Physicl has raised a $10 million Seed round led by Inovia Capital, with participation from US Venture Partners. The company builds exact physical training data for robots, autonomous systems, and world models. Its platform focuses on rare edge cases that break physical AI models.
Funding and team
The new funding supports the company's teams in Paris and the San Francisco Bay Area. It also funds the platform behind Studio. Studio is free during the beta period.
Company context and advisors
Physicl started after its team spent ten years digitizing objects and rooms in 3D for large retailers. Robotics, world model, and vision teams later asked for labeled versions of the cases their models failed on. Advisors include Thomas Wolf, Christoph Lassner, Matthias Niessner, Yaser Sheikh, and Sébastien Deguy.
How Studio works
Studio lets researchers recreate model failures and generate variations for training. Depth, normals, segmentation, boxes, and camera pose come from the scene itself, not from later estimates. It includes 10,000 environments and 100,000 assets built from real geometry with known materials.
Studio is opening to research teams in waves. Labs, universities, and companies building vision, world-model, and robotics models can apply. SpAItial, an early user, more than halved its reconstruction error on real rooms after training with Physicl data.