eLend's Michael Brenning on vendor software that 'wasn't built for you' and what it builds instead
The latest convention season brought a new wave of mortgage technology, largely fueled by rapid advances in artificial intelligence (AI).
Vendor rooms were full of either new tech companies that have sprung up since last year or existing companies rolling out their latest and greatest new products.
And while there is value in all of this new technology, there is also a growing trend where lenders and brokerages are deciding to build their own technology instead of buying it from third-party vendors.
Michael Brenning (pictured top), chief operating officer of eLend, said the lender hired a technology chief from outside mortgage and is rebuilding its tools around AI.
"It's always been the build versus buy conversation," Brenning told Mortgage Professional America. "But now it's build. Build, build, build."
From vendors to engineers
Two years ago, eLend built about 10% of its technology and bought the rest. Those percentages have now reversed.
"I'm hard-pressed now to have a vendor convince me I don't want to just build it myself," Brenning said. "Because then you've got to take what they've done, integrate it with your platform. It wasn't built for you. It was built for kind of a generic solution."
This raises the question as to what will happen with all of these third-party companies that are staying in business by developing software for the industry.
"If I were a software-as-a-service company supporting the mortgage industry, I'd be scared stiff," Brenning said. "I think that model is going to break because people are going the route we're going."
In a June white paper on AI in the mortgage industry, the Mortgage Bankers Association (MBA) says lenders are increasingly developing their own AI tools, though many companies may still deploy systems from third-party vendors.
The most recent change in eLend’s tech development is a hire from outside mortgage. The company’s new chief technology and innovation officer started earlier this year, and Brenning said his past work includes logistics models for Chewy, a trading platform for Charles Schwab, and machine learning and AI at Cisco.
"Most people in mortgage have gray hair like me," Brenning said. "We can have some original thoughts, but we can't be as original or new or fast-moving. For somebody that comes from a completely different industry, that's innovating at a much more rapid pace than mortgage."
The new officer has reshaped the technology team around AI engineers, developers and programmers, in place of database administrators. Their output is the small apps and bots eLend has been rolling out almost weekly.
"We've kind of swapped out about 20 people on the team for old-school resources versus new-school resources," Brenning said. "And now, we're developing at a pace that I never thought was possible."
The company completed 10 or 15 major tech enhancements in 2024, about 100 in 2025 and several hundred so far this year, and it expects 300 to 350 by year-end. It also has 138 smaller projects in the queue.
"We're finally, for the first time, getting through the backlog of the 138, like 10 or 15 at a time, 10 or 15 in a week versus 10 or 15 in a year before," Brenning said.
AI agents and human underwriters
Industry professionals expect AI to take over document collection, data entry and CRM follow-ups before anything else, according to a July survey by AD Mortgage.
As for eLend, it already has agents in production that review powers of attorney, trust documents, homeowner's insurance declaration pages, flood certifications and title commitments. The pace of that work has picked up over the last month, Brenning said.
"Not just review it, but know it's there, it came into the folder, ‘Oh, I got a new thing to look at,’ runs it through our engine, comes back with a result to the ops associate," Brenning said. "So the ops associate no longer has to stare and compare. They get results."
Operations associates look at each result and either clear the item or pend the file for a change or clarification. Each agent covers one document type.
"That's starting to change the amount of man-hours it takes to get us through a loan," Brenning said.
The company has been building its own loan origination system (LOS) for more than a year. The new system is meant to go well beyond an LOS, with a product and pricing engine, a scenario builder for loan officers and an automated pre-approval driven by AI.
"We're moving to our own LOS sometime before this conference next year," Brenning said.
The underwriting engine inside it draws on Fannie Mae and Freddie Mac's automated underwriting systems and on FHA, VA, USDA and agency guidelines. It is built to learn from those rules over time.
Freddie Mac updated its Seller/Servicer Guide in March to require governance frameworks for AI and machine learning, and Fannie Mae's April lender letter calls for policies and procedures covering AI and ML systems, according to MBA's white paper.
"We get those underwriting results when we roll this out in 30 seconds versus 2 days or 3 days," Brenning said.
The engine produces the result, and a person still makes the final decision.
"We're never, ever, ever, ever going to have AI just signing off on loans or making decisions because legally you can't," Brenning said. "Fannie, Freddie, Ginnie, they're not going to allow a bot to sign off on something. That's why it goes to our underwriter. They have a checklist they run against."
Stay updated with the freshest mortgage news. Get exclusive interviews, breaking news, and industry events in your inbox, and always be the first to know by subscribing to our FREE daily newsletter.