AI that runs science on its own.
Iterative Labs builds self-driving laboratories and the training environments around them. The goal is an AI agent that can operate a lab end to end: design an experiment, simulate it, run it on real hardware, make sense of the result, and decide what to try next.
The bottleneck is not intelligence. Models can reason about science on paper, but they have never had to learn from what happens at the bench. We give them a bench.
Today that means reinforcement-learning environments and training data for the science teams at frontier labs. Each environment teaches the whole loop, including the noisy, corrupted, and failed runs. The physical side starts in biology, with automated protocols such as cell-free protein expression run on a wet-lab bench and every real outcome cached as ground truth. Chemistry and materials follow.
Backed by Y CombinatorPreviously at Neuralink, Figure, NVIDIA, and Tesla. Computer engineering at the University of Toronto and the University of Waterloo.
We are a small team building the lab and the training loop around it. If you want models to learn from real experiments rather than from text about them, we are hiring engineers and scientists.
Open roles