How Atlantic Health cut legal document search time by 42% with Amazon Bedrock metadata filtering

This article describes how Atlantic Health implemented metadata-filtered retrieval augmented generation (RAG) to improve legal document search across multiple hospital systems.

• Achieved 42% reduction in query response time and 87% improvement in retrieval accuracy for hospital-specific searches
• Used Amazon Bedrock Agents with metadata filtering on hospital name and effective year to distinguish similar legal terms
• Stored document embeddings and metadata in Amazon OpenSearch Service for efficient vector and structured filtering
• Deployed serverless application on Amazon ECS with AWS Fargate for automatic scaling
• Applied Amazon Bedrock Guardrails for safety, privacy, and responsible AI controls
• Reduced false positives from wrong hospital or outdated documents, eliminating compliance risks

The solution enables legal teams managing multi-entity document repositories to retrieve precise, contextually relevant information while reducing compliance risk and improving user satisfaction to 94%.