From Manual Transcription to Automated Data Extraction: Transforming Client Onboarding for a Personal Loan Provider
Leveraging AI, we automated the onboarding process, reducing document processing time by 70-80%.
In an effort to streamline their client onboarding process, Gran Cooperativa faced inefficiencies due to manual transcription of documents. By implementing a hybrid OCR solution utilizing AI, the financial institution significantly reduced onboarding times and improved data accuracy.
- React Router v7
- TypeScript
- Prisma
- PostgreSQL
- BullMQ
- Redis
- AWS S3
- Tesseract OCR
- OpenAI GPT-4o
- Sharp
- pdf2pic
- pdf-parse
- Docker
The Situation
Manual transcription of multiple documents required for loan applications.
High error rates in data transcription leading to operational risks.
Limited processing capacity due to reliance on manual operators.
What We Worked On
Initial Manual Process
Operators manually transcribed critical client documents into the system, which led to significant processing delays.
Implementation of Hybrid OCR
Developed a hybrid OCR pipeline that utilized Tesseract for structured documents and GPT-4o Vision for complex document types.
Data Processing Automation
Documents were stored in AWS S3, processed asynchronously to extract structured data and reduce manual input.
Error Reduction and Efficiency Gains
Automated data extraction significantly reduced typing errors and processing times, streamlining the onboarding workflow.
Outcomes
What Changed
Significantly decreased time required for client onboarding.
Enhanced accuracy of data entry reducing operational risks.
Increased processing capacity allowing for higher volume of client applications.
Worth Noting
The automated onboarding solution is currently live and continues to improve the efficiency of Gran Cooperativa's operations, providing a robust framework for future enhancements.
How can we help you revolutionize your onboarding process?
We find the leaks. Then we fix them.