As one of the largest mortgage lenders and servicers in the U.S., Pennymac manages millions of documents every month—yet manual workflows and inconsistent accuracy were holding back operational scale.

Working with AWS, Pennymac modernized its document-processing pipeline using Amazon Textract and Amazon Bedrock to automate data extraction and classification across vast volumes of mortgage files. The enhanced system now delivers over 99% extraction accuracy, enabling teams to process documents with greater reliability, reduced manual effort, and lower operational risk.

This approach is helping Pennymac run document-processing workloads 24/7, onboard more customers, and keep costs under control—while laying the foundation for future, ML-powered automation.

Key points covered:

• How Pennymac automated millions of mortgage-related documents

• The shift from 70–80% accuracy to over 99% using AWS ML services

• A scalable, 24/7 document-processing workflow

• The operational impact: fewer errors, faster turnaround, and lower costs