Startup story #26 - InVirtuoLabs

Traditional drug discovery is one of the most complex and expensive processes in science, taking typically over 10 years and more than $2 billion to bring a single drug to market, with fewer than 10% of candidates surviving clinical trials. InVirtuoLabs is on a mission to radically transform this situation by combining generative AI, molecular simulations, and machine learning in a proprietary next-generation computational platform, InVirtuoGEN, capable of molecular property prediction, de novo molecule generation, and fragment-constrained design to discover drugs faster, cheaper, and with higher chances of success. In this interview, co-founder and CEO Gianvito Grasso shares the story behind InVirtuoLabs and their vision of the next frontier in AI-powered drug design.

How did the InVirtuoLabs story begin?

I hold a PhD in Computational Science from USI and have spent the past decade conducting research at IDSIA, with a focus on molecular simulations and machine learning. At the beginning of 2024, following the advent of generative AI, I realised we were witnessing a technological shift, especially in the field of drug design, where new models were being published almost daily in leading scientific journals. With a couple of startup experiences behind me, I felt the moment was right to explore how these emerging tools could be applied to transform the traditional drug discovery process. Confident in the direction, I began building a team, reaching out to Stefano, a fellow computational science PhD and long-time collaborator in various projects, and Sertaç, whom I met during a previous startup experience.

What exactly does your technology do?

Drug development today is costly, slow, and highly uncertain, with fewer than 10% of drug candidates reaching approval. This process leaves many diseases unaddressed, and we are trying to tackle this issue by developing InVirtuoGEN, our proprietary platform that merges generative AI and molecular simulations. These models act like large language models (LLMs), but instead of generating sentences, they generate chemical structures to computationally test billions of potential molecules. This approach allows us to simulate each molecule to evaluate its validity, chemical feasibility, and potential to bind to biological targets, overcoming the lack of high-quality and diverse data in the field and continuously improving the accuracy and robustness of our models.

How is InVirtuoLabs different from other competitors?

First, most companies in this space were founded before the generative AI revolution, while we started from day one with generative chemistry at the foundation of our business. Second, we aligned platform development with real-world drug discovery from the beginning, ensuring a continuous feedback loop between tech and science, while many companies build platforms first and look for use cases later. Lastly, rather than relying solely on existing datasets, we generate our own using molecular simulations to overcome the widespread problem of poor data.

What challenges have you faced so far?

Without a doubt, building the right team, the one truly functional for the company’s purpose, has been both the hardest and most critical part because drug discovery requires expertise across different domains like machine learning, chemistry, molecular biology, and pharma, fields that often don't speak the same language.

What are your next milestones?

By the end of this year, we aim to prove the real-world value of our technology, both through finalising the infrastructure needed to scale our proprietary pipeline and by forming strategic collaborations with biotech and pharma companies. In the long term, our business model is to develop drug candidates to early clinical stages and then out-license them.

3 quick questions to wrap it up:

• If you weren’t in biotech, what would you be doing? Probably a lawyer.
• What’s the best advice you’d give to new founders? Build a strong network. It's more important than cash early on because it helps you find co-founders and gain credibility.
• If you could have a conversation with anyone in history? Albert Einstein.