Container Apps for the API, PostgreSQL Flex with row-level security for tenant isolation, Redis for SSE pub/sub, Azure Static Web Apps for the dashboard, and Azure Entra for identity. The edge half is silicon-agnostic.

Book a working session at datafactz.ai/edge-ai/contact.