Disinformation tracking
Coordinated narrative detection across video platforms — bot-network cadence, mirrored clips, manipulated audio. Routed to the analyst desk with provenance and reach signals attached.
Use case · Public institutions
Disinformation tracking, public-safety signals, and compliance evidence — analyzed from public-media and CCTV streams with audit trails your records team already accepts. Privacy-preserving by default; face recognition off until a documented legal basis turns it on.
What the engine does for public-sector teams
Behaviour, density, and content classification — without identity. Every event ships with a self-contained case file: the clip, the contributing signals, the policy IDs invoked, the operator action. Exportable in formats your records team already accepts.
Coordinated narrative detection across video platforms — bot-network cadence, mirrored clips, manipulated audio. Routed to the analyst desk with provenance and reach signals attached.
Crowd density, abandoned objects, traffic incidents on existing city and transit cameras. Behaviour-only models — face recognition stays off unless authorised under a documented legal basis with its own audit trail.
Every event logs the model version, the contributing signals, the policy IDs, the timecode, the operator action. Exportable for FOIA, regulatory inquiry, or court-admissible chain-of-custody.
How it lands in your environment
Deployment options designed for jurisdictions that take data residency and procurement controls seriously — not as an afterthought.
EU, US, or in-country managed cloud; or self-hosted in your data centre / VPC for sensitive workloads. Customer choice, documented at procurement.
Available on common cooperative-purchasing schedules (Sourcewell, state MSAs on request). Pilot pricing fits inside discretionary spend for most agencies.
Every major VMS over RTSP / ONVIF; structured events out to the CAD or ERP system you already run, via webhook or message bus.
Our promise to public-sector teams
We are building a map-first operator console with behaviour, density, and abandoned-object models trained on real urban and platform footage. Face recognition stays off unless a documented legal basis switches it on, with a separate audit trail. We are pre-launch in smart-city / public-sector; these commitments are how we will structure pilots, not metrics from a fielded deployment.
No FR by default
Behaviour models — identity not required for the job
Edge-blur ready
Identifiable detail blurred before frames leave the camera
Audit-grade
Every event ships with a self-contained case file
Residency choice
EU, US, in-country, or self-hosted in your data centre
From the questions that come up in council review, the procurement office, and FOIA-readiness conversations.
Off by default. The behaviour and density models do not need identity to do their job. If your jurisdiction has authorised it under a documented legal basis, a feature flag turns it on with a separate audit trail and operator role-gating.
Customer choice — on-prem in your data centre, in-country managed cloud, or hybrid. Operators see only the streams their role allows; every view is logged and exportable for records review.
Every event ships with a self-contained case file: the clip, the timecode, the contributing signals, the model version, the operator action. Exportable in formats your records team already accepts.
Available on common cooperative-purchasing schedules. Sourcewell and a state-specific MSA list available on request. Pilot pricing fits inside a discretionary spend authority for most agencies.
Yes — every major VMS over RTSP/ONVIF, structured events out to your CAD over webhook or message bus. We do not replace the systems you already pay for.
Public institutions
Tell us the corridor, station, or platform you want to start with. We will scope a pilot inside three weeks, with the privacy-impact assessment in writing.