Technology

Technology

More AI compute than a typical camera, less exposure than cloud inference, and camera images that stay on a private local network. This page describes the technical design rather than a shipped spec sheet.

Home pilot underway since 5 October 2026. Camera counts, UPS runtime, and model names will be published only after they are measured.

Local AI computer illustration

The three common architectures

ArchitectureAdvantageTrade-off
In-camera AIPrivate, low latencyVery limited compute budget
Cloud AILarge models, easy scalingImagery or derived data leaves the premises; vendor and internet dependency
ilonai local AI applianceMuch more compute than a typical smart camera, imagery stays localRequires a local compute box and a dedicated network

That middle ground is the technical story. For a general audience we call it context-aware local AI. For technical readers: a fine-tuned multimodal vision-language model running on dedicated local hardware. Model brand names do not belong on a customer homepage.

System diagram

System diagram: cameras connect to a managed switch, then to a local AI computer. The computer handles reasoning, a short buffer, fall decision, and health. Independent 5G carries alerts only.
Cameras on a closed network of ilonai's own (RTSP / ONVIF), wired wherever possible; wireless on that same network where a cable can't be run. The local AI computer handles context-aware visual reasoning, a short encrypted rolling buffer, fall decision, camera and node health, and local encrypted incident storage. Independent 5G/eSIM carries heartbeat and emergency alerts only. Optional UPS sits on the compute side.

Your home or facility Wi-Fi does not need to become a surveillance network. Cameras are not meant to sit on ordinary household Wi-Fi or to require a camera-vendor cloud account.

Why centralized local compute?

More AI capability than a typical camera, less privacy exposure than cloud inference.

Putting a tiny detector in every camera is convenient and private, but the model has to fit a small processor. Sending video to a large cloud model is capable and operationally easy, and it moves sensitive imagery off site.

One local AI node, or a few of them, can serve many inexpensive cameras. The intelligence lives on the box, so the camera stays a replaceable sensor. We will publish home-node and facility-node camera ranges only after decoder, network, and model benchmarks exist. Nobody should promise “50 cameras easily” before measuring it.

For facilities, spreading the work across several compute nodes keeps one hardware fault from taking out an entire building.

Cameras: standards-based, approved list

We are not building a proprietary camera cloud.

Intended properties:

  • approved network cameras (wired/PoE where possible; wireless on the same closed network where needed)
  • local IP networking
  • standard local video protocols (RTSP / ONVIF class)
  • cameras replaceable without rebuilding the AI system
  • no camera-vendor cloud required for core detection

That keeps per-room hardware cost down at scale, makes replacement simple, avoids lock-in, and uses cabling facilities already know. The camera stays “dumb” while ilonai makes the decision.

We qualify camera models for power-cycle recovery, RTSP reconnects, night mode, image quality, and stability. We do not promise arbitrary third-party camera compatibility.

None of this makes it a “smart camera.” That phrase usually means a vendor app, a cloud account, and remote viewing, which is the opposite of what we are building.

Isolated network: cameras to a private switch, then the local AI computer, then independent 5G for heartbeat and alerts only.
Cameras stay on ilonai's closed network (wired where possible). The local AI computer never needs household Wi-Fi. Independent 5G/eSIM is for heartbeat and emergency alerts, not a camera cloud.

Why this matters:

  • the install does not depend on customer IT
  • networking behaves predictably
  • the attack surface is smaller than cameras on everyday Wi-Fi
  • no ordinary household or office device can reach a camera
  • alerts go out over an independent cellular link
  • installation needs no port forwarding and no home Wi-Fi credentials for the cameras

Detection keeps running locally even when the external connection is unavailable.

Rolling buffer and incident storage

Video is not retained as a diary of the day. A short encrypted rolling buffer (about 10–15 seconds) exists locally and is continuously overwritten. If a possible fall is detected, that window can be frozen, plus a short post-event period, as a local incident clip.

Access is authorized and logged, retention is short, and ordinary footage is never uploaded to a video cloud.

Heartbeat and failure detection

A safety system has two jobs: spot the emergency, and notice when it can no longer spot anything at all.

What the system watches:

  • whether each camera is still delivering valid frames
  • the health and heartbeat of the AI computer
  • missed heartbeats, which usually mean a lost cellular path
  • mains power lost for longer than a defined period, five minutes for example, which triggers a notification
  • recovery, reported when power returns

We do not promise a UPS runtime until the hardware package is fixed.

What ilonai can see remotely: a camera is offline, a heartbeat was missed, which software version is installed.
What ilonai cannot see remotely: what the person is doing in the room.

See also How it works, Privacy & Trust, and Evidence.