Frequently asked questions

Axiom Guard is sold as a one-time Pilot for evaluation, annual Starter and Enterprise subscriptions for production rollouts, and a quote-driven Multi-Site tier for large operators with 50+ facilities. Per-camera and per-site caps are listed on /pricing; Multi-Site carries a volume-discount curve past 50 sites. There is no usage-based metering on inference, no per-event charges, and no hidden seat fees — your tier is priced up-front against your camera and site count.

Raw video is never stored by Axiom Guard — frames are analyzed in memory and discarded as soon as inference completes. Detection events (anonymized metadata: timestamp, location, classification, confidence score, and a thumbnail) are retained per your tenant policy, defaulting to 90 days and configurable up to 7 years. Operator audit logs are retained for 13 months. Full retention windows, right-to-erasure paths, and processor terms are documented in the DPA.

Every camera runs a per-tenant confidence threshold plus a rule engine that filters on schedule, zone, class, and an explicit allow/deny list. Every surfaced event carries its confidence score and a thumbnail an operator can dismiss in one click; dismissals feed straight back into per-camera tuning so noisy sites quiet down within the first two weeks. Reported false-positive rate on production tenants is under 5% after week-two tuning, and we publish the methodology in the deployment report.

Most pilots are ingesting live video within 48 hours of credential hand-off. Axiom Guard supports ONVIF and RTSP out of the box, plus native connectors for Genetec, Milestone, Avigilon, Axis, and Hanwha — so there is no rip-and-replace of cameras or recording hardware. A solutions engineer joins your first sync, validates the connector end-to-end, and stays on call through the first detected event.

Production tenants carry a 99.9% monthly uptime SLA with service credits, a 15-minute P1 response window 24/7, a named Technical Account Manager, and a dedicated solutions engineer on call through the first quarter. The public status page, the historical uptime report, and our incident-response runbook are all published, and you can request the runbook under NDA at any time.