An oil services company in Oman approached us post a successful migration of their data to Dynamics 365 and SharePoint. Their challenge? In their retired system, PDF text was easily visible within records, but after the migration, accessing PDF text became cumbersome in the new setup, prompting the need for a quick solution to restore this essential functionality
To handle text recognition for thousands of individual PDF files, we established a primary Power Automate flow that could be manually initiated to locate the PDF files. This primary flow would in turn trigger a child Power Automate flow responsible for running the AI Builder model and updating the Dynamics 365 record with the extracted data. The child flow would be initiated with the ID of the relevant Dataverse record to carry out the necessary actions on that specific record
Anticipating potential errors, we faced uncertainty regarding the length of content within each of the PDFs and whether the allocated 50,000 characters per field would be sufficient for the extracted text. To resolve this uncertainty, we initially ran the flow once, solely for character count purposes. For character count calculations, we used a length expression as outlined below
We utilized the expression length(variables('ExtractedContent')) to ascertain the character count of the content, adding this count to an Excel sheet within the flow to record PDF character counts. Storing the results of all extractions in the Excel sheet, a sample run, taking over 24 hours, revealed the insufficiency of the 50,000-character field for hosting the extracted data. Consequently, we extended the multi-line field to accommodate 150,000 characters prior to the final run, which itself took an additional 24 hours to complete. This extra day of preparation before the final run ensured we could avoid any potential errors
Efficient Text Extraction and Versatile Processing: AI Builder excels in extracting text line by line, offering an automated solution for managing extensive text data across diverse document structures. Its versatility extends to extracting text regardless of the line's position within a paragraph, ensuring adaptability to various document formats.
Seamless Data Integration and Structured Output: AI Builder integrates seamlessly with tools like Power Automate, enhancing the processing and management of extracted text. The tool's ability to structure text with headers and paragraphs not only improves readability but also facilitates user-friendly data incorporation into reports and analyses.
Moreover, AI Builder brings forth a host of benefits, including time savings through automation, increased accuracy in text extraction, scalability for handling large datasets, consistency in extraction and formatting, and overall cost-efficiency by automating tasks that would traditionally demand manual labor. This comprehensive set of advantages positions AI Builder as a robust solution for diverse projects with varying data sizes, ensuring streamlined and reliable text extraction processes.