Rentosertib, a drug whose target and design both came from AI, shows early promise against a deadly lung disease

A medicine called rentosertib has been described as the first drug for which both the biological target and the molecule itself were discovered using generative artificial intelligence, and recent results suggest the approach can produce something that actually helps patients. Developed by Insilico Medicine and previously known as ISM001-055, the drug targets a protein called TNIK to treat idiopathic pulmonary fibrosis, a progressive scarring of the lungs that affects roughly five million people worldwide and carries a median survival of only three to four years.

Insilico says it identified the disease target with its PandaOmics biology engine and designed the compound with its Chemistry42 generative-chemistry system, completing that early work in about 18 months, far faster than the years such steps traditionally take. In a Phase IIa trial, patients given the highest 60-milligram dose over 12 weeks showed a mean improvement of 98.4 milliliters in forced vital capacity, a key measure of lung function, while the placebo group declined by 62.3 milliliters. The trial met its primary safety goals and pointed to dose-dependent benefits and reduced cough.

“Developed by Insilico Medicine and previously known as ISM001-055, the drug targets a protein called TNIK to treat idiopathic pulmonary fibrosis, a progressive scarring of the lungs that affects roughly five million people worldwide and carries a median survival of only three to four years.”

The drug's progress has drawn major industry attention, including a multibillion-dollar collaboration to advance Insilico's molecules toward the global market, and more than 170 AI-designed drugs are now in human testing across the field. The U.S. Adopted Names Council granting rentosertib an official generic name marks a formal step on the long path toward potential approval.

The cautions are important and real. Phase IIa is an early, relatively small study; promising lung-function numbers over 12 weeks are not proof of long-term benefit, and most drugs that look good at this stage still fail in larger trials. AI shortened the discovery phase, but the slow, expensive work of clinical testing remains fully human and far from finished. Even so, for a disease with few good options, a genuinely novel candidate that emerged from an AI-driven pipeline is a meaningful sign that these tools can help bring real new medicines toward patients.