Summary
A leading financial institution wanted to streamline credit card application processing, which relied heavily on manual effort and carried both error and compliance risk.
We implemented an end-to-end RPA solution on UiPath covering document extraction, validation, mainframe entry and memo generation. The result was faster turnaround, higher accuracy and the ability to process multiple applications concurrently.
The client
The bank handles thousands of credit card applications monthly, involving data capture from various formats, validation against internal policies, and processing through legacy mainframe systems. Credit Card Operations, Compliance and the IT & Automation CoE all had a stake in the process.
The problem
Manual processing was creating delay, error and compliance exposure at the same time.
- 01Manual data entry from scanned application forms
- 02Validation against internal policy, applied inconsistently
- 03Processing through a legacy mainframe with no API
- 04Memo preparation for successful and exception cases
What it was costing
- Manual entry from scanned forms causing delays and errors
- High processing time per application, creating backlogs at peak
- Frequent compliance issues from inconsistent validation
- No way to scale operations during demand spikes
Objectives
Goals
- Reduce manual effort and processing time per application
- Improve data accuracy through automated extraction and validation
- Ensure compliance through standardised rule-based validation
- Enable scalability for high application volumes
Success criteria
- 60–70% reduction in processing time per application
- 95% accuracy in data entry
- 100% compliance with predefined validation rules
- Improved throughput — applications processed per day
What we built
Intelligent extraction feeding rule-based validation, with nothing written to the mainframe until it passes. Queues and parallel robots handle the volume.
Intelligent document extraction
UiPath Intelligent OCR reads scanned PDFs and forms across varying layouts.
Rule-based validation
Extracted data is validated via VBA macros against internal policy before any commit.
Mainframe integration
Validated data is entered into the legacy mainframe directly — no API required.
Automatic memo generation
Memos generated for both successful applications and exception cases.
Exception handling with audit trails
Failures route to a defined workflow, with a full trail behind each decision.
Delivery
4 months — requirements to post-launch support
- Month 1 — requirement gathering and process mapping
- Month 2 — PoC development and validation
- Month 3 — full development and testing
- Month 4 — production deployment and post-launch support
Challenges and how we handled them
Stakeholders
Results
| Measure | Before | After |
|---|---|---|
| Average processing time | 15 mins/application | 5 mins/application |
| Accuracy rate | ~85% | >98% |
| Applications per day | 200 | 300+ |
| Manual effort | High | Significantly reduced |
| Compliance issues | Frequent | Minimal to none |
Quantitative
- Processing time reduced by 65% — from 15 minutes to roughly 5 minutes per application
- Data accuracy improved to over 98%
- Cost per application decreased by 40%
- 50% increase in daily throughput using parallel robots and queues
Qualitative
- Improved employee morale with repetitive tasks removed
- Faster application turnaround for customers
- Strengthened compliance posture through rule-based validation
Business value delivered
Scale without headcount
Volume spikes are absorbed by adding robot capacity rather than people.
Reduced operational risk
Standardised, compliant workflows replaced inconsistent manual validation.
A platform for what's next
The engagement positioned the bank for automation across other departments.