Intelligent Document Processing for Lenders Mortgage Insurance (LMI) Applications
Overview
A leading mortgage insurance provider needed to get ahead of a growing document processing burden: 1,600–2,000 Lenders Mortgage Insurance (LMI) applications every month, arriving from more than 25 banking and financial institutions. Each application package came bundled with application forms, valuation reports, income proofs, servicing calculators, payslips, lender notes, and checklists — all of which required manual review before underwriting could even begin.
TechTiera engineers designed and implemented an enterprise Intelligent Document Processing (IDP) solution on Tungsten TotalAgility (formerly Kofax TotalAgility) to automate document classification, data extraction, validation, and workflow routing across the entire application package. The solution significantly cut manual processing effort, improved data quality, and accelerated underwriting turnaround — all backed by ongoing managed services once it went live.
The Challenge
This wasn’t a case of one difficult document type — it was volume, variety, and consistency all colliding at once. Specifically:
- High monthly application volumes required extensive manual document processing
- Every application package contained multiple document types, each submitted in its own format
- Manual classification and indexing delayed underwriting before it could even start
- Data extraction from financial documents was repetitive and susceptible to manual error
- Validating extracted information required significant operational effort on its own
- Processing delays were directly impacting customer turnaround times and SLA commitments
- Growing application volumes demanded a platform that could scale without sacrificing accuracy or compliance
By the time TechTiera got involved, the manual intake process had become the primary bottleneck standing between application receipt and underwriting readiness — and with more than 25 institutions submitting documents in their own formats, that bottleneck was only going to get worse with volume.
Solution
Solution Architecture
TechTiera engineers built a comprehensive Intelligent Document Processing platform automating the complete intake process — from document receipt through classification, extraction, validation, and workflow routing — combining OCR, machine learning-based classification, business rule validation, and workflow automation into one processing environment:
Given that each application package could contain any mix of document types from any of 25+ institutions, the classification and extraction layers had to be genuinely flexible rather than tuned to a single lender’s format.
Solution Components
Intelligent Document Intake
The solution automatically received mortgage application packages from participating financial institutions and prepared them for automated processing — handling mortgage insurance applications, checklists, property valuation reports, servicing calculators, cover sheets, payslips, income documents, lender notes, and supporting financial documentation.
Automated Document Classification
TechTiera engineers developed intelligent classification models to automatically identify and categorize each document within an application package — classifying document types, grouping related documents, indexing application packages, and routing documents for downstream processing. This is what let a stack of unsorted documents from any lender become a properly organized, processable application package without a human sorting it first.
Intelligent Data Extraction
The platform extracted business-critical information from each document — applicant details, loan information, property information, income details, valuation data, financial figures, reference numbers, and supporting metadata — automatically structuring it for downstream underwriting systems.
Customized Validation Framework
TechTiera engineers built customized validation forms and business rules to verify extracted information, covering mandatory field verification, business rule validation, data completeness checks, confidence score review, exception handling, and user-assisted correction where needed. This was a deliberate design choice — rather than pushing every extraction straight to underwriting, low-confidence fields get flagged for a quick human check first, which matters a lot in a regulated financial process like this one.
Workflow Integration
Validated application data was routed directly to downstream underwriting and business systems, significantly reducing manual intervention while preserving process governance and auditability throughout.
Production Support & Managed Services
Following deployment, TechTiera continued providing production monitoring, application support, enhancement development, performance optimization, incident management, and continuous improvements.
