EdgeTrace

The core of the product is a search box that takes ordinary English. You describe the object, the action, or the moment, and EdgeTrace returns matches ranked by score with timestamps attached. Queries run across every connected camera at once, so a search doesn't stop at the edge of one parking lot.

EdgeTrace joins Genetec, Milestone, Avigilon, and S3 storage rather than asking you to rip out a working system. It indexes in place and can phase out the old VMS later if you want. That keeps the migration cost down and lets a small agency pilot the search layer without replacing hardware.

Beyond object search, EdgeTrace flags actions: someone reaching into a baggage carousel, loitering past a threshold, or crossing a line you drew on the building model. Alerts arrive as a location on that model, not as a camera number, which shortens the gap between a flag and a response.

Once a subject is identified, re-identification follows that person or vehicle across the fleet and stitches the appearances into one thread. Investigators get a route instead of a folder of unrelated clips, and the same logic supports live interception while a call is still open.

Chain of custody video export is built into the same screen as the search. Every query, viewer, export, and edit is signed and timestamped. Exports carry hash-signed metadata and query lineage, plus a printable affidavit describing how the evidence was surfaced. For a prosecutor, that record is the difference between a clip and an exhibit.

The appliance sits inside the agency network, and footage plus indexes stay there. A fusion or high-security site can run fully isolated, with model updates delivered on signed media. EdgeTrace states contractually that it doesn't train on, retain, or repurpose agency video.

Faces, plates, and bystanders are blurred on export according to policy, and reversing a redaction requires logged supervisor approval. That cuts the manual editing step that usually delays a public records release or a defense disclosure.