Novo Nordisk cuts clinical study report generation time by 90% with generative AI on AWS

Novo Nordisk's traditional method for creating clinical study reports was slow and labor-intensive, requiring 40–50 professionals and up to 15 weeks per report. These reports are mandatory for regulatory submissions and must meet strict standards for accuracy, consistency, and traceability, making automation challenging. As documentation scale and complexity grew, the company needed to rethink its workflows to remain competitive.

Novo Nordisk partnered with AWS, MongoDB, and Anthropic to build a secure, compliant generative AI solution using Anthropic's Claude 3.5 deployed via Amazon Bedrock. MongoDB Atlas provided a flexible data foundation for near real-time collaboration, integrated with AWS Lambda and Amazon S3 for scalable, serverless operations. A user-friendly interface allows medical professionals to submit documents, review AI-generated summaries, and provide validation feedback to continuously improve the system.

The time required to generate critical clinical documentation was reduced by over 90%, compressing a 15-week, 40–50-person process down to minutes with a team of three. This acceleration frees resources and shortens time-to-submission for new therapies, enabling faster delivery of treatments to patients. Novo Nordisk also confirmed zero risk of data leakage within its AWS Cloud environment, satisfying strict regulatory compliance requirements.

• Partnering with external AI and cloud experts was essential to meet both operational and regulatory requirements in a highly regulated industry.
• Keeping humans in the loop for validation and feedback ensures accuracy and accountability in AI-generated regulatory documents.
• Generative AI can compress multi-week, large-team documentation workflows into minutes without sacrificing compliance or quality.