CapiFin Technologies

CapiFin was processing 3,200 personal loan applications per month entirely manually — document collection, CIBIL pulls, eligibility calculation, and LOS data entry consuming 14 FTE-days per month. Month-end backlogs regularly hit 400+ applications, pushing approval TAT to 5–7 days and triggering borrower drop-off.

• 01Deployed an unattended UiPath bot fleet on AWS EC2 with Terraform-managed auto-scaling — 2 instances off-peak, scaling to 8 during month-end surge
• 02Integrated AWS Textract for OCR extraction of PAN, Aadhaar, bank statements, and salary slips from the borrower portal uploads
• 03Connected to CIBIL API for automated credit score pulls, DTI calculation, and eligibility scoring with results written directly to the LOS
• 04Built a CloudWatch monitoring stack with PagerDuty alerts for bot failure, queue depth, and processing rate — ops team knows before borrowers do
• 05Implemented CI/CD pipeline for bot releases — updates deploy in under 12 minutes with zero downtime using blue/green PM2 reload

• 75% reduction in loan processing TAT — from 5–7 days to under 6 hours for standard applications
• 73% reduction in processing cost — 14 FTE-days per month replaced by bot infrastructure running at a fraction of the cost
• 11% exception rate routed to human credit officers — 89% of applications handled end-to-end without manual intervention
• Payback period: 5.5 months. ROI in year one: 218%
• Zero production bot failures in 6 months post-launch — attributed to DevOps-grade infrastructure and monitoring from day one