AI hits underwriting: 57% of pros predict change

Underwriting processes will see the greatest transformation — and the biggest lift — from artificial intelligence this year, and the outcomes it produces should aid in instilling confidence among buyers and lenders. 

Processing Content

In a recent National Mortgage News survey, 57% of respondents predicted AI-driven underwriting would create the greatest mortgage industry change this year, with technology now at a point to begin achieving breakthroughs discussed for decades.  

The forecasts come from a survey of over 150 mortgage professionals conducted in November and December 2025.  

A clear set of established rules and guidelines that can be read, learned and then applied to files without potential human bias creeping in makes underwriting like a "perfect use case," according to Theo Ellis, CEO of fintech platform Friday Harbor.

"I don't think it's happened in 30 years because the real world's really messy," he said, alluding to different versions of analysis used or extra forms that might appear in a file. "It's very hard to shoehorn that into AI," Ellis added. 

With the rapid improvement of large language models, though, AI today can handle the messiness. The earlier it can be applied in the lending process means much of the heavy administrative lift can be done in the pre-approval stage of originations, where it can read files and orchestrate and route workflow, identifying anything that might stall or suspend a loan. 

"It really is designed to enhance workflow, and that's what people are leveraging it for today," said John Brumund, senior vice president at Quontic Bank's mortgage division and a former broker. "You're going to get more efficiencies out of a loan that's been put through some kind of AI system before it gets to underwriting."

At the same time, the functions that feed into the underwriting process also illustrate the extent to which the mortgage industry can expect to see AI develop.     

Fifty-one percent of the survey respondents similarly expect improvements in credit scoring and analysis due to use of AI, while 49% predicted changes to help streamline real-time employment and income verification processes. 

The information that AI can make readily available means loan officers are able to deliver a clear picture of a borrowers' finances early, a benefit likely to prove attractive to selling agents.  

"It actually makes more sense to equip originating teams with that capability, so they can get more confident preapprovals," Ellis said. "Underwriters have a way of now focusing on the true risk management decisions."

Other factors driving the pace of adoption

While clear data-focused guidelines open the door to underwriting adoption this year, the relaxed regulatory stance in the second Trump administration also may encourage AI growth. 

Forty-one percent expect an overall loosening of mortgage regulation at the federal level, with another 8% seeing little change. A 5% share said the U.S. government's focus will highlight artificial intelligence, data privacy and cybersecurity matters. 

In thinking about AI and how policy affects their plans, over a third, or 37%, of respondents said the current regulatory environment encouraged them to speed up their use of artificial intelligence in underwriting. The share came in just under the 39% who said they would pick up AI implementation for marketing purposes, where identifiable data generally does not come into play.

Still, mortgage leaders caution about deprioritizing compliance simply due to what they see as easing policy. 

"It's not relaxed at the state level," Ellis said. "Everyone's mindful of that, and even at the federal level, most large lenders want to do the right thing. They're thinking longer term."

No matter what goes on among lawmakers, compliance and security should remain top of mind among lenders by necessity, Brumund said. 

"If you expose somebody's personal data, it's a huge problem for everybody; personal identifiable information right now is what we have to protect. The unknown of what happens when it goes into an AI system is not acceptable today." 

While caution could factor into the pace of AI uptake, tech providers will also still have industry inertia to confront, according to other leaders. 

"Besides the obvious governance and accuracy concerns, there's operational risk to consider. Using AI to underwrite loans means lenders will need to redesign their processes, and in our industry, that does not happen very quickly or cleanly." said Flyhomes CEO Tushar Garg. 

Currently, adoption momentum can't be tied entirely to a single underlying development. Instead, it appears to be emerging organically as businesses  and their lending teams recognize the possibilities when they see peers' gains from AI implementation, according to Ellis.

"There's a tremendous amount of grassroots pull from loan officers and their teams," Ellis said.

"We're going to see shorter cycle times. Every day that you can bring in clear-to-close is worth several basis points of value to the lender," he added. 

"When someone sees some major lenders that are gaining huge efficiencies or delivering a great customer experience, and they're doing it at scale, not in pilot toy programs, then I think there will be a rush to do it."

For reprint and licensing requests for this article, click here.
Underwriting Artificial intelligence Editorial Research
MORE FROM NATIONAL MORTGAGE NEWS