El Agente F\'armaco: A Scalable AI Scientist for Drug Discovery
Summary
- Researchers introduce El Agente Fármaco, an LLM-powered AI scientist that transforms natural language discovery goals into complex, scalable computational drug screening workflows.
- The system utilizes a shared scientific memory called Fármaco Atlas to store molecules, structures, and results alongside their complete provenance for traceable decision-making.
- Calculations can be launched for entire collections of compounds at once without requiring separate per-compound LLM decisions, reducing token use and peak context size.
- In a retrospective comparison, the agent proposed chemically distinct candidates with competitive docking scores and higher calculated drug-likeness compared to expert-led campaigns.
Why it matters
El Agente Fármaco demonstrates how agentic LLM systems can efficiently automate complex scientific discovery pipelines while maintaining a traceable record of computational work, lowering barriers in drug development.