Atriva Atriva

Deployment

Every Atriva solution runs on the same stack. If a site has cameras and power, it's deployable.

STAYS ON SITE — NO VIDEO LEAVES THE NETWORK IP Camera RTSP / ONVIF + more cameras Atriva Edge Device on-prem · runs inference Your systems JSON events · webhook / API Atriva dashboard live alerts · reports · history
Runs on
NVIDIA Jetson, Intel, Rockchip, Hailo, MemryX
Cameras
Any RTSP/ONVIF IP camera — existing cameras are fine
Network
Local processing. No cloud dependency, no video leaves the site
On-site footprint
One edge device per 8–16 cameras
Output
Structured JSON events to your systems, or our dashboard
Time to stand up
1–3 days per site once hardware is on-site

Integration outputs: MQTT · Webhook · REST API

Deployed as containers. The same image runs across every supported accelerator — the runtime is selected at install, so switching hardware doesn't mean rebuilding the application.

Common questions

Can it use our existing cameras and VMS?
Yes. Atriva works with any IP camera that streams over RTSP or ONVIF, which covers the majority of cameras installed in the last decade. We don't replace your VMS — we sit alongside it, reading the same stream your VMS already records. You keep your existing footage, access control, and recording workflows unchanged.
What happens if the network drops?
The edge device keeps processing. Detection, alerting, and local logging all continue without a cloud connection — there's nothing cloud-side that the on-device runtime depends on. Events queue locally and sync when connectivity is restored. The only thing that goes offline is remote access to the dashboard.
Who installs it — you or us?
Either, depending on what makes sense for your site. For first deployments we typically install ourselves so we can calibrate detection zones and validate performance on-site before handover. For subsequent sites we can train your team or a local integrator to handle it. The containers are designed to be reproducible — the same config that worked on site one works on site ten.
Can we run it on hardware we already own?
Possibly. If you have an x86 server or an NVIDIA Jetson on-site, we can often deploy to it. We'll tell you upfront if the hardware won't meet the throughput requirements for your camera count rather than letting you find out after install. We'd rather right-size the hardware at the start than debug performance issues later.
What does ongoing support cost?
Support is scoped per deployment and quoted before you sign — not hidden in a per-camera annual fee after go-live. It covers threshold tuning, model updates when cameras change position or lighting conditions shift seasonally, and a response SLA for incidents. Most sites run on a flat monthly fee.
Talk to us about your site