Key points from article :
A new San Francisco–based startup, Medra, is betting that tightly linking artificial intelligence with robotics in the lab—what it calls “physical AI”—could dramatically speed up drug discovery, including efforts to target aging and age-related diseases. Backed by a $52 million Series A round, Medra aims to solve a long-standing bottleneck in pharma R&D: AI systems can generate hypotheses rapidly, but experimental validation remains slow, fragmented, and hard to feed back into learning models.
Medra’s platform combines two layers. A “scientific AI” system proposes hypotheses, interprets results, and suggests next experimental steps in natural language, while a “physical AI” layer uses general-purpose robots to autonomously run experiments on standard lab equipment. Crucially, results from each experiment are fed straight back into the models, creating a continuous, self-improving loop. Founder and CEO Dr Michelle Lee argues this approach allows researchers to explore vast biological spaces far more efficiently than traditional one-experiment-at-a-time methods.
The company sees particular promise in longevity research, where aging is driven by interacting pathways such as metabolism, inflammation, protein homeostasis, and stress responses. Medra’s system could systematically test combinations of pathways, doses, and timing to uncover synergistic effects that are difficult to detect with conventional workflows. The same logic could apply to complex age-related diseases like neurodegeneration and fibrosis, where biological context matters as much as the target itself.
Experts say the idea addresses a growing mismatch between prediction and validation. Patrick Hsu, a professor at UC Berkeley, notes that AI models now generate insights faster than labs can test them, while existing automation is often too rigid to adapt. Medra’s flexible robotics-plus-AI approach, he says, could help close that gap by scaling data generation and accelerating discovery. Early partnerships, including a collaboration with Genentech, suggest industry interest in this “lab-in-a-loop” model, which could reshape how drugs—especially those targeting aging—are discovered.
