MicroFleet Pulse
Failure prediction for small mixed fleets without manufacturer telematics subscriptions.
- Category
- Fleet maintenance
- Model
- Hardware + SaaS
- Core platform
- ALIA Connectivity Core
- Status
- PLANNED
01 Problem
Small mixed fleets fix vehicles when they break because the telematics products that could warn them are priced and built for enterprises. OEM subscriptions bill per vehicle per month and rarely cover a real fleet’s mix of trucks, vans and equipment.
The cost is roadside failures, jobs missed while a vehicle sits, and repairs that cost multiples of the maintenance that would have prevented them.
02 Specification
This is the specification we are building toward. The status column shows what exists today.
- OBD and CAN signal extractionPLANNED
High-value signals are taken straight off the OBD port and CAN bus, without manufacturer telematics subscriptions.
PLANNED - Failure prediction from sparse signalsPLANNED
Models built for sparse, noisy fleet data flag likely failures early enough to schedule the repair.
PLANNED - Duty-cycle and abuse detectionPLANNED
Hard use, overloading and abusive driving patterns are detected and attributed per vehicle.
PLANNED - Condition-based maintenance schedulingPLANNED
Maintenance is scheduled from measured condition instead of mileage intervals alone.
PLANNED - On-vehicle data retentionPLANNED
Raw data stays on the vehicle and only findings leave it, so there is no raw feed to a cloud.
PLANNED - Mixed-fleet coveragePLANNED
One system covers trucks, vans and equipment across manufacturers and model years.
PLANNED - Fleet health dashboardPLANNED
The whole fleet’s condition is visible on one screen, ranked by what needs attention first.
PLANNED - Low-cost hardware targetPLANNED
The on-vehicle hardware is engineered to a cost target under 300 dollars per vehicle.
PLANNED
Updated as capabilities ship. BUILT means working and validated; nothing gets that label early.
03 System
- 01Vehicle bus
- 02Edge decoder
- 03Prediction models
- 04Fleet health view
- 05Maintenance scheduling
04 Engineering
Connectivity and signal core
The planned architecture pairs our protocol and connectivity core with the ALIA Signal Engine. The connectivity core handles OBD and CAN decoding across manufacturers and model years, including signals the manufacturer never documented, and the Signal Engine turns the decoded stream into condition estimates and failure predictions designed for sparse, noisy fleet data.
Deployment model
Processing is on-vehicle by design. Raw bus data is retained on the device and only findings and health summaries leave it, so there is no raw vehicle feed to any cloud. The on-vehicle hardware is engineered to a cost target under 300 dollars per vehicle.
Integration surface
The device is designed to install at the OBD port with no vehicle modification. Findings are intended to surface in the fleet health dashboard and flow into whatever maintenance calendar the fleet already keeps, with no manufacturer telematics subscription involved at any point.
05 Why this is hard
The genuine problem is twofold: reliable decoding across vehicle years, makes and buses that were never meant to be read by a third party, and useful failure prediction from signals that are sparse, irregular and different per vehicle. A model that works on one van and fails on the truck parked beside it is worthless to a mixed fleet. The engineering work is a decoding library that degrades gracefully and prediction models that are honest about their own confidence.
06 Design constraints
- No manufacturer telematics subscriptions
- Raw data never leaves the vehicle
- One system across makes, years and vehicle types
- Hardware cost target under 300 dollars per vehicle
- Designed to install at the OBD port, no vehicle modification
07 Design partner program
Design partner program.
We are looking for a small number of design and pilot partners. A partner gets the problem solved for their operation first, direct influence on the specification above, and pilot pricing. In exchange we get reality: real constraints, real data, real operators. No external partner is engaged on this product yet.