Hundreds of routers and gateways across sites, vehicles and cabinets. Multi-WAN links that fail over at 3am. Bridged Matter, Thread and Zigbee estates nobody can see. This is the stack that turns all of it into one operated fleet.
Edge estates accumulate hardware generations and vendors. Each brings its own portal, and none of them shows the whole fleet.
Without remote diagnostics, Wi-Fi surveys and guided field workflows, every anomaly becomes a site visit.
Fleet-wide firmware rollouts without canaries, waves and abort conditions are how one bad image becomes an outage across every site at once.
Matter, Thread and Zigbee devices behind the gateway carry real service value and usually zero management visibility.
A modern C++ client built for OpenWrt-class routers and gateways, in production on live reference hardware.
The same carrier-grade core that runs tens of millions of devices: every gateway interaction flows through its northbound APIs.
The fleet operations console: one product that resolves into purpose-built views per persona. Currently a design preview you can click through, open to design partners.
Designed against live reference hardware, not lorem ipsum: the demo fleet includes a real OpenWrt router running the Friendly UltraPRO client.
Link health, failover events and Starlink terminal telemetry per site, so the fleet view explains the outage before the customer reports it.
Firmware (Object 5) and software (Object 9) campaigns with canary staging, wave scheduling and regression aborts. Fleet-wide change without fleet-wide risk.
Firewall and VPN state, patch level and configuration drift across every site, visible to the security officer view as a posture, not a spreadsheet.
Profile state and lifecycle operations for cellular-backed gateways, managed alongside the device rather than in a separate operator portal.
Matter, Thread and Zigbee devices surfaced through the gateway's LwM2M Gateway object and managed as first-class fleet citizens.
Model deployment and inference monitoring for TFLite Micro and ONNX Runtime workloads running on the edge devices themselves.
Fleet Ops, Analytics, Auto-Fix and Security Sentinel agents acting under tiered auto-approval: humans approve by risk class, every action audited.
The site visit that does happen gets a guided mobile workflow with the device's live state, not a printout and a hunch.
Backhaul on 5G/LTE or fibre; LoRaWAN, Thread/Matter/Zigbee and Wi-Fi to the last device; one management plane behind all of it. The gateway is where those lanes converge.
| Lane | Range | Topology | Carries |
|---|---|---|---|
| LoRaWAN | 1 to 5 km | Star of stars, multi-year battery | Remote telemetry, outbuildings, sub-metering |
| Thread / Matter / Zigbee 3.0 | 10 to 100 m | Self-healing mesh | Sensors, locks, lighting; bridged via LwM2M Object 25 |
| Wi-Fi 6 (DBDC) | 100 to 500 m | Star, high throughput | Cameras and heavy data |
| Backhaul WAN | Site uplink | 5G/LTE cellular standby (2× M.2 WWAN) + gigabit Ethernet or fibre | Everything northbound, with failover |
The point that matters: the lane is a deployment choice; the LwM2M session behaviour and the management evidence are the same on every one of them.
The operator relationship should be a fleet decision, not a soldering decision. SGP.32 makes the SIM a managed object, and the gateway already carries the machinery.
Models on the edge are fleet assets: deployed, versioned, monitored and rolled back with the same discipline as any other software on the gateway.
A backend abstraction over TensorFlow Lite and ONNX Runtime, on CPU everywhere and on GPU or NPU where the silicon SDK provides one.
A dedicated Edge-AI object with 36+ resources exposes model identity, version, health and inference statistics to the management plane.
Models ship through the same Object 9 software campaigns as any package: canary sites first, waves after, abort conditions armed.
Edge Command surfaces model deployment state and inference monitoring per site, next to the WAN, Wi-Fi and security posture it already shows.
Local gateways keep the site operating; the fleet view reconciles when the link returns.
NTN and Starlink links managed with the same tooling as urban fibre sites.
HVAC, occupancy and energy behind a site gateway, bridged into one operational view.
Substation gateways with failover assurance and campaign-grade update discipline.
Edge Command is white-label capable by design. Operators and MSPs run managed edge-router fleets as a branded service: the NOC wall for operations, the mobile app for field techs, the executive view for the quarterly meeting.
Enterprises with edge estates get the same machinery pointed inward: retail, energy and transport sites with visibility, campaigns and failover assurance across hundreds of locations.
Honesty first: today Edge Command is a design preview you can click through, pinned to live reference hardware and a simulated fleet. That is exactly why now is the time to shape it as a design partner. UltraPRO and the One-IoT platform beneath it are in production.
A working session against the live demo fleet, plus the design-partner programme details.
Submissions route to Friendly's CRM. No newsletter, no drip by default.
