AI/MLHealthcareFeatured

Healthcare ERP ETL AI Agent

A regional hospital network operating Epic EHR alongside an Oracle ERP struggled with manual, error-prone ETL between clinical and financial systems. Claims reconciliation, supply-chain accruals, and ...

9 months
Duration
7 people
Team Size
$220,000 - $280,000
Budget
AI/ML & Data
Service
Healthcare ERP ETL AI Agent
5
Key Results

Project Overview

Project Information

IndustryHealthcare
ServiceAI/ML & Data
Duration9 months
Team Size7 people
Budget Range$220,000 - $280,000

Technologies Used

Amazon BedrockAWS GlueAmazon OpenSearch ServerlessAWS LambdaAWS Step FunctionsAmazon TextractAmazon Comprehend MedicalAmazon SageMakerAmazon S3Amazon Aurora

Our team leveraged modern technologies and best practices to deliver a robust, scalable solution that meets enterprise requirements.

The Challenge

Every great solution starts with understanding the problem. Here's what they were facing.

A regional hospital network operating Epic EHR alongside an Oracle ERP struggled with manual, error-prone ETL between clinical and financial systems. Claims reconciliation, supply-chain accruals, and cross-system reporting required a team of analysts to hand-key and reconcile records, while unstructured documents (EOBs, claim denial letters, purchase orders, contracts) sat outside any structured pipeline. Reconciliation cycles ran 9-11 business days and claim denials from miscoding exceeded 12%.

Our Solution

We developed a comprehensive solution that addressed their challenges and delivered measurable results.

We delivered an autonomous AI agent that orchestrates end-to-end ETL between the EHR and ERP. AWS Step Functions coordinate the workflow: Amazon Textract ingests EOBs, denial letters, and POs; Amazon Comprehend Medical extracts clinical entities and de-identifies PHI; AWS Glue performs schema transformation and loads into Amazon Aurora. The agent uses Retrieval-Augmented Generation (RAG) over an Amazon OpenSearch Serverless vector store indexed with payer contracts, billing policies, and HCPCS/CPT coding rules; Amazon Bedrock (Claude) grounds its classification and coding decisions in that knowledge base, generating reconciling journal entries and denial-appeal narratives. AWS Lambda handles event-driven triggers; Amazon SageMaker hosts a fine-tuned coding classifier. The whole pipeline runs under a HIPAA BAA with end-to-end KMS encryption and CloudWatch audit trails.

Results & Impact

The measurable outcomes that were achieved through our partnership.

Reduced EHR-to-ERP reconciliation cycle from 9 days to under 8 hours

Cut claim denial rate from 12% to 3.4% via RAG-grounded coding

Automated ingestion of 40,000+ unstructured documents per month

Saved finance team ~1,200 staff hours per month

Achieved 99.2% straight-through processing on supply-chain accruals

9 months
Project Duration
7 people
Team Members
5
Key Outcomes

Technology Stack

The tools and technologies we used to build this solution.

Other

Amazon BedrockAWS GlueAmazon OpenSearch ServerlessAWS LambdaAWS Step FunctionsAmazon TextractAmazon Comprehend MedicalAmazon SageMakerAmazon S3Amazon Aurora

Each technology was carefully selected to ensure scalability, maintainability, and optimal performance for the project requirements.

"The AI agent fundamentally changed how finance and revenue cycle interact with clinical data. RAG over our own payer contracts means the coding decisions are explainable and auditable, which is non-negotiable in healthcare."
Dr. Anita Rao
VP of Revenue Cycle
Regional Hospital Network

Ready to Start Your Project?

Inspired by this case study? Let's discuss how we can apply similar strategies and technologies to solve your unique challenges.

Quick Start

MVP development in 8-12 weeks

Get Started

Free Consultation

Discuss your project requirements

Schedule Call

View More Work

Explore our full portfolio

See Portfolio

Want to Discuss a Similar Project?

Every project is unique, but we apply proven methodologies and cutting-edge technologies to deliver exceptional results.