For years, we heard from scientists that biology doesn't have enough clean data for AI. That it’s too variable, too hands-on, too hard to scale, and too expensive to do in the US. Even after two decades of lab automation, only about 5% of lab instruments are actually automated, and much of that work has been outsourced overseas. Science needs AI that learns from variability, and that capability is too important to build outside of the US.
That is why we built the Physical AI Scientist: a system that perceives the lab, runs experiments, and continuously improves its own experimental design. In just three years, Medra’s Vision Language Lab Action model has learned to operate more than 75% of the instruments scientists already use. It sees what is happening on the bench and catches errors as they occur. It reads the literature, analyzes results, and decides what to try next.
We’re running in production with partners like Genentech, across antibody discovery, protein engineering, gene editing, genomics, and cell biology.
Now we are scaling: 38,000 square feet, 100s of robots. Running 24/7. The largest autonomous lab in the United States, built under 90 days. A single platform that covers the entire design-make-test-analyze cycle: hypothesis generation, experiment design, scientific reasoning, and physical execution, all under one roof - in Medra Lab 001 (ML001).
We built ML001 to serve two kinds of partners. For the foundation model teams now building the next generation of models for biology, ML001 is a data foundry. It delivers the same post-training loop that has made language models so capable, now for science, and without the need to stand up a wet lab of your own. For pharma companies thinking about owning their autonomous labs, ML001 is a blueprint: a working reference for what can be built, and a system designed to be reproduced faster than it can be built anywhere else.
The physical layer for AI in science is finally here. And it is being built right here in the United States.
We are hiring across science, engineering, deployments, and operations. Join us.
Or come and see what we’re building.