Pharmaceutical Leader Scales to 2,500+ AI Chatbots with Amazon Bedrock Self-Service Platform

Novo Nordisk, a multinational pharmaceutical company founded in 1923, develops therapies for chronic diseases including diabetes, obesity, and rare blood disorders. While the organization had experimented with generative AI and large language models starting in 2023, employees faced significant barriers to innovation. Many colleagues possessed viable ideas for using AI to solve business problems in nonregulated processes, but lacked the ability to develop applications or maintain supporting infrastructure. The company struggled to predict which use cases would deliver value, and needed a way to enable secure, compliant, and cost-effective experimentation across its workforce. In partnership with Cloud2 Oy, an AWS Partner, Novo Nordisk designed a solution to address these constraints.

The company built a self-service generative AI platform on Amazon Web Services (AWS) leveraging Amazon Bedrock, a fully managed service providing access to high-performing foundation models. The platform enables employees to create and customize chatbots and AI agents for specific use cases without requiring extensive infrastructure management. Built on serverless architecture using Amazon DynamoDB for database management and AWS Lambda for compute, the solution minimizes operational overhead and scales automatically based on demand. Employees can develop chatbots from scratch in days rather than months, test multiple foundation models as they are released, and share applications across teams for iterative improvement. Use cases span document drafting with automated error correction, information retrieval from source documents, and virtual collaboration agents that enhance team decision-making and reduce knowledge silos.

Since launch, the platform has driven significant adoption and measurable impact. Over 25,000 employees have created chatbots for 2,500+ unique use cases, with the general-purpose chatbot alone processing 26,000+ prompts monthly. The solution maintains cost efficiency at approximately $10 per chatbot per month through serverless infrastructure, while accelerating innovation cycles from several months to just 2 days, or 1 hour for proof-of-concept development. Users report completing work in 30 minutes that previously required a full day, and the platform has enabled employees to shift focus from routine tasks to higher-value work. Novo Nordisk views this implementation as a catalyst for enhancing employee productivity and job satisfaction rather than replacing roles, with plans to expand capabilities through graph database technology to build comprehensive knowledge graphs and deliver additional patient value.