Ono Pharmaceutical brings the Biomni Lab AI agent into drug discovery

Ono Pharmaceutical brings the Biomni Lab AI agent into drug discovery

On July 28, Ono Pharmaceutical and Phylo announced that researchers at the Japanese pharmaceutical company will use Biomni Lab, an agentic system for biomedical tasks. According to the companies, it compiles experimental histories, works with internal data, proposes the next experiment, and performs computational analyses.

Drug discovery does not depend on a single successful hypothesis. Researchers select a target, compare the results of earlier experiments with new data, decide how to test an idea, and interpret the outcome. At every stage, they must find records, run software, and explain to colleagues why a particular experiment should come next.

On July 28, Ono Pharmaceutical and Phylo announced that Biomni Lab will become part of this process for researchers developing new drugs. A scientist defines the task, and the agent selects a sequence of actions: it connects the experimental history with internal data, helps plan the experiment, and runs computational biology tasks.

Biomni Lab is intended to preserve not only the answer, but also the process used to reach it. In a Science paper on Biomni, the authors described an agent that selects databases and software, creates a plan, executes code, and saves intermediate results. This allows researchers to see which data supported the analysis and how that analysis led to the proposed next experiment.

“Our researchers quickly learned to use Biomni Lab and recognized its potential to accelerate everyday drug discovery,” said Seishi Katsumata, Corporate Officer and Executive Vice President of Research and Drug Discovery at Ono.

In the laboratory, such an agent does not develop a biological hypothesis in place of a human researcher. It connects the testing of that hypothesis with the next decision: it compiles results, prepares the analysis, and helps design the experiment. If this workflow becomes established at Ono, researchers will have a shared record of how data led to an experiment and how that experiment led to the next task.

For Phylo, this marks a move from a standalone demonstration into the daily work of a pharmaceutical company. Ono is introducing the agentic system into a process in which researchers repeatedly decide which experiment to conduct next.