AI fleet health & predictive maintenance
XFleet reads the live condition of every asset and turns it into one honest health score — engine, electrical, brake and fluid — with a breakdown-risk estimate and days-to-failure. It runs on whatever telematics hardware you already have, fused onto a single record per truck.
The problem
A roadside breakdown costs far more than the part — towing, a stranded load, a missed delivery window, a driver sitting idle. Yet most fleets still run to failure or to a fixed calendar, because the one thing they can't see is the true condition of each unit, right now.
Calendar-based PM over-services healthy trucks and misses the one quietly heading for a failure.
Geotab on some units, another vendor on others — and no single view of condition across the whole fleet.
Without condition fused to cost, the unit that's quietly bleeding margin looks the same as the rest.
Hundreds of raw diagnostic codes, no sense of which actually threaten an unplanned stop this week.
How it works
XFleet pulls signals from whatever device is on the truck, unifies them onto one asset record, and scores condition the same way for every unit — so a number on the Norfolk refuse fleet means the same thing as a number on the Palmetto drayage fleet.
Diagnostic and GPS signals are pulled from Geotab, Geometris and other devices on a continuous cadence.
Every signal maps onto one canonical record per asset — vendor-agnostic, so the whole fleet is comparable.
Engine, electrical, brake and fluid subsystems each get a score, rolled into one overall health number.
Current condition becomes a breakdown-probability estimate and a days-to-failure projection per unit.
An AI narrative explains why a score moved and lists the recommended actions, in plain language.
Risk tiers sort the fleet so your shop works the trucks that actually threaten an unplanned stop first.
Health scoring
XFleet doesn't hand you a green/yellow/red light and leave you guessing. Each asset carries a current overall score built from four subsystem scores, a breakdown-risk estimate, a days-to-failure projection, and a written explanation of what's driving it — recomputed every day from the live signals.
Coolant, RPM, fuel and fault patterns roll into an engine subsystem score that flags trouble before it becomes a tow.
Battery and charging-system voltage tracked against healthy ranges, so a dying battery surfaces before a no-start.
Brake-system signals scored as their own subsystem, surfaced wherever the hardware reports them.
Coolant level, DPF soot and aftertreatment signals feed a fluid score that catches regen and overheat risk early.
Predictive, not just reactive
A score on its own is a snapshot. XFleet turns it into a decision: how likely is this unit to fail, and roughly when — so you schedule the fix on your terms, in your shop, instead of on the shoulder of an interstate.
Hardware-agnostic by design
Most platforms are really a front-end for the hardware they sell, so a mixed-vendor fleet never unifies. XFleet works the other way around: it ingests from the devices you already run, normalizes them onto one canonical record per asset, and scores every truck on the same scale. One fleet, one language for condition.
Geotab GO9 / GO10 and Geometris today, with one record per asset regardless of which device reports it.
A single digital twin per truck that every Prometheus module reads from — the same record XFuel and XShield use.
An 82 on a refuse packer means the same as an 82 on a sleeper — vocation-aware scoring on one consistent scale.
Ask questions about the fleet and get a real, score-backed answer — the fleet health gauge reads true condition, not a guess.
Raw diagnostic codes are classified and ranked, so the noise drops away and real threats stand out.
Condition flows into the wider platform — feeding cost-per-mile, dispatch and accident reconstruction from one source of truth.
Request a demo
Connect the hardware you already run and watch XFleet score every unit's condition, surface the trucks heading for trouble, and tell you exactly what to fix first.