—
AI-Powered OCR Extraction from Inbound Emails
From inbox to intelligent automation: how OCR and AI cut 30 hours of manual work per week
For businesses drowning in documents, the inbox can quickly become a bottleneck. An organisation found itself processing over a thousand PDF attachments via email each month, including delivery confirmations, invoices, and other essential documents. The process was entirely manual. Every file had to be opened, saved, and reviewed by staff who then copied key data points into Excel. The result? Nearly 30 hours a week spent on repetitive admin, high error rates, and mounting reconciliation delays.
The Challenge: A Costly, Manual Workflow Holding Back Productivity
With staff spending the equivalent of almost four workdays per week just to process documents, the business faced a significant drag on efficiency. Manual handling meant errors were frequent, data was often incomplete, and there was little visibility into what had or hadn’t been processed. Reconciliation cycles were slow, and the risk of audit issues grew alongside volumes.
The Solution: A Fully Automated OCR Pipeline with Built-In Intelligence
To solve the issue, I designed and implemented a scalable, end-to-end automation solution using a mix of AI, scripting, and visual analytics.
Key elements included:
Email Automation via IMAP: The solution connected directly to the company’s inbox to fetch relevant emails and attachments automatically.
OCR Extraction with AWS Textract: PDF documents were processed using Textract to extract structured fields such as delivery dates, item references, and quantities.
Python Scripting for Workflow Orchestration: Extracted data was parsed, cleaned, validated, and written to a structured output, ready for downstream use.
Exception Tracking with Power BI: A dashboard provided live tracking of document processing status, flagged missing fields, and enabled staff to resolve issues quickly without digging through inboxes.
The Result: Faster Processing, Greater Accuracy, Smarter Oversight
The automation pipeline was embedded into daily operations and adopted as the standard approach for document handling. It enabled the business to:
Cut reconciliation time from more than 8 hours to under 45 minutes per week
Improve data accuracy and reduce human error across all processed documents
Strengthen audit compliance by creating a transparent trail of processed and flagged items
Free staff from low-value admin so they could focus on higher-impact client service tasks
Establish a foundation for broader workflow automation in finance and operations
Skills & Tools Applied
AWS Textract for document OCR
Python (IMAP, PDF parsing, automation logic)
Power BI dashboarding for exception handling and reporting
Data validation and cleaning
Workflow automation and operational efficiency
Conclusion
This project shows how applying intelligent automation to everyday pain points such as document processing can unlock major time savings, improve data reliability, and set the stage for deeper digital transformation. When you give time back to people, you give focus back to what matters.