Ricoh USA partnered with AWS to build a scalable, multi-tenant intelligent document processing (IDP) platform for healthcare clients using the AWS GenAI IDP Accelerator. The solution combines Amazon Textract for OCR with Amazon Bedrock foundation models for document classification and data extraction, deployed on a serverless architecture using Lambda, SQS, S3, and DynamoDB. Key outcomes include reducing customer onboarding time from 4–6 weeks to 2–3 days, cutting engineering hours per deployment by over 90%, and scaling monthly throughput from ~10,000 to a projected 70,000+ documents. The architecture supports HIPAA, HITRUST, and SOC 2 compliance through encryption, least-privilege IAM, and comprehensive audit logging. A configurable AWS SAM/CloudFormation deployment model enables rapid onboarding without code changes, while confidence scoring drives a human-in-the-loop review process achieving 98–99% accuracy.
Table of contents
Customer overviewChallenges with complex document processing workflowsSolution overviewArchitecture detailsResults and outcomesBest practices and lessons learnedGetting startedConclusionSort: