AI Agents That Reason Over Biomedical Knowledge Graphs

Biomedical knowledge graphs connect research findings across genes, diseases, drugs, and biological processes. Exploring these connections can reveal valuable insights—but often requires specialist knowledge and technical querying skills.

At Innolabs, we developed AI agents capable of reasoning over the existing Monarch and Hetionet knowledge graphs. Our goal is to make this structured knowledge easier to investigate through natural-language questions, while exploring how agents can follow relationships across a graph to address complex research problems.

These demonstrators help us investigate a practical question: how can AI turn connected scientific data into useful, traceable reasoning? They provide a setting for examining how agents explore evidence, connect findings, and where their reasoning still falls short.