Genentech leverages generative AI with Amazon Bedrock Agents to accelerate drug discovery
Genentech uses Amazon Bedrock Agents with Anthropic Claude Sonnet 3.5 to automate complex biomedical research workflows for biomarker validation. The gRED Research Agent processes and synthesizes information from millions of scientific data sources using multi-agent collaboration and Retrieval Augmented Generation (RAG). This automation reduces manual research time from weeks to minutes, freeing scientists to focus on high-impact tasks and accelerating drug discovery.
- Organization
- Genentech
- Industry
- Healthcare
- Location
- United States
- Published
- April 2026
Reported outcomes
43 hours
timeTime & speed
Strategic outcomes
Primary read
Use case focus
Showing 3 of 3
- 1Generative AI
- 2Autonomous AI Agents
- 3Drug Discovery Automation
- Manually sifting through millions of biomedical data sources for biomarker validation, a process that takes years and is highly complex.
- Scientists faced difficulty validating biomarkers across vast scattered datasets requiring long, manual searches.
- Developed gRED Research Agent on AWS Amazon Bedrock, utilizing autonomous generative AI agents to break down complex queries and access multiple scientific data sources simultaneously.
- Implemented multi-agent collaboration leveraging Retrieval Augmented Generation (RAG), internal APIs, and scientific databases for comprehensive, cited synthesis of scientific data.
- Automated over 43,000 hours of biomarker validation, reducing data analysis time from weeks to minutes.
- Accelerated time-to-target identification and enabled faster delivery of lifesaving medicines to patients.
- Enhanced speed and precision of drug discovery research with autonomous AI agents collaborating across specialized knowledge domains.
Architecture
The architecture involves the gRED Research Agent built on Amazon Bedrock Agents platform using Anthropic Claude Sonnet 3.5, integrating Retrieval Augmented Generation (RAG) and multiple data sources including PubMed, Human Protein Atlas, and internal repositories. Multi-agent collaboration enables specialized sub-agents to query distinct data domains and synthesize findings with traceable citations.
Sources & evidence1
AI-generated summary. Verify important details with the linked sources before relying on this case.