Ann Arbor-based Laplace Intelligence is building automated red-teaming for robotics and physical AI , betting that the only way to make physical AI safe is to have AI test it. CEO Howie Sun says the goal is to help companies find and fix problems before deployment. Over time, Laplace also wants to build a shared foundation of evidence that customers, insurers and regulators can trust. The company is one of 20 startups competing in Startup Battlefield 200 at TechCrunch Disrupt.
Sun and chief scientist Henry Liu founded Laplace a year ago out of the University of Michigan. In 2023 the pair published a paper on challenging autonomous driving AI in simulation that made the cover of Nature, and AV companies soon reached out. The company is named after Laplace's demon, an 1814 thought experiment about predicting the future with total precision. Sun says the company is essentially using simulation to predict the real world's future.
Laplace's simulation environment is a post-trained version of Nvidia Cosmos , grounded in real-world data such as Street View imagery and map data, so scenes are true to scale. Developers can change scenes in real time with language prompts, and live traffic data comes from providers like TomTom. Other actors in the scene react realistically: in a demo, a car moved out of the way to avoid a collision instead of being driven through. Sun describes it as a world model with interactive participants.
Laplace's main selling point is its AI agents, which automatically design corner cases from sources such as news stories and police collision reports. This moves humans further out of the testing loop. AV developers today mostly turn recorded footage into simulations or rely on humans to design scenarios, though some, like Waabi, already use AI. Sun argues the need will grow with generalized robotics, where there is little data and nobody knows what the real-world edge cases will be.
Laplace already generates revenue and a modest profit, mainly from autonomous driving customers including Bot Auto and a major Japanese automaker. It is also running a free smart city pilot in San Jose with Nvidia and T-Mobile, and is in early work with warehouse humanoid robot developers. The company aims to move from project-based contracts to recurring revenue and to raise funding to compete with Applied Intuition, which has raised $1.2 billion at a $15 billion valuation. Sun calls AI testing AI a new paradigm and the company's biggest moat.