Pharmaceuticals and biotech / Pharmaceuticals
Novo Nordisk: Retrieving manufacturing knowledge by effective document version and applicable scope
- Company
- Novo Nordisk
- Country
- Denmark
- Adoption stage
- Limited operation
- Source published
- Date basis
- The date the source was published. It can differ from the date adoption started.
- How the source was checked
- Read the full source text
The work problem
Novo Nordisk's Tablet Manufacturing Technology team needed approved, effective versions matching the site, product and equipment. Keyword searches and fragmented tracking obscured document relationships and change impacts.
Technology and data
The team put an AWS GraphRAG assistant into use. It links a Neptune relationship graph with OpenSearch semantic and keyword retrieval; Bedrock AgentCore agents compose answers. Ambiguous scope triggers clarification first. Answers default to the latest applicable approved, effective documents and flag missing or ambiguous content for expert review.
Results
Novo Nordisk and AWS described expert search, document updates and adaptation dropping from hours or days to minutes, depending on the task. The graph traces references and outdated links. Full-text indexing is incremental by onboarded domain, not complete across the entire vault.
Limits and open questions
The start month, user count and task-level measurement samples were not disclosed. The Tablet team's use does not establish company-wide automatic approval or complete full-text indexing.
Sources
- Novo Nordisk modernizes manufacturing process documents with Amazon Bedrock AgentCoreAWS / Novo Nordisk, Accessed
Compiled from public sources. These are not results from ATF Works customers.