OddessyDOCUMENTATION
GET STARTED Live

From invitation to first heartbeat

Sign in, enrol an edge node and verify its telemetry in the Oddessy Platform.

Last updated 27 July 2026 ยท Documentation version 2026.07

Before you begin

You need an Oddessy Platform invitation and a supported Windows or edge-capable Linux node. Linux targets include Jetson, QCS6490, Raspberry Pi 5, generic ARM64/x86_64 and Yocto-capable product images. The enrolment token is shown once and expires after 30 minutes.

Oddessy device enrolment lifecycle from tenant operator through a one-time token to outbound authenticated device heartbeats
One-time bootstrap establishes a durable device identity without opening an inbound management port.
Request an invitation

Open platform.oddessy.io and choose Request a platform invite. The structured request captures company size, intended deployment, systems and region, then appears in the reviewed Oddessy admin inbox.

1. Install the runtime

Windows

Download the approved x64 MSI from Releases. It installs the machine-wide OddessyEdge service and stores configuration under C:\ProgramData\Oddessy\Edge.

Linux / Yocto-capable edge runtime

Create the device in Platform, then use the reviewed provisioning path supplied for your pilot. The npm command below is the intended registry interface; @oddessy-io/cli remains a release candidate pending publication. It requests a 15-minute signed release, verifies SHA-256 before installation, enrols the host and starts the service. Leave hardware_profile set to auto, then confirm the detected target and model-scope status in telemetry.

sudo npx --yes @oddessy-io/cli provision   --platform https://platform.oddessy.io   --token "$ODDESSY_ENROLMENT_TOKEN"   --install

2. Create the enrolment

  1. Open Devices and choose Add device.
  2. Enter a descriptive name, physical location and hardware type.
  3. Copy the one-time command and run it with the required administrator privileges.
  4. Confirm the device reports Operational and inspect its model-provenance state.

3. Verify locally

# Ubuntu
sudo oddessy-edge status
sudo oddessy-edge telemetry

# Windows PowerShell
Get-Service OddessyEdge
& "$env:ProgramFiles\Oddessy\Edge\oddessy-edge.exe" status
& "$env:ProgramFiles\Oddessy\Edge\oddessy-edge.exe" telemetry

Compact Linux / Yocto vision

Where a VLM is unnecessarily large, install OpenCV in the target image and use oddessy-classical-vision for local Canny edges, contours, baseline motion or a compact ONNX classifier through OpenCV DNN. Results remain advisory and behind the same PLC/safety boundary.