What happened
Insilico Medicine is advancing its AI-discovered drug, rentosertib, to Phase III human trials. The drug targets idiopathic pulmonary fibrosis (IPF), a fatal lung disease. The move follows a Phase II trial where patients on a 60mg daily dose showed a mean forced vital capacity gain of +98.4 mL over 12 weeks, compared to a 20.3 mL loss in the placebo group, as reported by AI News.
The US FDA granted the drug 'Orphan Drug Designation' in February 2023. Rentosertib was identified and designed using Insilico's proprietary Pharma.AI platform, which handled everything from target discovery to generative molecular design.
How the room's reading it
The AI drug discovery field sees this as a pivotal moment. For years, the story has been about speed — how AI can accelerate target identification and molecule generation. Insilico's progress shifts the narrative to clinical translation. Practitioners are pointing to this as a complete, documented case of an AI platform originating new biology and chemistry that survives the rigours of human trials.
Deep tech watchers and investors are framing this as hard evidence that AI can deliver in capital-intensive, highly regulated industries. The consensus is that this isn't just an optimisation play. It’s about using AI to solve problems that were previously intractable, providing a crucial proof point that the technology can create novel therapeutic opportunities, not just find them faster.
Sailfish's take
We see this as a lighthouse for builders in every hard-tech sector. The win here isn't just the drug — it's the validation of an end-to-end, AI-native workflow in a domain with zero tolerance for error. For too long, "AI" has been synonymous with software and language. This is a powerful reminder that the same principles can be applied to atoms, not just bits.
We've seen countless pitches for AI in materials science or industrial processes that get dismissed as too ambitious. Insilico provides the proof point they've all been missing. We think this signals the real beginning of generative AI for the physical world. If you're building for complex, regulated systems, this is the case study to have in your back pocket.