Whereas top lenders on day one urged firms to maintain a human touch, tech lenders on day two advised competitors to get their entire staff up to speed on AI, before it potentially replaces them. 

Here are more takeaways from the second day of the Digital Mortgage conference. 

What are the gaps in lenders' tech stacks?

Lenders prioritize data security enhancements but are still seeking solutions around borrower documentation, according to new research by Arizent, parent company of National Mortgage News.

Among 128 professionals surveyed this summer, executives cited inconsistencies or incomplete documentation from borrowers as a leading challenge. They're seeking to reduce gaps in data coverage for borrowers including gig workers, self-employed and unbanked consumers. 

Originators told Arizent they want automation to reduce cumbersome manual reviews. It's unclear however how far potential artificial intelligence solutions will go in the industry. Only a small minority of those surveyed said they're using agentic AI or copilot tools. 

Will your workforce adopt AI peacefully?

Some experts cautioned to ease integration among teams that may not be ready to mesh with the tech. Mark Hansen, executive vice president of product at CMG Financial, suggested unprepared employees would produce so-called "AI slop," or low-quality media. 

"The challenge with that is, and this is the honest truth, not all of your employees are going to be able to use AI," he said.

Still, everyone needs to be familiar with the emerging tools. Dawn Svedberg, executive manager of lending and fintech at CI&T, said her firm trains every single person at a financial institution to learn and use its CI&T/Flow AI platform. 

Teaching your most qualified workers to use AI will also be a challenge, Hansen added.

"We call it the two-week rabbit hole," said Hansen. "You've got to learn it, figure out the best practices, and then eventually you sort of plateau, and then you get it, and now you're like this AI genius. How do you replicate that amongst your whole staff?"

The return-on-investment potential also expands to data projects. Agentic AI, the human-sounding chatbots, could supercharge ROI on vast databases by meshing with it, said Joe Welu, founder and CEO of Total Expert. 

Jay Promisco, president of Sierra Pacific Mortgage whose origination assets are being acquired by Union Home Mortgage, continued to urge lenders to heed the rapid tech advancements that could enhance, but replace, their colleagues. He explained that upgrades in Sierra Pacific Mortgage's fulfillment have helped his staff exponentially multiply their loan output while reducing the company's offshore capacity dramatically. 

The executive also predicted that in two years, an originator will only need a database and native language model to create origination volume. Lenders are eventually going to face difficult decisions in reducing their workforces.

"If you started a mortgage company today, would you build it differently?" he asked attendees. "You would."

The Mortgage Bankers Association's second quarter Mortgage Bankers Performance Report found lenders slashed expenses by 60 basis points per loan, or over $1,600, via factors including less hiring. 

Joel Kan, the trade group's vice president and deputy chief economist speaking Wednesday at Digital Mortgage, pointed to a Federal Reserve study showing service firms using AI to retrain their workforce rather than replace them. 

"Do we think this is going to show up more broadly across the job market?" asked Kan, speaking Wednesday at the conference. "I think it's kind of a wait-and-see."