Bringing Intelligent Automation to the $10 Trillion Mortgage Industry

We previously wrote about how financial services are being transformed by intelligent automation. Let’s have a closer look at how this is actually happening in one of its largest asset classes — mortgages.

Mortgages in the U.S. today total more than $10 trillion. But it’s also a highly regulated industry, which has resulted in a complex mortgage process. The processing of mortgage applications is largely manual, involving many legacy systems and databases. The applications themselves often include more than 500 pages of documents, with 1,700 data points that need to be reviewed and extracted into systems. As a result, the process can be very error-prone.

It’s no surprise, then, that mortgage processing is inefficient and expensive, as lending costs have risen to around $7,000 per loan.

A better approach involves using Robotic Process Automation (RPA), or software bots for task execution, combined with cognitive automation solutions such as IQ Bot to automate the mortgage process end-to-end.

A typical mortgage process starts with an application form. The multi-page PDF file is often hundreds of pages long with multiple document types and formats such as W-2, 1040, bank statements, standard application forms, and so on.

A cognitive automation solution like IQ Bot can pick out the important pages of interest. It can split the document to identify relevant pages such as loan application forms, bank statements, and pay stubs. A human mortgage analyst can then use IQ Bot to identify the form fields such as name, type of credit, date of birth, social security number and so forth that need to be extracted from the relevant pages.

Using artificial intelligence techniques, IQ Bot can then be trained to extract the right data from the borrower’s loan application form. Once IQ Bot completes extraction on the sample form, the human worker can review and guide the bot to the right field if it needs correction. Each field identified can be edited, marked optional, and re-edited if needed by the user.

Once the training for IQ Bot is completed, IQ Bot can be put into production. Data is then extracted by IQ Bot from bank statements, W-2s, 1040s, and other documents into a structured CSV format. From there, other bots can pull information from the CSV to upload it into the right ERP and other systems for processing. In production, a human mortgage analyst can continue to validate the data as it’s extracted. IQ Bot data extraction capabilities improve over time via supervised learning.

By using cognitive solutions together with RPA, banks can reduce mortgage origination times through faster document processing. This also frees up human workers from manual data entry to focus on more value-added activities. And the reduction in mortgage approval times can help increase customer satisfaction with the bank and improve loyalty.

You can try out IQ Bot and see how it can be applied to the mortgage industry and others by downloading Automation Anywhere’s Community Edition.