Laplace Intelligence tests robots and self-driving cars in simulation

In short: Laplace Intelligence says it uses AI-run simulations to find safety and reliability problems in robots and self-driving vehicles before they are deployed.

What happened

Laplace Intelligence is a startup from Ann Arbor, Michigan, created by researchers from the University of Michigan. It builds software that “red-teams” robots and autonomous vehicles, meaning it deliberately tries to make them fail so companies can fix problems early.

The company does this inside an interactive simulation, which is like a very detailed video game version of the real world. Laplace says it can recreate specific places, such as a highway in Arizona or a busy crosswalk in Paris, using real-world map and street data. Developers can also change conditions on the fly, like adding afternoon traffic.

Laplace also builds AI agents, which are programs that can act on their own toward a goal. In this case, the goal is to invent tricky, rare situations that might confuse a robot or self-driving system, based on things like news reports or police collision reports.

TechCrunch reports the company is already bringing in revenue and says it has turned a modest profit. Customers mentioned include Isuzu and Bot Auto, and CEO Howie Sun says Nvidia is also a customer. Laplace is competing in TechCrunch Disrupt 2026’s Startup Battlefield pitch competition.

Why it matters

As more AI moves from screens into the physical world, mistakes can cause real damage. Testing in simulation first is like doing a fire drill before an actual fire, and it can help companies, regulators, and insurers feel more confident that these systems will behave safely.

Source: TechCrunch AI