Mortgage Processing Demo

End-to-end document packet processing with high variability and complex validation rules

Consumer finance teams have to evaluate many different types of documents in order to determine whether a customer is eligible for the loan they are applying for. The process for evaluating a mortgage application is one of the most complex types of loan reviews, requiring careful analysis of identity documents, tax documents, bank statements, paystubs, and property inspection reports, among other forms of paperwork.

Additionally, these many disparate documents are submitted as a single packet. The documents within the packet have dependencies that can affect applicant’s eligibility and which need to be analyzed in context in order to decide whether or not to approve the loan. Before approving the loan, the packet has to be evaluated and determined to be in good order / not in good order.

Hyperscience streamlines the mortgage application process by ingesting documents at scale and processing each packet successfully, in line with the accuracy requirements at the field, document, packet, and in-document validation rules required.

In this demo, you’ll see:

• Document packet ingestion and classification at scale
• Data Extraction with clear definition of fields, page layouts and the business-required accuracy levels
• Human-in-the-Loop and Business Decisioning
• Paystub processing for income verification – a complex semi-structured document automation flow that uses sophisticated inference layering with CPU and GPU based ML models for optimal performance and accuracy