Interview with Hans Melo

As artificial intelligence continues to redefine industries, its impact on drug discovery is becoming increasingly profound. Yet beyond the headlines and hype lies a more nuanced reality—one shaped by data limitations, scientific complexity, and the need for collaboration.

In a recent conversation with Oxford Global’s Cerlin Roberts, Hans Melo, CEO of Menten AI, shared his perspective on how generative AI is beginning to transform one of the most challenging frontiers in pharmaceutical research: peptide macrocycle design.

AI Before the Hype

Menten AI was founded in 2020—before the widespread attention generated by large language models and tools like ChatGPT. At the time, the company recognised a critical opportunity: machine learning could address some of the most difficult problems in drug discovery.

Rather than pursuing well-trodden areas, the team focused on peptide macrocycles—a class of molecules known for their therapeutic potential, but also for their complexity.

“It’s a challenging space,” Melo explained, “and that’s exactly why it matters.”

The Data Dilemma

Despite its promise, AI in drug discovery faces a fundamental obstacle: data scarcity.

Machine learning models typically rely on large, high-quality datasets. Yet in many areas of pharmaceutical research—particularly novel modalities—such data simply does not exist.

This creates a paradox. The areas where AI could have the greatest impact are often those where it has the least data to learn from.

Compounding this challenge is the siloed nature of the industry, where valuable datasets are often isolated within individual organisations. Without broader data sharing and collaboration, progress risks being slower than it could be.

AI Across the Discovery Pipeline

Even with these challenges, AI is already reshaping drug discovery—from target identification to hit discovery, molecular design, and beyond.

Melo points to the growing ability of AI to uncover novel biological targets that may have been overlooked using traditional methods. At the same time, generative models are enabling the design of entirely new molecules, spanning small molecules, peptides, antibodies, and RNA-based therapies.

The implications are significant: faster discovery timelines, improved safety profiles, and more effective treatments for patients.

Breaking Beyond Protein-Centric Models

One of the most exciting frontiers lies in moving beyond traditional, protein-based datasets.

Many existing AI models are built on protein data, which is inherently limited to the 20 natural amino acids. But the future of drug design may lie in expanding beyond this biological constraint—exploring non-natural amino acids and entirely new chemistries.

This shift, however, introduces a new challenge: how to train AI models in environments where data is sparse or non-existent.

The solution, Melo suggests, will require new methodologies—approaches that allow AI to function effectively without relying solely on large datasets.

Rethinking AI Adoption

For organisations navigating this evolving landscape, Melo offers a clear message: do not fear AI.

“There’s a misconception that AI will replace scientists,” he noted. “But in reality, it should be seen as a co-worker.”

Rather than replacing human expertise, AI has the potential to augment and accelerate decision-making—acting as a “co-pilot” that enhances productivity and insight.

Equally important is the need for a cultural shift. Successful AI adoption depends not just on technology, but on collaboration and openness—sharing data, exchanging knowledge, and learning collectively.

The Next Five Years

Looking ahead, the pace of innovation in AI-driven biomedicine is expected to accelerate rapidly.

Advances in generative models, novel chemistry, and data integration will likely expand the boundaries of what is possible in drug design. At the same time, new approaches to working with limited data could unlock entirely new areas of discovery.

For companies like Menten AI, the goal is clear: to harness AI not just as a tool, but as a transformative force in how medicines are conceived and developed.

Looking Ahead

As drug discovery becomes increasingly complex, the integration of AI will be less about replacing traditional methods and more about redefining them.

The future will belong to those who can combine computational power with scientific insight—leveraging AI to explore new chemical spaces, uncover hidden biological pathways, and ultimately deliver better therapies.

In this emerging landscape, one thing is certain: the intersection of AI and biology is no longer optional. It is foundational.