How Insilico Medicine Is Using AI To Reinvent Drug Discovery

Drug discovery has always been one of the slowest and most expensive games in business, often taking around a decade and more than $2 billion to bring a new treatment to market. Insilico Medicine believes generative AI can radically rewrite that equation.

The company has already achieved something no one had done before, using AI from end to end to design a drug candidate that reached phase 2 clinical trials. Now it is tackling an even harder question. Can one remarkable breakthrough be turned into a repeatable system capable of producing new treatments at scale?

That is where this story becomes relevant far beyond pharmaceuticals. Insilico is showing how companies can move from an impressive AI proof of concept to the infrastructure, platforms and business models needed to transform an entire industry.

How Has Insilico Used Generative AI?

Insilico’s big medical breakthrough, published in Nature Medicine in 2025, involved using generative AI to design a molecule that can be used therapeutically to treat pulmonary fibrosis.

Its first notable breakthrough was getting its candidate through phase 2 trials, achieving clinical validation of a drug designed end-to-end by AI for the first time. The process took under 18 months, rather than a decade as is usually expected.

This was done using Chemistry42, a tool for generating virtual molecules, which evaluated 78,000 potential candidates before outputting a list of the 60 that it considered most likely to be successful.

So far so good, but to grow a pharmaceutical business, this needs to be done at scale. So it worked on further developing its proprietary discovery platform, Pharma.AI. This is built around Chemistry42 as well as other tools it has developed like PandaOmics, for ingesting and processing scientific data and trial results.

Most recently, the platform was bolstered by the addition of PandaClaw, adding agentic capabilities that allow it to work on longer and more complex experiments with less human involvement.

However, even with all this AI power, there are only so many new drugs that one company can discover.

So, Insilico licensed its platform to other pharmaceutical researchers. As of writing, 13 of the world’s top 20 pharmaceutical companies are using it to run their own drug discovery projects. Between them, work is underway to create new treatments for life-threatening diseases including cancer, heart disease and degenerative neurological conditions.

What this demonstrates is businesses beginning to put in infrastructure in place to enable AI to reach the scale needed for it to be truly transformational. Not just for businesses, but entire industries, as well as our understanding of critical scientific fields like medicine.

What Can We Learn From This?

Insilico started with identifying a problem that AI was uniquely equipped to solve, and developing a process for doing it again and again.

When it realized it had built something that worked for its own problems, it packaged it into a product it could sell to others to help them do the same thing.

Why let its competitors have access to its revolutionary new technology? Because it allows it to position itself not just as another pharmaceutical research company, but as a provider of a solution that can create value for the whole industry. In other words, increasing the size of the whole pie, rather than just its own slice.

This comes down to thinking like an AI company, and, like Google and Amazon, understanding that owning the infrastructure empowering an industry can often be more rewarding than simply being a player in that industry.

For business leaders, the real opportunity lies in identifying the problems AI can solve better than any other technology, and using those solutions to create new sources of value and growth.

These kinds of problems exist in every industry, and those who are not just capable of identifying them, but building infrastructure to solve them at scale, will be the drivers of real, meaningful digital transformation.