
Robocurve, a three-month-old San Francisco startup, has secured $10 million in seed funding led by Initialized Capital to establish itself as an independent auditor for physical AI systems. The company plans to test how well frontier models control real-world robots and publish the results publicly, positioning itself as a Public Benefit Corporation with a legal duty to transparency. Early findings suggest general-purpose large language models can outperform specialized robotics models on certain tasks, challenging current industry assumptions. The startup aims to fund academic teams and provide hardware to create open-source benchmarks, addressing the growing need for third-party verification in the rapidly expanding physical AI sector.
Read original