Congruent is an automotive radar company solving a structural gap in how autonomous vehicles are trained. The most advanced AV systems today are trained end-to-end — a single neural network learns driving actions directly from raw sensor data. Every sensor in the stack must provide two things: raw data access and a high-fidelity simulator. Cameras have both. Lidars have both. Radar has neither.
Every automotive radar on the market discards over 99% of raw sensor data before the model sees it, and no world-model radar simulator exists for driving scenes. This locks radar out of modern AV training pipelines entirely.
Congruent solves both problems: a radar architecture that exposes fully unprocessed raw data, paired with the first-ever generative world-model radar simulator. Radar is the only depth sensor priced for mass deployment — tens of dollars per unit, already in 90% of US cars, and operational in rain, snow, fog, and dust. Congruent is building the radar that makes mass-market autonomous vehicles possible.
