UBI programs on real driving data
Continuous, consent-gated driving-behavior data from the vehicle itself — not the phone, not self-reporting. Distance-weighted UBI scores, harsh-event detection with GPS, and crash-proxy incidents designed for FNOL triage.
The problems we remove
Rating on proxies, not behavior
Traditional underwriting uses demographics and postal codes. Two drivers with identical profiles can behave completely differently on the road — proxies can't see it.
Behavior-based risk segmentation
Per-vehicle safety scores computed from actual braking, acceleration, and speeding across every driving day, distance-weighted so high-mileage evidence dominates. Continuous, not snapshot.
Claims with no data
When an incident happens, adjusters get a self-reported account and maybe a photo. Reconstruction depends on memory and witnesses.
Crash-proxy incident records
Harsh-deceleration spikes (≥25 kph/s) are captured as incident events with speeds before/after, decel magnitude, severity band, and GPS coordinates when consented — timestamped, objective, automatic.
Policyholder engagement gap
Insurers touch policyholders at renewal and claims only. Between those moments there's no relationship — and no reason to stay.
Engagement primitives included
Safety leaderboards, DriveCoins reward points tied to safe driving, and per-trip feedback loops — the mechanics that make telematics programs sticky, exposed via API.
What you can build on day one
Safety score (0–100)
Braking 30% + acceleration 30% + speeding 40%, scored per driving day, aggregated distance-weighted. Days under 1 km are unscored — no noise from driveway shuffles.
Component scores
Braking, acceleration, speeding, and eco scores independently — identify exactly which behavior drives a driver's risk profile.
Incident webhooks
vehicle.incident events push to your endpoint HMAC-signed the moment a crash proxy fires — straight into your FNOL workflow.
Behavior summaries
30-day aggregates per vehicle: distance, driving time, max/avg speed, harsh-accel and harsh-brake counts. One call for program reporting.
Consent audit trail
Every data grant is an explicit owner opt-in with revocable scope. Provenance fields mark every sample's confidence level.
Canadian compliance
Built for PIPEDA from the start: consent-first capture, purpose-scoped permissions, and full revocation paths.
Price risk on how people actually drive
One API for scoring, incidents, and engagement. Pilot with a handful of policyholders in days — no hardware program to stand up.