Novo Nordisk deploys an AI agent for manufacturing documentation management on the Amazon Bedrock platform
Novo Nordisk has deployed an AI agent in its tablet manufacturing team on the Amazon Bedrock platform, which uses a GraphRAG architecture (Neptune + OpenSearch) and Claude models to speed up manufacturing documentation management from hours to minutes.
Novo Nordisk, a pharmaceutical manufacturer of drugs for chronic conditions such as diabetes and obesity, has deployed an AI agent for managing manufacturing documentation within its Manufacturing Technology team (tablets area). According to the company, it manages thousands of active controlled documents (standard operating procedures, work instructions, forms, training materials) spread across divisions, departments, and product programs. Existing processes were largely manual, based on spreadsheets and disconnected systems, which led to the risk of outdated documentation, delays in updates, and gaps in regulatory compliance visibility.
The solution is built on a GraphRAG architecture: a knowledge graph in Amazon Neptune captures the structure and relationships between documents, while a hybrid search index in Amazon OpenSearch Serverless provides both semantic and exact search across document content. The reasoning layer runs on Amazon Bedrock AgentCore, the agentic runtime is built on the Strands Agents SDK, and it is powered by Claude models from Anthropic deployed via Amazon Bedrock. Documents pass through an ingestion pipeline using Amazon Bedrock Data Automation for multimodal processing, orchestrated via AWS Step Functions and AWS Lambda, with infrastructure defined using AWS CDK.
According to the company, the system allows subject matter experts (SMEs) to complete work that previously took hours or days within minutes. The assistant also features a mode that actively clarifies ambiguous queries before running a search, so that answers are more precise and defensible in a regulated environment. See the source article for details.
Why it matters
This showcases a concrete production deployment of a GraphRAG architecture in a strictly regulated pharmaceutical environment, where traceability and accuracy of answers are critical. For companies with complex, frequently changing documentation (manufacturing, compliance, quality), it is a documented example of how to shorten the time needed to locate the current procedure and reduce the risk of working from an outdated document.
Two audiences, two different impacts
What this means
For individuals
For developers and architects, this is a concrete production pattern for combining a knowledge graph, hybrid search, and an agentic framework to deliver citation-backed answers in a strictly regulated industry.
For a business
The deployment shortens the time needed to find and update manufacturing documentation from hours or days to minutes, improves visibility into document approval status, and reduces the risk that an outdated procedure remains in circulation in manufacturing.
ProcessesCheck the original
Event sources
only one source so far · 1 publisher, 0 independent. We count feeds from the same owner only once.