As the firm aggressively scaled its HELOC and secured business loan volume, the underwriting team was crushed by unstructured data. Processors were manually downloading complex UK Self-Assessment (SA302) tax returns and non-standard Profit & Loss statements from email threads. They were forced to manually read these PDFs, extract net profits, add back depreciation, and re-key the data into spreadsheets to calculate debt-service ratios. This "Human API" created a massive bottleneck, risking processor burnout and slowing down turnaround times.
Automated UK Tax Parsing
& Inbox Triage
The Bleeding Neck
Underwriters acting as highly paid inbox sorters.
Architecture & Prototype
Qeiva OCR Engine: UK Tax (SA302) & P&L Parser
1. Inbox Extraction: Raw PDF Data (Unstructured)
Qeiva deployed an Automated Document Ingestion & P&L Extraction Pipeline. The custom Python architecture actively monitors the intake inbox, intercepts the email attachments, and runs targeted OCR against the complex UK SA302s and P&Ls. It automatically extracts the required income entities and pushes the clean, calculated data straight into the Loan Origination System.
The ROI
40+
Hours Saved/Month
100%
Inbox Automation
Max
Data Extraction Velocity
Before: Underwriters spent 15–20 minutes per application merely acting as inbox sorters—downloading SA302 tax returns, renaming files, and manually re-keying P&L data.
After: The automated ingestion pipeline intercepts, reads, and pushes the data to the Loan Origination System instantaneously without human intervention.
Net Impact: Protects margins by allowing the firm to scale aggressive loan volume without hiring manual data-entry clerks, completely eliminating cognitive fatigue.