Oldcastle's AI-Driven Document Processing Revolution with AWS
Discover how Oldcastle transformed its inefficient document processing with AWS and Amazon Bedrock, achieving 90% accuracy and significant cost savings. Lear...
Key Takeaways
- Oldcastle achieved 90% accuracy in document processing using AWS Bedrock and Textract.
- The solution reduced manual labor by 80%, freeing up valuable human resources.
- Scalability from thousands to hundreds of thousands of documents per month without performance issues.
- Cost-effective processing at less than $0.04 per page.
Oldcastle's AI-Driven Document Processing Revolution with AWS
Oldcastle, a global leader in architectural products, faced significant challenges with its document processing workflow, particularly with proof of delivery (POD) documents. The company was processing 100,000–300,000 ship tickets per month across over 200 facilities, but their existing optical character recognition (OCR) system was unreliable, requiring constant maintenance and manual intervention. This bottleneck was not only time-consuming but also error-prone, leading to significant inefficiencies. By partnering with AWS, Oldcastle implemented a transformative solution using Amazon Bedrock and Amazon Textract, achieving remarkable results.
The Challenges of Manual Document Processing
The primary challenges Oldcastle faced included:
- High Volume of Documents: Processing 100,000–300,000 ship tickets monthly.
- Inconsistent Inputs: Dealing with rotated pages and variable formatting.
- Low Accuracy: The existing OCR system could only accurately read 30–40% of documents.
- Manual Intervention: Dispatchers spent 4–5 hours daily on manual processing.
- IT Maintenance: Constant development and maintenance efforts were required.
Solution Overview
AWS Solutions Architects collaborated with Oldcastle engineers to build a robust, scalable solution. The architecture leverages several AWS services to create an end-to-end document processing workflow:
- Email Ingestion: Amazon Simple Email Service (SES) receives ship tickets directly from drivers in the field.
- Event-Based Processing: Amazon S3 Event Notifications trigger the processing pipeline.
- Document Analysis: Amazon Textract uses the Start Document Analysis API with Layout and Signature features to handle PDF files.
- Data Extraction: AWS Lambda microservices resolve rotation issues and generate markdown representations of the text.
- AI-Driven Extraction: Amazon Bedrock efficiently extracts key values from the markdown text.
- Data Storage: The results are saved to an Amazon Relational Database Service (RDS) for PostgreSQL database.
The Architecture in Detail
The following diagram illustrates the solution architecture:
This architecture ensures that the system can handle large PDF files at scale, with the output of Amazon Textract containing the necessary geometries to fix layout issues and generate quality markdown. Amazon Bedrock's cost-effectiveness and ability to process formatted shipping tickets with consistent fields were crucial in this solution.
Results and Benefits
The implementation of this architecture brought numerous benefits to Oldcastle:
Business Process Improvement
- Automated Document Processing**: Reduced the need for manual intervention.
- Enhanced Accuracy**: Improved data extraction accuracy to over 90%.
- Signature Validation**: Added the capability to validate signatures and reject incomplete documents.
- Real-Time Visibility**: Provided real-time insights into outstanding PODs and deliveries.
Productivity Gains
- Fewer Manual Hours**: Significantly reduced the time spent on manual data entry and document processing.
- Value-Added Activities**: Freed up staff for more strategic tasks.
- Reduced IT Efforts**: Decreased the burden on the IT team for development and maintenance.
Scalability and Performance
- Seamless Scaling**: Handled an increase from a few thousand to 200,000–300,000 documents per month without performance issues.
- Robust Performance**: Maintained high performance even with increased volume.
User Satisfaction
- High User Confidence**: Increased trust in the system's accuracy and reliability.
- Positive Feedback**: Business users appreciated the ease of use and effectiveness.
Cost-Effectiveness
- Affordable Processing**: Achieved processing costs of less than $0.04 per page.
The Bottom Line
Oldcastle's successful implementation of AI-driven document processing with AWS demonstrates the potential for transformative change in the industry. By leveraging advanced technologies like Amazon Bedrock and Textract, companies can achieve significant improvements in efficiency, accuracy, and cost-effectiveness. This solution not only addresses current challenges but also positions Oldcastle for future expansion into other use cases such as AP invoice processing, W9 form validation, and automated document approval workflows.
Frequently Asked Questions
What technologies did Oldcastle use for document processing?
Oldcastle used AWS services including Amazon Bedrock, Amazon Textract, Amazon SES, Amazon S3, AWS Lambda, and Amazon RDS for PostgreSQL to build their document processing workflow.
How did AWS help Oldcastle improve document processing accuracy?
AWS's Amazon Textract and Amazon Bedrock provided advanced OCR and AI capabilities, improving data extraction accuracy from 30-40% to over 90%.
What was the impact on Oldcastle's manual labor?
The automated solution reduced manual labor by 80%, allowing staff to focus on more value-added activities.
Can this solution scale to handle more documents?
Yes, the solution seamlessly scaled from processing a few thousand documents to 200,000-300,000 documents per month without performance issues.
What is the cost of processing documents with this solution?
Oldcastle can process documents at less than $0.04 per page, making it a cost-effective solution.
