Reliable drug discovery agents require structured, harmonized, and domain-aware data connectors that make scientific retrieval and downstream reasoning repeatable. This presentation examines how such connectors support agentic workflows across chemistry, pharmacokinetics, and drug safety.
We will describe an approach that combines controlled taxonomies, data science, domain expertise, and software engineering to retrieve, normalize, harmonize, and package scientific information for downstream agents. We will also set out the design principles that make these connectors reproducible, auditable, and suitable for scientific decision-making.
We will demonstrate the approach through a pharmacokinetics data connector that draws evidence from Elsevier’s discovery, preclinical, and clinical sources and harmonizes the results into a form that agents can use consistently. This example illustrates how integrated scientific data services support agentic workflows that reason over harmonized evidence rather than unstructured retrieval alone.
Speaker
Ivan Krstic is Senior Director of Product Management at Elsevier, responsible for agentic solutions in life sciences. He leads the strategy and delivery of a portfolio of modular data connectors that draw evidence from Elsevier’s discovery, preclinical, and clinical sources and harmonize it into a form that scientific agents can use consistently.
This helps customers reduce the complexity of working with fragmented scientific information and build AI-enabled workflows they can trust in drug discovery and development. Ivan has previously led product management for Reaxys, the chemistry information solution, and Embase, the biomedical information solution.
Ivan holds a PhD in biophysical chemistry from Goethe University Frankfurt.