Laplace Intelligence is building simulation and AI-agent tools to uncover potential failures in autonomous vehicles and other physical AI systems before deployment.
Laplace Intelligence is developing a testing system that uses simulated environments and AI agents to probe robots, self-driving vehicles and other physical AI for potential failures. The Ann Arbor startup, founded by University of Michigan researchers Howie Sun and Henry Liu about a year before the October 19, 2026 report, grew out of their academic work on challenging autonomous-driving systems in simulation.
The company’s platform combines a high-fidelity virtual environment with agents designed to generate difficult test situations, including rare scenarios that developers may not have encountered in recorded data. Its simulation is based on a post-trained version of Nvidia’s Cosmos platform. Laplace says it can reconstruct places such as an Arizona highway or a Paris pedestrian crossing using real-world information, and developers can adjust conditions with language prompts. The system also models how other participants respond: in a demonstration, a car moved to avoid a test vehicle that was directed into oncoming traffic.
Laplace’s approach is intended to help developers find and address weaknesses before deployment. Sun told TechCrunch that the company also wants to build a shared body of evidence that customers, insurers and regulators could trust. The company draws on sources such as collision reports and news to design challenges for autonomous vehicles and other physical AI systems.
The startup already has customers, including Texas-based autonomous trucking company Bot Auto and a major Japanese automaker, and the article reports that it has generated revenue and a modest profit. Most contracts are project-based, while Laplace works toward recurring revenue. It is also collaborating with Nvidia and T-Mobile on a free smart-city pilot in San Jose focused on high-level traffic operations, and is in early discussions with partners developing warehouse humanoid robots.
Laplace is competing in TechCrunch Disrupt’s Startup Battlefield 200. The company is seeking funding to scale and faces competitors including Applied Intuition, which the article describes as a later-stage startup. Laplace’s current customer work is concentrated in autonomous driving, while its founders see broader robotics testing as a potential application.