4. TechTiera Responsibilities
TechTiera was responsible for the complete implementation lifecycle, including:
- Solution architecture
- Project planning and timeline management
- Document classification design
- Extraction group development
- OCR configuration
- Validation form development
- Business rule configuration
- Unit testing
- User acceptance support
- Production deployment
- Hypercare
- Application support
- Managed services and continuous optimization
5. Technology Stack
RPA Platform: | Tungsten RPA (Kofax Kapow) 9.7.3 / 9.7.4 / 10.2 |
Intelligent Document Processing: | Tungsten TotalAgility (formerly Kofax TotalAgility (KTA)) |
OCR & Classification / Workflow: | Intelligent Document Classification & OCR (KTA) |
Data Extraction: | Automated Field Extraction |
Validation: | Custom Validation Forms & Business Rules |
Workflow: | Intelligent Document Routing |
Database: | Microsoft SQL Server |
Operating Environment: | Windows |
Managed Services: | Production Monitoring & Application Support |
6. Results and Business Impact
Measurable outcomes
| Area | Before | After |
|---|---|---|
Application volume handling | Manual review of 1,600–2,000 packages/month | Automated intake and classification at scale |
Document classification | Manual sorting by document type | Automated classification and indexing |
Data extraction | Manual entry, error-prone | Automated field extraction across document types |
Validation | Fully manual review | Rule-based validation with confidence scoring and exception routing |
Underwriting readiness | Delayed by manual intake | Faster routing to underwriting teams |
Business Benefits
Benefits achieved by TechTiera’s automation implementation:
7. ROI
Major outcomes for engagement include:
- Return on investment within 6–12 months
- Significant reduction in operational processing costs
- Improved productivity of underwriting support teams
- Faster customer response times
- Reduced rework through improved extraction accuracy
- Enhanced scalability without increasing processing headcount
8. Project Execution Timeline
9. Project Hurdles and Lessons Learned
Key Challenges
Mortgage application packages contained multiple document types with varying layouts and scan quality, which meant the classification models had to be genuinely robust rather than tuned to a handful of clean templates. Financial documents varied significantly across the 25+ lending institutions submitting them, so extraction and validation rules needed real flexibility rather than a one-size-fits-all configuration. Getting extraction accuracy high enough for critical financial data took iterative tuning of both OCR and business rules — this wasn’t something that worked perfectly on the first pass. Custom validation forms had to strike a careful balance between full automation and efficient human review for low-confidence cases, since regulated financial processes don’t tolerate silently wrong data. And keeping performance consistent as document volumes grew required continuous monitoring and optimization rather than a single tuning pass at go-live.
Lessons Learned
Intelligent document classification made a real difference in cutting manual indexing effort, especially when processing mixed packages with several document types bundled together. Bringing underwriting teams into the design process early improved both validation rule design and exception handling — they knew which fields actually mattered and where errors were costly. Configurable extraction models made onboarding new lenders and new document formats meaningfully easier than building bespoke logic for each one. Human-in-the-loop validation turned out to be essential, not optional, for maintaining confidence in a regulated financial process — full automation without a review layer for low-confidence extractions would have been the wrong tradeoff. And ongoing managed services were critical for sustaining extraction accuracy over time, since document templates from participating lenders kept evolving even after go-live.
10. Best Practices Implemented
- End-to-end Intelligent Document Processing architecture
- Automated classification of mixed document packages
- Configurable extraction groups for diverse document formats
- Custom validation forms aligned with underwriting requirements
- Business rule-driven exception handling
- Human-in-the-loop review for low-confidence extractions
- Comprehensive audit logging and document traceability
- Structured deployment, testing, and change management
- Continuous production monitoring and optimization
11. Security & Compliance
The solution was designed to support the stringent governance and regulatory requirements of the financial services sector, including:
- Secure processing of customer and financial information
- Role-based access controls for document processing and validation
- Comprehensive audit trails for document receipt, extraction, validation, and workflow routing
- Version-controlled deployment and change management processes
- Secure storage and transmission of mortgage application documents
- Alignment with enterprise information security, operational governance, and financial services compliance requirements
- Controlled exception management to support underwriting quality and regulatory auditability
12. Business Value Delivered
TechTiera transformed the mortgage insurance application intake process through an enterprise Intelligent Document Processing solution. By automating document classification, OCR-based data extraction, customized validation, and workflow routing, the solution significantly reduced manual effort, accelerated underwriting preparation, and improved data quality. Through end-to-end ownership — architecture, implementation, production deployment, and ongoing managed services — TechTiera delivered a scalable, secure, and highly efficient document automation platform that enabled faster mortgage insurance processing while supporting the organization’s growth and operational excellence.