AiChemy: Next-Generation Agent with MCP, Skills and Custom Data for Drug Discovery

AiChemy is a multi-agent AI assistant built on Databricks for drug discovery workflows. It integrates external MCP servers (OpenTargets, PubChem, PubMed) with proprietary chemical libraries using Databricks Vector Search and Genie (text-to-SQL). The system supports disease-to-target-to-drug workflows, including chemical similarity search against the ZINC15 database using ECFP molecular fingerprint embeddings. It can be assembled via no-code Agent Bricks or coded with LangGraph, deployed as an MLflow AgentServer with a React UI on Databricks Apps, and all invocations are traced via MLflow with OpenTelemetry for observability and evaluation.