About Radar
A world worth
learning from.
We’re building datasets and environments for the next generation of intelligence—grounded in complex physical systems and real work.
Intelligence meets the field.
Our physical-world work starts with agriculture: living systems, changing conditions, and decisions whose consequences unfold over days, weeks, and seasons. We’re interested in the connections between what is observed, what is done, and what happens next.
Longer horizons. Richer worlds.
We’re developing physical-world datasets, complex simulations, and long-horizon reinforcement learning environments for frontier AI labs and robotics teams. Agriculture, biology, and robotics are connected parts of that direction.
From FDE to RL.
Forward-deployed engineering and managed workflow delivery remain part of Radar. We build and operate software around difficult work, then explore how permissioned execution data can inform evaluation and learning environments. Our active project work continues alongside physical-world research.
Built with the people who know.
Expert judgment gives observations context. Our approach centers on the people closest to the work, with permission required for each source and intended use.