On June 27, 2023, Insilico Medicine announced it had dosed the first patients in a Phase II clinical trial of INS018_055, an oral drug candidate for idiopathic pulmonary fibrosis (IPF), a chronic scarring lung disease with a median survival of about three years. Insilico describes INS018_055 as the world’s first anti-fibrotic small-molecule inhibitor whose biological target was discovered by AI and whose molecule was also designed by AI, advanced to mid-stage human trials.
Two of the company’s platforms did the core work. PandaOmics, a target-discovery engine, scored genes and pathways to nominate a novel target - the kinase TNIK, which sits at the intersection of several pro-fibrotic pathways. Chemistry42, a generative-chemistry engine, then designed candidate molecules against that target. Insilico says the path from target discovery to a Phase I trial took under 30 months, roughly half the time it estimates for a traditional campaign.
The “first” claim is worth stating precisely, because earlier AI-designed molecules had reached the clinic. What Insilico emphasizes is the combination: AI choosing the target and AI designing the compound for the same program. That makes INS018_055 a test of the end-to-end promise of AI drug discovery rather than of one isolated step.
A drug entering Phase II is a milestone of process, not of proof. Phase II measures safety and early efficacy in a larger group; most candidates that reach it still fail, and an AI origin does not change that base rate. The significance is that the slow, expensive front end of drug discovery - picking a target and inventing a molecule - was driven by machine learning for a program that a regulator allowed into human trials.