R&D · Engineering notes
How we read cars without opening the hood.
Notes from the team behind DeepDiag, the data and intelligence layer of Steer AI, the AI infrastructure for the used-car economy: road tests, signals, physics-informed models and the product built on them.
11 notes
Research programmes9 min read
Why Two Identical Cars Never Have the Same Price
Read
Models4 min read
Reading a car without opening the hood
Read
Models5 min read
Physics-informed models, explained
Read
Data3 min read
Same car, same verdict
Read
Models3 min read
What a U-turn tells us about a car
Read
Data3 min read
Battery health from the car's own signals
Read
Models3 min read
Reading the brakes in one controlled stop
Read
Product4 min read
Why we keep everything on one record per car
ReadOur approach
Measure, model, and publish the limits.
Steer AI was founded by an engineer formerly at Renault. We publish our methods, and their limits, in our white papers.
01
Measure on the road
Accelerate, brake, U-turn: three moves on a short road test, recorded through the car's own port.
02
Model with physics
The laws of vehicle behaviour plus the recorded data, so every finding can be explained.
03
Say what we can't see
Each method has limits. We document them, so a report is never read for more than it says.
4Systems measured: engine, battery, suspension, brakes
3Moves on every road test
≈15 minInspection report and trade-in value
÷5Time compared with a standard inspection
The method, in full.
Our white paper covers the road test, the models and their limits.