Obvious Ventures led a $20 million Series A for Mithrl, with Headline, AGI House and a group of pharmaceutical executives also backing the round, which pays for a second generation of the company’s AI platform for drug research.
Two pieces do the work. One is a biomedical world model; the other is what Mithrl calls an agentic harness, sitting on top to sort model routing, trim token use and orchestrate content. The company’s stated goal is to move research teams from an idea to an investigational new drug filing 50% faster.
September 21 is when early access to the second-generation platform begins. Mithrl says the models rely on validated biology curated from decades of peer-reviewed literature plus public data, rather than on scale alone.
Founder Vivek Adarsh started the company three years ago, after his grandfather died of acute kidney disease. No FDA-approved drug existed for the condition then, he has said, and comparable gaps persist in other disease areas. His question, he told GEN, was what technology could do to get medicines to patients far more quickly.
Why it matters: world models, which learn a domain’s underlying dynamics instead of one narrow task, are becoming a structural bet in drug discovery. For buyers, the claim that matters is not benchmark performance but whether a platform shortens the long and expensive stretch between a validated target and a filing. Mithrl is chasing research organizations that already run models of their own, so the sale rests on orchestration at least as much as prediction.
