Infosys Mortgage Solutions, part of Infosys Cobalt encompass several business process automation solutions for the mortgage industry, with a primary focus on document-centric processes that are manual and effort intensive. The solutions are built on open-source and employ state-of-the-art techniques like computer vision, NLP and machine learning including correlation, predictive analytics and classification. Deploying these solutions has helped our mortgage clients achieve significant value.

Automation solutions delivering value to mortgage processing


Optimize processes and deploy automation and AI to drive efficiency, predictability and governance, while enhancing compliance and customer experience. Use robotics to automate manual and repetitive functions. Use AI driven tools for administrative tasks. Leverage the power of intelligent analytics, with flexible technology, to demonstrate higher relevance and take away market share.


Challenges & Solutions

Infosys Document Review solution automates loan auditing with the following features:

  • ML-driven document classification to identify the right document types and versions
  • Extracting relevant data from multiple loan document types (using OCR and NLP)
  • Enriching the extracted data to ensure accuracy in standardized formats
  • Applying audit rules as per process requirements
  • Efficient and rich UI for user validation/checks

This solution delivers more than 40% effort and time savings. For one of our clients, it has generated over US $2 million in savings across 2 years

Infosys Due Diligence solution automates high effort tasks involving data extraction from different loan documents and analysis of unstructured data. Its features include:

  • ML-driven document classification to identify the right document types and versions
  • Extracting relevant data from multiple loan document types (using OCR and NLP)
  • Identifying sentiment from unstructured contact history notes to uncover loans with negative/similar sentiment
  • Extraction of relevant data regarding loan-specific transactions from large prior servicer payment documents (comprising 10,000 lines or more)
  • Combined and correlated summary view for the underwriter to take quick and informed decisions

A large US-based mortgage servicer implemented this solution and achieved 50-60% savings within a single year

Infosys Customer 360 Insights solution involves the following:

  • Ingestion of data from 9 different sources, some of which include customer contact history, letter sent history, note history, payment history, call transcripts, and dispute activity
  • Learning from past cases of disputes and defaults to uncover data patterns by analyzing more than 1 million cases
  • Leveraging patterns to predict which of the current borrowers are most likely to default/file a dispute (using techniques like sentiment analysis, predictive analytics and correlation analysis)

The insights generated are analyzed by business groups to initiate actions that mitigate borrower dissatisfaction and adverse behavior

Infosys Prior Servicer Payment History Solution

Typically, input document files are large in size (over 1GB), contain data of 100 to 1000 loans, are stored in various file types (CSV, PDF, text), and comprise multiple document formats depending on the source. Our solution supports the seamless extraction of loan data with features like:

  • Classification of semi/unstructured payment history files based on the document format
  • Separating loan-wise payment history from the consolidated file and extracting all transactions for a single loan
  • Transformation of the data to the desired format
  • Automatic validation of payment history against loan boarding details

A US-based mortgage servicing company used this solution to process over 700,000 cases and slash effort by 90% within one year.

Infosys Automated Letter Auditing Solution extracts, enriches and validates data in letters sent to borrowers. It uses system data to identify discrepancies. The solution supports:

  • Straight-through validation of letters when no discrepancy is found
  • Improved scalability through 100% auditing of all letters generated on a daily basis
  • Reduction in manual effort thanks to straight-through processing for correct letters