Laplace Intelligence, an Ann Arbor-based startup, has developed an AI-powered simulation platform designed to identify safety risks in autonomous vehicles, robots, and other physical AI systems before they operate in real-world environments. The company uses automated testing agents to generate challenging scenarios and detect potential failures without exposing people or equipment to danger.
Founded in 2025 by CEO Howie Sun and chief scientist Henry Liu, Laplace was selected as one of 20 startups competing in TechCrunch Disrupt 2026’s Startup Battlefield competition. The company presented its technology on October 13 in San Francisco, demonstrating how AI can test other AI systems through interactive simulations.
How Laplace Tests Physical AI Systems
Laplace combines high-fidelity virtual environments with autonomous testing agents that deliberately search for weaknesses in AI-controlled systems. Its simulation platform uses a modified version of Nvidia’s Cosmos technology to reconstruct real locations, including highways, intersections, and pedestrian crossings.
The system combines map data, photographs, and other geographical information to create environments with realistic dimensions. Developers can modify conditions using natural language instructions, while traffic information from providers such as TomTom helps reproduce actual congestion patterns.
Unlike simulations that simply replay recorded events, Laplace models how vehicles, pedestrians, and other participants respond to changing conditions. During one demonstration, an approaching vehicle reacted to another car entering its lane by changing course to avoid a collision.
The company’s AI agents can also generate unusual or dangerous scenarios using information from news reports, collision records, and traffic data. These tests are intended to identify situations that autonomous systems may encounter but have not previously experienced during development.
Research Origins and Commercial Customers
Sun spent eight years researching autonomous driving simulations before establishing Laplace with Liu, his former doctoral adviser at the University of Michigan. Their research included a 2023 study on autonomous vehicle safety published in Nature, which subsequently attracted interest from automotive companies.
Laplace currently generates revenue primarily from autonomous driving applications, with customers including Japanese truck manufacturer Isuzu and Texas-based autonomous trucking company Bot Auto. Sun said the startup has already achieved a modest profit, although he did not disclose its revenue figures.
Nvidia is also a customer, according to Sun, and the companies are collaborating with T-Mobile on a free smart-city pilot in San Jose, California. The project focuses on traffic operations, while Laplace is separately beginning work with companies developing humanoid robots for warehouses.
Laplace Plans to Expand Its Simulation Business
Most of Laplace’s customer agreements are currently project-based, although the company plans to develop recurring revenue arrangements. Sun is also seeking additional funding to expand operations and compete with established simulation providers.
One competitor is Applied Intuition, which has raised approximately $1.2 billion and reached a $15 billion valuation. Laplace is focusing on automated scenario generation and interactive testing as it expands from autonomous vehicles into robotics and other physical AI applications.
Featured image credits: Magnific.com
For more stories like it, click the +Follow button at the top of this page to follow us.