Technology
Built like the trucks it watches
Industrial hardware on the vehicle, deep-learning models behind it, and an architecture that keeps your fleet's raw data in your hands. Here's what's under the hood, for the people who will ask.
On the truck
A sealed, industrial-grade edge computer rides on each vehicle, rated for minus 40 to 85 degrees Celsius, reading 1,000-plus J1939 CAN bus signals per second across 200-plus truck types: powertrain, braking, thermal, and emissions systems. Anomaly detection runs on the device itself in milliseconds, so a full offline mode keeps watch on low-connectivity routes and forwards the data when coverage returns. Safety-critical alerts never wait on the cloud.
The models
Deep-learning models trained on billions of data points from millions of vehicle-hours learn what the run-up to a failure looks like, then watch for it 48 to 72 hours before any fault code triggers, at a projected 85 to 90 percent accuracy with a sub-5 percent false-positive rate. The models weigh more than engine data: weather, traffic, route profile, driver behavior, and each truck's maintenance history all feed the same prediction. Every confirmed fault feeds back in, so accuracy compounds with every mile the fleet runs.
Your data stays yours
The platform uses federated learning: model improvements travel, raw telemetry doesn't. Your fleet's operating data stays in your domain, encrypted in transit and at rest with AES-256, on infrastructure designed to SOC 2 Type II and ISO 27001 specifications and built for 99.95 percent uptime. Anything shared beyond your fleet is anonymized, aggregated, and governed by your agreement, never sold out from under you.
It fits the stack you have
The Canary runs alongside your existing ELD and telematics, not instead of them, and connects at the diagnostic port without rewiring. Rollouts are progressive: start with the trucks and use cases that create the most value, prove it, then expand. Alerts route wherever your operation already lives, email, SMS, and the dashboard, and the platform is designed to connect with the ERP and TMS systems larger fleets run on.
Bring your hardest questions
The demo is run by people who can answer the technical ones.