Databricks launches AiChemy: multi-agent AI for drug discovery

Databricks launched AiChemy on April 6, 2026, a multi-agent AI system designed to accelerate drug discovery by connecting internal enterprise data with external scientific databases.

The system is built on Databricks' Data Intelligence Platform and uses the Model Context Protocol (MCP) to allow specialized AI agents to communicate with each other. Agents have distinct skills: some query scientific literature, others retrieve chemical and molecular data, perform similarity searches, and synthesize evidence from multiple sources.

AiChemy targets early-stage research, where identifying biological mechanisms and evaluating candidate molecules is time-consuming and expensive. By automating these tasks, pharmaceutical companies can potentially reduce the time from hypothesis to clinical testing.

For technology leaders in the healthcare sector, this is a practical example of vertical agentic AI, where general LLMs are orchestrated by domain-specific tools toward a bounded business objective. The reference architecture is openly available via GitHub.

Launch confirmed by InfoWorld and Databricks engineering blog, April 6, 2026.