Benchmarking Engine

The math that makes physical AI insurable.

One shared safety benchmark for every robot.

0 1000ABCD

A shared benchmark

One score. Three winners.

ProvidersMake the case for better terms
OperatorsShow your safety record
InsurersGet a benchmark

The proof

Insurers already price safety.

Same loop, now for physical AI.

  1. Safety featureProven on the road
  2. Claims dataFewer losses
  3. PriceBetter terms
92%fewer injury claims for a self-driving fleet than human drivers
~25%fewer injury claims with automatic braking

Sources: Swiss Re and Waymo claims study (2024); Highway Loss Data Institute (2021).

The engine

Answers in. A rating out.

RobotSiteEvidenceENGINEScore / 1000Tier A–DPart by part RobotSiteEvidenceENGINEScore / 1000Tier A–DPart by part

Weights, thresholds and calibration stay inside Dunster. Ask us about the method in a demo.

The line-up

One engine. Three products.

LiveSafety RatingRate a robot for free
PilotEngine licenseFor insurers and brokers
PlannedLive risk feedRatings that update from the field

In person

See the method.

We walk insurers, brokers and robot makers through the engine in person.

Ratings are indicative only: not a certification, an insurance quote or an offer of cover. Built on the Dunster Physical AI Safety Principles.