What challenges did Novo Nordisk face before AI implementation?

Manual, disconnected ML-development steps were difficult to scale. Quality-assurance tasks such as cartridge counting and agar plate analysis were time- and resource-intensive. Novo Nordisk needed traceability and monitoring for regulated manufacturing workflows.

What was the impact of the AI solution?

Reduced manual labor in manufacturing quality workflows. Improved scalability and traceability of ML models in production. Shortened time to market for quality-assurance use cases by automating deployment and monitoring.

Source

Novo Nordisk Uses ML for Computer Vision to Optimize Pharmaceutical Manufacturing on AWS

Use case typeIndustrial inspectionUpdated Jun 13, 2026

Novo Nordisk A/S uses computer vision and machine learning on AWS to automate manufacturing quality tasks such as cartridge counting and anomaly detection for agar plates. The company built a prototyping solution to train, deploy, monitor, and manage ML models for edge devices and to support regulated pharmaceutical operations.

Organization: Novo Nordisk
Industry: Pharma
Location: Denmark
Published: May 2026

Reported outcomes

Strategic outcomes

Scale & capacityImproved scalability of ML production workflowsRisk & complianceImproved traceability in regulated manufacturingSpeed & agilityShortened time to market for QA use casesCost efficiencyReduced manual labor in quality workflows

Primary read

Use case focus

Showing 3 of 4

1Computer Vision
2Machine Learning Operations
3Quality Inspection

Manual, disconnected ML-development steps were difficult to scale.
Quality-assurance tasks such as cartridge counting and agar plate analysis were time- and resource-intensive.
Novo Nordisk needed traceability and monitoring for regulated manufacturing workflows.

Built an automated ML pipeline on AWS using Amazon SageMaker Pipelines and Amazon S3.
Packaged and deployed models with Amazon SageMaker Edge and Amazon SageMaker Edge Manager.
Used AWS IoT Greengrass for edge deployment and Amazon QuickSight for monitoring and anomaly review.

Technologies

Amazon SageMaker Pipelines Amazon S3 Amazon SageMaker Edge Amazon SageMaker Edge Manager AWS IoT Greengrass Amazon QuickSight

Reduced manual labor in manufacturing quality workflows.
Improved scalability and traceability of ML models in production.
Shortened time to market for quality-assurance use cases by automating deployment and monitoring.

Architecture

Novo Nordisk built an automated ML pipeline on AWS to process images, train and tune models, evaluate results, register models, compile them for edge deployment, and monitor production inference. The workflow uses Amazon SageMaker Pipelines, Amazon S3, Amazon SageMaker Edge, Amazon SageMaker Edge Manager, AWS IoT Greengrass, and Amazon QuickSight.

Sources & evidence1

Groundedness: 5/5Type: Customer StoryPublished: May 27, 2026Publisher: AWSEvidence: PrimaryConfidence: High

Primary source

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

Explore related AI use cases

Customer evidence