Vivodyne’s robotic labs test drugs on human tissue

Inside Vivodyne’s facility just outside San Francisco, robotic systems are growing human tissues, adding drugs and tracking what happens next. The startup’s HIVE machines can cultivate 20 kinds of human tissue and run hundreds of thousands of ongoing experiments, turning a familiar bottleneck in drug discovery into an automated stream of biological observations.

The bottleneck is not simply a shortage of computing power. AI models often learn from animal tests or static studies of individual cells and proteins. They can identify that a cell is in state A or state B, Vivodyne CEO and co-founder Andrei Georgescu told TechCrunch, but not necessarily that state B was caused by inflammation in state A. HIVE is designed to record that chain of events by exposing diseased human tissue to different stimuli and measuring the response.

Vivodyne says the early results are promising. Its liver cells showed 94% predictive accuracy against human toxicity trials, while its airway tissue matched the behavior of real human tissue 96% of the time. Its bone marrow reached 100% concordance in tests of 20 chemotherapy drugs. The company has raised just under $80 million across two rounds led by Khosla Ventures and says it is working with multiple major pharmaceutical companies, though it has not named them publicly.

The immediate payoff would be practical: better evidence about which drug candidates are worth taking into clinical trials, where the process typically costs tens of millions of dollars. That matters because 90% of drugs that work in animal testing and enter clinical trials do not receive regulatory approval for humans. If HIVE’s results hold up, researchers could discard more weak candidates earlier and focus scarce trial capacity on treatments with a stronger biological case.

The larger ambition is to give AI something it currently lacks: causal human-biology data for models that can reason about effects, not just patterns. That could become especially useful for combination therapies, where several biological pathways must be targeted at once. But the evidence remains company-reported, and no independent trial of these performance claims has been published in the account examined here. HIVE is operating as an experimental platform, not a proven replacement for human clinical testing.