Oldcastle APG, a leading global manufacturer of architectural products, faced significant challenges with its document processing workflow. The company struggled to efficiently handle hundreds of thousands of proof of delivery (POD) documents, commonly known as ship tickets, each month across over 200 facilities. Their reliance on an outdated optical character recognition (OCR) system proved unreliable, demanding constant maintenance and substantial manual intervention. This created inefficiencies and hindered operational effectiveness; therefore, Oldcastle sought a transformative solution to streamline its processes. Consequently, they partnered with AWS to revolutionize their approach to document processing.
Understanding the Challenges in Document Processing
Oldcastle’s primary concern revolved around automating the handling of high-volume POD documents while minimizing manual effort. Specifically, they needed a solution capable of accurately processing a substantial workload—ranging from 200,000 to 300,000 ship tickets monthly—and adapting to inconsistent inputs like rotated pages and varying document formats. Furthermore, improving data extraction accuracy beyond the existing 30–40% was crucial, alongside adding functionalities such as signature validation. The current system required dispatchers at each of their numerous facilities to dedicate approximately four to five hours daily solely to manually processing ship tickets, highlighting a considerable drain on resources.
The Limitations of Existing OCR Systems
The previous OCR solution presented several limitations that hampered efficiency and accuracy. For instance, its inability to consistently extract data from diverse document formats necessitated frequent manual corrections and interventions. Additionally, the system’s vulnerability to variations in page orientation and layout further reduced its reliability. Consequently, Oldcastle’s IT team spent a significant portion of their time maintaining and troubleshooting the outdated OCR system.
Expanding Beyond Ship Tickets: Invoice Processing
Beyond ship tickets, Oldcastle also encountered similar challenges with supplier invoice processing, requiring matching against purchase orders. These invoices often exhibited varying formats and structures, mirroring the complexities observed in POD documents. As a result, automating this process proved equally vital to achieving overall operational efficiency and reducing errors; therefore, a comprehensive solution needed to address both scenarios.
The AWS Bedrock & Textract Solution for Document Processing
To overcome these hurdles, Oldcastle collaborated with AWS Solutions Architects to develop an innovative workflow leveraging Amazon Bedrock and Amazon Textract. This architecture enabled a significant upgrade over the previous OCR system. The solution employs Amazon Simple Email Service (Amazon SES) for receiving ship tickets directly from drivers in the field via email, which is then triggered by Amazon S3 Event Notifications. A key component of the workflow involves automatic scaling compute jobs to orchestrate document processing.
Workflow Breakdown: From Receipt to Extraction
- Incoming PDF files are submitted to Amazon Textract using the Start Document Analysis API, specifically utilizing layout and signature features for comprehensive data extraction.
- Subsequently, Amazon Textract‘s results are processed by an AWS Lambda microservice that corrects page rotation issues and generates a markdown representation of the text for each page.
- The generated markdown is then passed to Amazon Bedrock for efficient key data extraction. This allows for more nuanced and accurate information retrieval compared to previous methods.
Scalability and Real-Time Visibility
Importantly, this new architecture allows for effortless scalability to accommodate fluctuating volumes of document processing. The event-driven design ensures the system can handle peaks in demand without compromising performance or accuracy. Furthermore, the solution provides real-time visibility into outstanding PODs and deliveries, enabling proactive management and improved operational control. Consequently, Oldcastle’s ability to efficiently manage its document flow has been dramatically enhanced.
Benefits & Future Implications for Document Processing
Oldcastle’s partnership with AWS resulted in a significant transformation of their document processing capabilities. The new solution not only automated the handling of hundreds of thousands of POD documents each month but also drastically improved accuracy and reduced manual effort. Moreover, it provided valuable real-time insights into delivery status. Notably, this approach serves as a practical blueprint for other businesses facing similar challenges with document management or seeking to leverage generative AI for business process optimization. The solution’s modular design allows for easy adaptation and integration with existing systems, making it a versatile tool for achieving operational excellence.
Source: Read the original article here.
Discover more tech insights on ByteTrending.
Discover more from ByteTrending
Subscribe to get the latest posts sent to your email.
