As
The best move for mortgage companies to get up to speed on the technology, specifically when it comes to preparing their personnel, is to first decide upon a clear purpose and plan that plots the course ahead, industry experts say.
If a company knows exactly what efficiencies they want to create with AI, that's one thing, "but if it's just because we like the word, and we want to put a hashtag behind it, maybe don't train the employees yet," said Suzanne Krause, founder and CEO of Simplify the Machine, a mortgage industry management and technology consultancy firm.
In research from Arizent, parent company of National Mortgage News, published in 2024,
Coming up with a clear AI plan and value statementCompany leaders looking to build their artificial intelligence know-how should look at their own specific circumstances before getting started, Krause said.
"If there can be a value statement for your business when it relates to AI — if you can come up with what that is and if the messaging can come from the tech team — you will certainly have much more control over the use cases," she said.
"I think that any place that we have a known pain point is exactly where we should point it," Krause added.
With mortgage professionals viewing AI differently depending on their role or responsibilities in an organization, balancing expectations is an important component for smooth adoption across a company as well, according to Sridhar Sharma, chief information officer
"Yes, this can do wonderful things. But you also have to be a little careful in how you roll this out," he said.
"Compliance is always a worry. I think that's part of life," he said. "The other thing I'm noticing recently is this fear of missing out." But moving too quickly can also lead to problems when outside vendors become the sole source for AI tools and training.
"Don't give all your eggs to a vendor basket," Krause noted. "Make sure you know how you can use these tools to help yourself be sufficient and help your organization be the best that it can be."
The tools have great potential to solve thorny industry problems, "as long as we're bringing some knowledge in-house," she said.
Addressing security and compliance issuesDeveloping some internal expertise also serves a mortgage business from a security and compliance standpoint. More than one-quarter of financial industry professionals cited increased threat of cyber attacks or fraud as a deterrent toward AI adoption
Several high-profile cyber hacks over the past two years
"I think what's happened is just some of these big stories getting headlines of breaches is really everyone's biggest concern. In this space of AI, they're leaning out to third-party vendors to have competency, and that can be inherently risky," Krause said.
At the same time, artificial intelligence can only be effective as the information that goes in it. Much of the negative publicity in AI today often revolves around the data artificial intelligence works with rather than software.
The challenge lies in understanding that technology can learn good things as easily as bad, Sharma said. "It's the underlying data that we want the engine to learn, so we want to be very careful about what it's learning."
"We discovered this early on, and we had to pivot and make changes to our approach, which was part of the difficulty," he added. The process involved curating data to ensure accuracy of AI-generated output.
Curation is essential to determine what genuinely should be considered quality data. Applying arbitrary criteria can have shortcomings, Sharma continued, using a hypothetical example of trying to train AI on phone calls.
"A good phone call is a call that finishes, let's say, in six minutes," he said. "I have phone calls that finish in six minutes because the agent says, 'I don't know how to help you. Call me tomorrow.' That's not a great call."
Possible use casesWhen the data is correct, though, and artificial intelligence is able to analyze it effectively and quickly, the technology will open doors for lenders and servicers to "rewire" their approach to customers in a more holistic way, said Aditya Swaminathan, Americas consumer lending and mortgage leader at EY.
Mortgage employees, including processors or loan collection agents, in turn, will move away from rote information retrieval or outbound calling towards tasks that develop reasoning and outreach skills in a compliant manner.
"You move from outbound dialing for dollars to inbound customer engagement for those customers who are willing but not able to pay, and you actually are upskilling the collections agent," Swaminathan said.
"Rather than just being an outbound, 'Hey, you owe $1,000,' to actually becoming some degree of financial advisor, because they're talking about a loan modification that may be needed," he added.
On the lending side, many loan-origination systems have relevant procedures and policies and workflow management now baked in thanks to generative AI. The upgrade in technology means a processor can develop into a role more akin to a junior underwriter. With document preparation becoming automated, time opens up to examine complex situations and determine when more work on a file is needed.
"That's one area we see in that processor-underwriter bucket where the upskilling is going to happen," Swaminathan said.
Just because the potential of AI may seem unlimited, a thoughtful approach may be the best strategy for long-term success. While headlines may tout artificial intelligence capabilities, the reality of applying new technology is often more complicated.
"Everyone makes it sound like you install it, and you're good to go. It's not really going to work that way," Sharma said.
"This is a new area," he said. "I think there's a lot more merit in being very pragmatic and being methodical about it, than to try and rush into it."
