BLOG VIEW: The mortgage lending industry has come a long way from the days of simple automation to create document sets for borrower closings. Today, IT teams are bringing increasingly sophisticated tools to the mortgage process – the latest being artificial intelligence (AI).
AI-powered chatbots like ChatGPT and Google’s Bard have taken the world by storm, with consumer adoption growing faster than any other technology. But consumer adoption and adoption by mortgage lenders are two very different things. It takes a lot more than deep fake artwork and an easy way to write a college application letter to get lenders to change their existing process and invest in new technologies.
But it will happen. Lenders are going to adopt AI. It’s a matter of “when,” not “if,” and, importantly, “where” they will use it.
At our firm, we see AI as being applicable to document processing – and lenders should be prepared to adopt. What follows are three reasons why:
Documents Are Still a Problem
In the beginning, loan applications were taken on a clipboard with paper and pencil and then documents were pulled manually from stacks to present to the borrower. It’s not likely that anyone reading this article will remember those days. It was before your time.
Automation, when it came, was a lifesaver.
Once we had the documents automated, the problems were (a) choosing the right doc set for the loan program and investor, (b) getting all the data required to complete the forms, (c) getting the data from the forms into the LOS, (d) getting all of the documents signed at the closing table and then (e) stacking them in the right order for the investor.
During the foreclosure crisis, we learned about a host of other problems we had with document custody, which created a new cottage industry devoted to checking all the documents and perfecting the files.
Later, the CFPB complicated the process again by adding the LE and CD and putting some timing restraints on the lender’s process through its TRID initiative.
Electronic documents have gone a long way to simplifying this process and taking the friction out of it, but we still have cubicles full of people swiveling their heads between monitors so they can index the data, make sure it matches the information in the LOS and that no requirement documents are missing.
There must be a better way.
Automation That Can ‘Think’
Automation can go a long way toward helping lenders move loans through their process more effectively and thereby become more efficient. But today’s automation technology can’t think. Or at least it couldn’t.
AI has changed all of that.
Today, we have AI models that have been trained on millions of mortgage documents so they can identify a form, ensure that it’s required for the loan origination process, index it for the digital loan file and pull all the data from it to populate the LOS. It can do all this much faster than humans using a fraction of the energy.
But the big win for AI for docs isn’t just that it will save lenders time and money by speeding up the process. It’s really about the positive impact this has on the borrower experience.
We know that prospective borrowers are completing loan applications with multiple lenders and then closing the loan with the one that doesn’t disappoint them. Winning this business means the lender must minimize friction and make the process as easy as possible on the borrower.
Virtually every touchpoint with the borrower involves a document, whether that be the application, the upfront disclosures or the VOI/E information provided by the consumer. There are a lot of opportunities to disappoint the borrower here.
Even if all the lender does is make sure all the required documents have been acquired and then never ask for the same information again, their customer satisfaction scores will rise.
This makes document management the low hanging fruit for any lender who wants to step into AI.
And every lender will step into it because the leaders have already become the first movers.
Lender Adoption Has Already Begun
Having been in the mortgage technology business for decades, we know how to identify a tipping point, that moment in time when all resistance to an idea breaks down and the early adopters are joined by the rest.
It took the GSEs about 10 years to get full adoption of their automated underwriting systems. It took the developers of remote online notarization about twice that long before the pandemic made their tools a requirement.
AI won’t take nearly that long. ChatGPT gained its first million users in a matter of days, a feat other popular online tools took months or years to achieve.
But it’s not just the technology that will lead to faster adoption in our industry.
The mortgage industry has an advantage over most industries in that our executive workforce is a smaller group made up of individuals that tend to be well connected. When something works, news spreads quickly.
But knowing about a tool and adopting it are two different things. Everyone knew that AU was coming and that RON tools were available. They just didn’t see the need to adopt.
The leaders who are adopting AI for docs now realize that competition is very intense for a smaller set of prospective borrowers. Customer satisfaction will win them more business while document-related problems are the easiest way to lose it.
AI is solving this problem and the news is spreading. And that’s all it takes to create a tipping point.
Jim Rosen IS EVP of Services at Mortgage Cadence.