Laplace Intelligence Uses AI to Crash-Test Robots Before Deployment

A one-year-old Ann Arbor startup founded by two University of Michigan academics is already profitable — and it counts Nvidia, Isuzu and Bot Auto among its customers. Laplace Intelligence sells automated red-teaming for robotics and physical AI: deliberate stress-testing designed to make autonomous systems fail in simulation before they meet the real world.

CEO and co-founder Howie Sun frames the problem bluntly. "AI agents are moving from computers into the real world, and we don't really want physical AI to fail in the real world," he told TechCrunch. The trigger for the company's pitch is recent history: unpredictable AI agent behavior of the kind seen in the OpenAI-Hugging Face incident.

What does Laplace actually build?

At the core of the product sits a high-fidelity simulation reconstructed from real-world data, paired with an AI agent that continuously probes for weaknesses. The simulation environment is a post-trained version of Nvidia's Cosmos, an AI world simulation platform, grounded in real-world inputs. Laplace blends a mesh of real-world information — a Google Street View image can serve as the reference from which an entire street is generated — combined with real map data so environments stay true to scale.

Developers can change settings in real time through language prompts, such as adding afternoon traffic. Sun says Laplace pulls live traffic and congestion data from navigation providers including TomTom. In one demo, Sun directed a test vehicle to swerve into oncoming traffic; the oncoming car moved aside rather than letting itself be driven through, a common flaw in some generative world models.

"It's truly an interactive simulation environment," Sun said. "Everyone is reacting to everything. …This is more like a world model with interactive participants."

The name is deliberate. "Laplace's demon," an 1814 thought experiment by French scholar Pierre-Simon Laplace, posited that knowing the exact state of everything in the universe would allow total prediction of the future. "We are essentially using simulation to predict the real world's future," Sun said.

Why AI-designed corner cases matter

The real differentiator, Sun argues, is not the simulator but the AI agents that automatically design corner cases — rare, tricky scenarios that could trip up an autonomous system — removing humans from the loop.

"They can design from the news, from police reports on collisions — all kinds of information to actively challenge the autonomous vehicles and other physical AI agents," Sun said.

Today, most autonomous vehicle developers train on edge cases using recorded footage converted into simulation, or rely on humans to invent new scenarios. Some AV companies have already embraced AI-generated edge cases — self-driving truck firm Waabi comes to mind. Sun believes the real demand arrives with generalized robotics: "There's less data on generalized robotics. No one knows what the edge cases will be in the real world."

The company's stated mission extends beyond testing. "Our goal is to help companies discover and fix problems before deployment—and, over time, build a shared foundation of evidence that customers, insurers, and regulators can trust," Sun said. That evidence base, if it materializes, could matter as much as the testing itself — though the company has not yet disclosed metrics on how many failures its platform catches.

Who pays, and how much?

Neither founder carries a Silicon Valley tech pedigree, yet Laplace has started generating revenue and turned a modest profit. Most of the money comes from autonomous driving use cases. Customers include Japanese trucking manufacturer Isuzu and Texas-based autonomous trucking firm Bot Auto.

Nvidia is also a customer. Alongside T-Mobile, the two companies are running a free smart city pilot for the City of San Jose to provide high-level traffic operations. Laplace is in early-stage work with partners building humanoid robots for warehouses.

The caveat sits in the revenue structure: most contracts are project-based today, and the company is working to convert them into recurring revenue — a shift it has not yet completed.

Can it survive Applied Intuition?

Laplace's chief rival is Applied Intuition, a late-stage startup that has raised $1.2 billion at a $15 billion post-money valuation. Isuzu uses both. Sun contends his core method — AI testing AI — is a "new paradigm that changes the testing of physical AI entirely," turning it into a "dynamic interactive testing process."

"That will be the biggest moat," he said. A methodology is not, by itself, a $15 billion balance sheet.

The academic roots run deep. Sun spent eight years researching how to challenge autonomous driving AI in simulation. In 2023, he and Liu — his then-PhD advisor and now co-founder and chief scientist — published a paper on the subject that landed on the cover of Nature. Calls from autonomous vehicle companies followed, and Laplace was founded a year ago.

Laplace is one of 20 startups competing in Startup Battlefield, TechCrunch Disrupt's three-day pitch competition in downtown San Francisco, presenting Monday afternoon, October 13. Sun hopes the exposure will help the company raise funds and scale — capital it will need if it intends to fight Applied Intuition for the physical-AI testing market.