Laplace Intelligence Builds Robots and Self-Driving Vehicles Before They Reach the Real World
Laplace Intelligence, a University of Michigan spin-off, uses AI agents and interactive simulation to stress-test robots and autonomous vehicles before deployment. The profitable startup is competing in the TechCrunch Disrupt Battlefield.
Disrupt 2026 has officially begun. Get your pass now to be in the center of the startup hub. Register now
Disrupt the doors opening on October 13th. Get your card and bring someone with you with a 50% discount. Register now.
If there's one lesson the OpenAI-Face-Hugging incident taught us, it's that AI agents can be unpredictable.
Howie Sun, CEO and co-founder of Laplace Intelligence, told TechCrunch: “Imagine if something like this happened in the physical world. "AI agents are moving from computers to the real world, and we really don't want physical AI to fail in the real world."
Ann Arbor-based Laplace is building automated red teaming — essentially stress testing a system by deliberately trying to fail it — for robotics and physical AI. The company bets that the only way to be sure its AI is safe is to test its AI.
"Our goal is to help companies discover and fix problems before deployment — and over time, build a common foundation of evidence that customers, insurers and regulators can trust," Sun said.
Laplace is one of 20 startups competing in Startup Battlefield, TechCrunch Disrupt's three-day competition in downtown San Francisco. It will be presented on the Startup Battlefield stage on Monday afternoon, October 13.
Sun, along with co-founder and chief scientist Henry Liu, founded Laplace out of the University of Michigan a year ago. Sun has spent the past eight years at the university researching how artificial intelligence can challenge self-driving simulations. In 2023, he and Liu—his then-doctoral advisor—published a paper on the subject that landed on the cover of the top science journal Nature. That's when the calls started coming from autonomous car companies interested in implementing their ideas.
"I realized I wanted to build this and take it from purely academic research to a product that is useful in the real world," Sun said.
Using their academic research as a foundation, Sun and Liu created Laplace, named after "Deu Laplace," an 1814 thought experiment by French researcher Pierre-Simon Laplace who hypothesized that if you knew the exact state of everything in the world, you could predict the future with perfect accuracy.
"We're basically using simulation to predict the future of the real world," Sun said.
At the startup's core is a high-fidelity simulation recreated from the real world, along with an artificial intelligence agent that constantly checks for weaknesses and points of failure. The goal is to ensure that any physical AI—be it a robot, autonomous vehicle, or smart city solution—is safe in simulation before being deployed in the real world.
The simulation environment is a post-training version of Nvidia's Cosmos, a real-world artificial intelligence simulation platform that allows Laplace to recreate anything from an Arizona highway to a Paris pedestrian crossing. Simulations are created by combining a network of different real-world information. A Google Street View image can serve as a reference image from which a complete street is generated. But instead of generating something random, the simulator combines real-world map data to recreate an environment that's true to scale.
Developers can then use language prompts to change settings in real-time, such as adding afternoon traffic. Sun says Laplace gets real-time traffic and congestion data from navigation providers like TomTom.
In Laplace simulations, not only the environment but also the behavior of other actors are simulated. For example, during a test demonstration, Sun directed the test agent's vehicle to swerve into oncoming traffic. The car it was about to hit – or pass, as is common among some of the world's production models – swerved to avoid a collision.
"It's really an interactive simulation environment," Sun said. "Everybody reacts to everything. ... It's more like a global model with interactive participants."
While the simulated environment is an important part of Laplace's product, the real unique selling point is the AI agents the startup is developing to automatically design corner cases (complex and rare scenarios that can break an AI system) that take humans further out of the loop.
"They can draw from the news, from police reports about collisions — all kinds of information to challenge autonomous cars and other physical AI agents," Sun said.
Today, autonomous vehicle developers mainly train artificial intelligence to handle edge cases using recorded footage and then turn it into a simulation, or rely on humans to come up with new ideas and design them. While many AV companies have done what Laplace advocates and are starting to embrace AI to create edge-frames for training — self-driving truck company Waabi comes to mind — the need for this kind of technology will really become apparent when we start putting generalized robots into the real world.
"There is less information about generalized robotics," Sun said. "No one knows what the edge cases will be in the real world."
Despite Sun and Liu coming from universities, without the glittering pedigrees of a Silicon Valley tech company, the two have been able to start making money and even turn a small profit. Most of the money comes from self-driving use cases, with clients including Japanese truck maker Isuzu and Texas-based autonomous transport company Bot Auto.
Sun says Nvidia is also a customer. The two companies, along with T-Mobile, are working on a free smart city pilot for the city of San Jose to provide high-level traffic operations. Laplace is also in the early stages of working with partners to build humanoid robotics for warehouses.
Today, most of Laplace's client contracts are project-based, and the company is trying to turn that into recurring revenue. The startup's short-term goal — which it hopes will be helped by competing at Disrupt — is to raise the capital and scale to compete with its competitors.
Among its competitors is Applied Intuition, a buzzy late-stage startup that has raised $1.2 billion so far at a $15 billion post-money valuation. Isuzu is also one of Applied's customers.
Sun is not too worried. He believes that Laplace's original method of using AI for AI testing is "a new paradigm that completely changes physical AI testing" and turns it into a "dynamic interactive testing process."
"It's going to be the biggest ditch," he said.
When you make a purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
Rebecca Blanc is a senior reporter at TechCrunch, where she covers business, politics and emerging trends in artificial intelligence. His work has also appeared in Forbes, Bloomberg, The Atlantic, The Daily Beast, and other publications.
You can contact or confirm Rebecca by emailing her at [email protected] or by texting rebeccabellan.491 on Signal.
Get 50% off a second pass The Disrupt experience is meant to be shared. Grab your pass and bring a colleague, partner or peer for 50% off. Cover more ground by building connections, building momentum, and discovering what's next in the startup ecosystem.
Elon Musk attacks Ambani over Jagmeet Singh's India Starlink launch delay
Batteries are already cheaper than the natural gas turbines used in many of Tim de Chant's data centers.
Remember Orkut? Its founder wants to bring back Jagmeet Singh
The United States excludes Microsoft, Adobe and large IT companies from the green card program for skilled workers. Foreign workers Ayesha Malik
Elon Musk questions Ambani influence as Starlink India launches on Jagmeet Singh stands
The 19-year-old founder of Cal AI just raised $10 million for his new AI startup Julie Burt.