Amazon Global Engineering Services: Automated operational readiness testing with Amazon Bedrock Nova Pro

Use case typeComputer vision inspectionUpdated Feb 10, 2026

Amazon Global Engineering Services (GES) built an Intelligent Operational Readiness (IORA) solution to automate testing for new fulfillment centers. The system uses Amazon Bedrock with Amazon Nova Pro for real-time image-based object detection and Anthropic Claude Sonnet 4.0 via Bedrock to generate standardized UIN descriptions and detection rules. Amazon API Gateway, AWS Lambda, Amazon S3, and Amazon DynamoDB orchestrate and store the workflow, allowing testers to verify installation status, detect defects, and review results with a production UI.

Organization
Amazon Global Engineering Services
Industry
Logistics
Location
United States
Published
February 2026

Reported outcomes

92%

precisionQuality & accuracy

2 secondslatency per image5 secondslatency per image60%total testing time

Strategic outcomes

Cost efficiencyReduced manual verification effortScale & capacityEnabled focus on missing componentsOther strategic outcomeImproved ground-truth data quality

Catalog median for quality & accuracy deployments: +90% across 281 reported metrics. Compare benchmarks →

Primary read

Use case focus

Showing 2 of 2

  • 1Computer vision inspection
  • 2Operations optimization
  • Operational readiness testing for new fulfillment centers required about 2,000 hours of manual effort per facility.
  • Teams had to verify more than 200,000 components across 10,500 workstations.
  • Manual review made it difficult to focus on missing components and scale testing efficiently.
  • Amazon GES developed an AI-powered Intelligent Operational Readiness (IORA) pipeline.
  • Amazon Nova Pro performs real-time image detection with bounding boxes and installation-status verification.
  • Anthropic Claude Sonnet 4.0 generates detailed component descriptions from reference images and helps create standardized detection parameters and false-positive rules.
  • Amazon API Gateway routes requests to AWS Lambda functions, which process images and call Amazon Bedrock.
  • Module images, reference images, results, and structured verification data are stored in Amazon S3 and Amazon DynamoDB with encryption enabled.
Technologies
Amazon BedrockAmazon Nova ProAnthropic Claude Sonnet 4.0Amazon API GatewayAWS LambdaAmazon S3Amazon DynamoDBAWS Key Management ServiceAWS Identity and Access Management
  • The prototype achieved 92% precision on representative test modules.
  • Latency was 2 to 5 seconds per image.
  • The solution reduced total testing time by 60% versus manual operational readiness testing.
  • Field teams could focus only on missing components, with the article noting that 40% coverage translates to 40% time reduction.
Architecture

A serverless orchestration flow uses Amazon API Gateway to invoke AWS Lambda functions. Lambda calls Amazon Bedrock for two paths: batch generation of standardized UIN descriptions and detection rules using Anthropic Claude Sonnet 4.0, and real-time image analysis using Amazon Nova Pro to detect UINs, bounding boxes, installation status, defects, and confidence scores. Images and structured results are stored in Amazon S3 and Amazon DynamoDB, protected with AWS KMS and IAM-based access control.

Sources & evidence1
Groundedness: 5/5Type: Blog PostPublished: Feb 10, 2026Publisher: AWSEvidence: VendorConfidence: 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
More from Amazon Global Engineering Services
Provider evidence
AWS provider pageAWS ranked cases
Market context
Logistics AI use casesLogistics industry insightsUnited States country insights
AI patterns
Vision
Browse the catalog
All AI use casesAI use-case typesComputer vision inspection use cases

Was this useful?