70% Reduction in Manual Data Entry for Loan Processing
How we deployed a secure, air-gapped Document Intelligence Agent to extract and verify Nepali citizenship and property documents.
The Challenge
Kathmandu Finance Cooperative was drowning in physical paperwork. Loan officers spent 40% of their day manually typing data from Nepali citizenship cards and Lalpurja (land ownership documents) into their core banking system.
The SUNEB Solution
We deployed a localized version of our Document Intelligence Agent. Crucially, because financial data is highly sensitive, we utilized an air-gapped Gemma model hosted on their internal servers, ensuring compliance with Nepal Rastra Bank guidelines.
The Result
The system now automatically scans, extracts, and validates document data, passing it directly into their ERP. Loan processing time dropped from 4 days to under 24 hours.