AstraZeneca Accelerates Drug Development with Amazon Bedrock Agents

AstraZeneca developed Development Assistant, an AI tool using Amazon Bedrock Agents with multi-agent architecture for fast natural language querying of clinical, regulatory, safety, and quality data. The solution unifies structured and unstructured data sources to provide transparent, actionable insights supporting faster decision-making in drug development. Multi-agent AI architecture routes queries to specialized agents for context-aware, high-performance responses across clinical trial and R&D domains. Development Assistant reduced insight generation time from hours to minutes and scaled to 1,000+ users, breaking down domain silos across pharmaceutical R&D. The tool provides transparent data source referencing and is progressing toward expansion across broader R&D functions to accelerate medicine development pipeline.

Organization
AstraZeneca
Industry
Pharma
Published
April 2026

Reported outcomes

Strategic outcomes

Speed & agilityAccelerated insight generationScale & capacityScaled to over 1,000 R&D usersCustomer experience & trustImproved trust in AI insightsNew product / capabilityBuilt multi-agent R&D querying tool

Primary read

Use case focus

Showing 3 of 4

  • 1Multi-Agent AI
  • 2Natural Language Querying
  • 3Clinical Trial Insights
  • Drug development is complex with disparate data systems and slow insight generation limiting timely clinical trial and regulatory decisions.
  • Manual data compilation across multiple domains caused bottlenecks and inefficiencies in pharmaceutical R&D teams.
  • Built a multi-agent AI solution using Amazon Bedrock Agents for natural language querying across structured and unstructured datasets.
  • Implemented a supervisor agent directing queries to specialized subagents optimized for clinical, regulatory, quality, and terminology data.
  • Combined text-to-SQL generation with retrieval-augmented generation to unify and query heterogeneous data sources rapidly.
  • Established controlled vocabularies and transparency features to ensure trust and interpretability of AI-generated insights.
  • Cut time to generate insights from hours to minutes, enabling faster, evidence-based decisions in drug development.
  • Scaled AI platform usage to over 1,000 R&D users, promoting cross-domain collaboration and breaking data silos.
  • Helped accelerate clinical trial pipelines and medicine development with actionable, trustworthy AI insights.
  • The modular multi-agent architecture increases flexibility and performance of AI applications in pharmaceutical R&D.
Architecture

The architecture employs a multi-agent AI model on Amazon Bedrock Agents, with a supervisor agent routing queries to specialized agents handling clinical, regulatory, safety, quality, and terminology data. It integrates text-to-SQL generation and retrieval-augmented generation to query unified structured and unstructured data. Data sources are harmonized via a Drug Development Data Platform providing standardized, interoperable datasets for AI consumption.

Sources & evidence1
Groundedness: 5/5

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases
This website uses cookies to enhance the user experience. Learn more.