Collect raw telematics events from the vehicle's OBD-II dongle, mobile SDK, or connected-car API, each event containing timestamp, GPS coordinates, speed, acceleration, and hard-braking indicators
Normalize events into trips by segmenting on ignition-on/off events or GPS gap thresholds; compute per-trip metrics: distance driven, time of day, average speed, hard-brake count, rapid-acceleration count, and distracted-driving events
Aggregate trip-level metrics over the policy scoring period (typically 30–90 days) to produce fleet-level or driver-level features: miles driven, percentage of night driving, hard-brake events per 100 miles
Apply your UBI scoring model—which may be a weighted composite or a machine learning model trained on historical loss data—to produce a normalized score that maps to a discount or surcharge percentage
Write the score and its component factors to the policy record; surface the factors to the insured in a driver feedback app to encourage behavioral improvement and support adverse-action disclosure requirements
Known gotchas
GPS signal loss in tunnels or urban canyons creates gaps that can cause trips to be split or merged incorrectly; implement gap-filling heuristics and flag trips with excessive data loss before scoring
Adverse-action disclosure requirements under the FCRA and state insurance regulations may apply if telematics data is used to increase a premium; document which factors drove the score and retain the underlying data
Battery drain and data transmission costs are real constraints for OBD-II dongles; design the data pipeline to tolerate burst uploads when the vehicle connects to WiFi rather than assuming continuous cellular transmission
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp