We’re an applied research lab, building the evals and data that advance AI

We’re a team of researchers, engineers, and domain experts from Meta, Google, and Autodesk, previously leading engineering and research at frontier labs. We build evals, benchmarks, and expert-grade training environments in domains where AI still falls short, from RSI to 3D design and engineering. We work with frontier labs, industry leaders, and robotics teams across the full cycle, from defining what to measure to delivering data ready for training.