Transforming Document Processing: How Oldcastle APG Enhanced Workflow with AWS Solutions

Introduction: A Partnership for Innovation

This post was written in collaboration with Avdhesh Paliwal of Oldcastle APG. Oldcastle APG, a leader in the architectural products industry, faced significant challenges in managing proof of delivery (POD) documents, navigating inefficiencies in their existing processes. This article delves into their journey to optimize document handling through advanced technologies offered by AWS.

Challenges with Document Processing

Oldcastle’s primary challenge was streamlining the processing of high volumes of ship tickets with minimal human intervention while addressing issues such as inconsistent inputs and low accuracy rates.

Solution Overview

A collaborative effort with AWS Solutions Architects resulted in a robust, scalable solution utilizing Amazon Textract and Amazon Bedrock, enhancing both the accuracy and efficiency of document processing.

Results of Implementation

The AWS-driven architecture yielded significant improvements in business processes, productivity, and user satisfaction, all while maintaining cost-effectiveness.

Conclusion: A Strategic Move Towards Efficiency

With a successful implementation in hand, Oldcastle is poised to explore additional use cases, leveraging AI for enhanced operational efficiency.

About the Authors

Learn more about the dedicated professionals who contributed to this transformative project.

Transforming Document Processing at Oldcastle APG with AWS: A Case Study

This post was written in collaboration with Avdhesh Paliwal of Oldcastle APG.

In today’s fast-paced business environment, efficiency is crucial. Oldcastle APG, a leading player in the architectural products industry, faced significant challenges in their document processing workflows. With 100,000–300,000 ship tickets to manage each month across over 200 facilities, their manual processes quickly became unsustainable. An outdated and unreliable optical character recognition (OCR) system could only read 30-40% of documents accurately, requiring constant human intervention and maintenance.

This post explores how Oldcastle partnered with AWS to revolutionize their document processing using Amazon Bedrock and Amazon Textract, moving toward a more automated, efficient, and scalable solution.

Challenges with Document Processing

Oldcastle’s primary challenge was clear: they needed a way to:

• Accurately process a vast volume of ship tickets with minimal human effort.
• Scale efficiently to handle between 200,000–300,000 documents each month.
• Adapt to inconsistent inputs such as rotated pages and diverse formatting.
• Significantly improve data extraction accuracy, moving beyond the existing 30-40% accuracy rate.
• Incorporate new features like signature validation for proof of delivery (POD) documents.
• Achieve real-time visibility into outstanding PODs and deliveries.

The company also faced challenges with processing supplier invoices and matching them to purchase orders, further complicating their document workflow. Dispatchers were spending 4-5 hours daily on manual processing, leading to inefficiencies and potential errors. The IT team was overwhelmed with the upkeep of the unreliable OCR system.

Solution Overview

To address these challenges, AWS Solutions Architects collaborated with Oldcastle engineers to devise an innovative solution. The new end-to-end workflow harnesses several AWS services, including:

1. Amazon Simple Email Service (SES): Used to receive ship tickets directly from drivers.
2. Event-based architecture with Amazon S3 Event Notifications: This allows for the efficient processing of emails at scale.
3. Amazon Textract: Documents were sent to Textract for analysis using the Start Document Analysis API, leveraging its layout and signature features.
4. AWS Lambda microservices: Process the results to adjust for rotation and generate markdown representations of the extracted text.
5. Amazon Bedrock: Utilized to efficiently extract key values from markdown text.
6. Amazon Relational Database Service (RDS): The solution stores results in a PostgreSQL database for easy access and reporting.

This architecture effectively addresses the requirements of accurately processing large PDF files while minimizing human intervention.

Results

The results of implementing this new AWS architecture were transformative for Oldcastle:

Business Process Improvements

• Reduced Manual Processing: The automation of ship ticket processing eliminated the need for manual intervention at each facility.
• Higher Accuracy and Reliability: Data extraction accuracy significantly improved.
• Document Validation: The system could validate signatures and reject incomplete documents.
• Real-time Visibility: Enhanced tracking of outstanding PODs and deliveries.

Productivity Gains

• Savings on Human Hours: There was a sharp decline in the amount of time staff spent on manual data entry.
• Value-added Activities: With less time spent on processing, employees could focus on higher-value tasks.
• Reduced IT Burden: The solution lessened the ongoing maintenance and development responsibilities of the IT team.

Scalability and Performance

• Seamless Scaling: Oldcastle was able to scale their processing capabilities from a few thousand documents to 200,000–300,000 per month without performance issues.

User Satisfaction

• Increased Confidence: Users reported greater confidence in the accuracy and reliability of the new system.
• Positive User Feedback: Business users commended the system’s ease of use and effectiveness.

Cost-effectiveness

• The cost for processing documents dropped to less than $0.04 per page, thus enhancing overall operational efficiency.

Conclusion

The successful implementation of this AWS-based solution has positioned Oldcastle for significant growth and operational efficiency. The company is already considering expanding its capabilities to other areas, such as accounts payable invoice processing and automated document approval workflows.

We encourage businesses facing similar document processing challenges to consider how intelligent document processing could revolutionize their workflows.

For Further Exploration:

Feel free to explore additional resources to understand better how AWS can help optimize document processing for your organization.

About the Authors

Erik Cordsen is a Solutions Architect at AWS based in Georgia, dedicated to applying cloud technologies and machine learning to solve real-world problems.

Sourabh Jain is a Senior Solutions Architect with over eight years of experience, specializing in architecting cloud solutions for better business outcomes.

Avdhesh Paliwal is an accomplished Application Architect at Oldcastle APG, bringing 29 years of expertise in various ERP modules to the table. His work focuses on creating enterprise solutions that drive operational efficiency and business value.

This case study highlights the potential of leveraging AWS services to transform document processing workflows, offering insights valuable to companies aiming to streamline their operations.