University of Michigan spinout Laplace Intelligence is developing AI-driven simulations to find failure cases in autonomous vehicles, robots and smart-city systems.
Laplace Intelligence is developing simulation software that uses AI agents to probe autonomous vehicles, robots and other physical AI systems for weaknesses before real-world deployment. The Ann Arbor startup’s approach combines environments modeled on real locations with agents designed to generate difficult or unusual scenarios for testing.
The simulation platform is based on a post-trained version of Nvidia’s Cosmos. Laplace combines map and other real-world information to build to-scale settings, and says developers can modify those environments with language prompts. The company also draws on live traffic and congestion information from providers such as TomTom. In a demonstration described by TechCrunch, simulated road users responded to a test vehicle’s actions rather than behaving as static scenery.
Laplace’s founders, CEO Howie Sun and chief scientist Henry Liu, came out of the University of Michigan. Sun had spent eight years researching ways to challenge autonomous-driving AI in simulation, and the pair published related research in 2023. Sun told TechCrunch that customer interest in applying that work helped move the project from academia toward a commercial product.
The startup is already generating revenue and a modest profit, according to the report. Sun said customers include Japanese truck maker Isuzu, autonomous trucking company Bot Auto and Nvidia. Laplace and Nvidia are also working with T-Mobile on a free smart-city pilot in San Jose, California, focused on high-level traffic operations. The company is in early discussions with partners working on warehouse humanoid robots.
Most contracts are project-based, and Laplace is seeking to develop recurring revenue as it raises funds to scale. It competed in TechCrunch’s Startup Battlefield 200 at Disrupt in San Francisco, held October 13–15, 2026. TechCrunch identified Applied Intuition, which serves Isuzu as well, as a competitor. Laplace’s near-term customer base is concentrated in autonomous driving, while its founders see the approach as potentially useful for robotics, where they say fewer real-world examples are available for testing rare edge cases.