Laplace Intelligence subjects robots and autonomous vehicles to tricky scenarios in simulation before they are deployed in the field

Spun out of the University of Michigan, the start-up uses AI agents and Nvidia’s Cosmos simulator to automatically generate edge cases. It is already profitable thanks to autonomous driving.

Founded a year ago by Howie Sun and Henry Liu as a spin-off from the University of Michigan, Laplace Intelligence carries out ‘red teaming’ for autonomous machines: before a robot or driverless vehicle is deployed, the start-up uses simulation to identify situations that will cause it to fail. Howie Sun spent eight years researching simulation for autonomous driving, and an article published in *Nature* in 2023 caught the attention of industry players.

Agents that devise worst-case scenarios

The platform combines Nvidia’s Cosmos world simulator with interactive participants and AI agents tasked with automatically devising edge cases. Clients include Bot Auto, a Texan specialist in autonomous lorries, a major Japanese car manufacturer, as well as Nvidia and T-Mobile as part of a smart city project in San José. Partnerships are also being established with manufacturers of humanoid robots designed for use in warehouses.

The company claims to be already profitable, with as yet modest revenues from one-off contracts that it aims to convert into subscriptions. It faces stiff competition from Applied Intuition, which has raised $1.2 billion at a valuation of $15 billion.