The clamor for artificial intelligence to augment lead generation, loan processing and pipeline hedging technology intensified in 2025 as mortgage vendors and lenders alike unveiled a range of AI-powered tools aimed at improving loan quality and accelerating closings.
So, a digital graveyard quicky fills with the corpses of chatbots usurped by more eloquent updates, littered with proofs of concept that had exhausted their usefulness in pitches.
Deciphering the true value of a novel piece of software from the marketing of a prototype polished for the conference circuit remains a challenge for mortgage lenders. But it nevertheless makes the difference between maintaining or squandering one’s competitive edge in an era of low production and rapid consolidation.
“I would say, in broader strokes, probably the majority of it is a massive letdown,” says Rick Chakra, director of AI solutions at Mortgage Capital Trading (MCT), a capital markets advisory firm. He joined MCT on a permanent basis at the end of 2025, having previously been a senior consultant in Deloitte’s applied AI division, focusing on financial services.
“The barrier to entry is so low,” he tells Scotsman Guide in a recent interview, noting that many companies simply wrap large language models around chatbots, leashing those chatbots to workflows like dogs to their kennels. Achieving “real outcomes” that meet business needs and square with internal audit controls separates the players from the pros, he says.
As mortgage lenders race to adopt AI, experts warn that poor strategy, weak governance and unclear goals can ultimately undermine efficiency gains, leaving providers increasingly stuck navigating hype, unrealistic expectations and the slow grind of change management.
“The winners with material outcomes with AI are the platform companies that have agents with really strong guardrails and really good transparency,” predicts Chakra. “You see companies dumping so much money into AI that actually adds very little value.”
Within such an environment, Chakra sees three major risks: investing in AI tools that fail to deliver measurable improvements; assuming compliance and data security hidden by inadequate governance; and adopting AI solutions too quickly, thus introducing new frictions.
Lenders must have AI strategies
But experts also tell Scotsman Guide that many mortgage lenders clamor for AI solutions without being able to articulate what they want AI to accomplish. Failure to align AI integration with existing workflow limitations can ultimately reinforce the siloed loan processing infrastructure many lenders struggle with now.
The irony of that fact is not lost on Jane Mason, founder and CEO of loss mitigation and default servicing platform Clarifire. She says the consistency and reliability of AI-powered outputs inherently depend on the quality of related inputs, as investors and regulators increasingly demand visibility into the logic behind AI decisioning.
“I don’t think they know where to start,” says Mason. “I believe if you don’t adopt AI, you’re going to lose. But if you do adopt it you could lose too, if you don’t think about it smartly.”
Mason started the year working with a large investor who requested algorithmic verification before accepting that a given loan decision was entirely positioned by AI. Only the “really smart mortgage entities” can demonstrate that level of provability, she says, underscoring significant governance gaps across mortgage AI.
Meanwhile, Mason says that “there’s not a lot of servicers that we’re talking to that really know where to start and what to do,” with most setting their AI sights on lower-hanging fruit like streamlined document processing, call dialing and chatbots on the front end. “If you can’t prove it, we can’t let you use it,” she recounted from conversations with Fannie Mae and Freddie Mac, describing the broader investor attitude toward AI decisioning.
“AI is an accelerator and an aggregator, so it doesn’t really matter what part of the mortgage industry you’re in, whether it’s originations or servicing,” adds Mason. “AI is not positioned yet and probably won’t be in 2026 to actually be able to make a decision and approve something from an underwriting perspective.”
Facing such constraints, lenders have an opportunity to tailor AI adoption more precisely to existing workflows.
Vendors holding lenders’ hands
Operations management veterans like Alok Datta, founder of Alchemist Solutions, believes tailored AI solutions will level the field between small lenders and larger competitors. He provides automated fulfillment services to small and midsize mortgage lenders, from pre-underwriting decisions to post-closing correspondent loan reviews.
“It’s like an elephant at this stage, and we are all blindfolded and we are touching different parts of the elephant,” explains Datta. “To me, the trunk seems to be AI and to somebody else the legs could be,” reinforcing Mason’s point that a general awareness of the AI elephant does not directly translate to productive AI integration or a return on investment.
OpenAI’s ChatGPT first launched to public fanfare more than three years ago in late 2022. Mortgage stakeholders are “starting to get a little jaded” about the degree of enhancement AI will bring to bear on the industry in the near term, says Datta, but he also understands the cash-intensive funding models driving innovation in the background.
“I think the providers haven’t helped their cause by claiming that they can boil the ocean when as of now the industry is still evolving,” Datta says. “We are in the business of change management, where we have to work with the lenders in terms of making them understand what the future looks like and holding their hands to take them to that future.”
He nevertheless believes it’s incumbent on executives to participate in the evolution of their companies with AI. Some choose to ignore the holistic change that AI solutions represent, wanting instead to only see the results.
Datta’s company operates on an evolved strategy, as he has been forced to adapt to what deliverables current AI technology can reliably support and what demand for such solutions the current market can afford.
“We went in saying we can do the whole nine yards from the time an application comes in until it’s set up for funding,” he explains, “only to realize that there were a lot of people who were excited by the program but nobody who wanted to buy it because that was a significant change to their existing process.”
Leveraging AI to fill market gaps
Construction loan management once represented a particularly siloed corner of the mortgage lending universe that should substantially benefit from AI-powered automation, believes Sean Faries.
He founded construction loan management platform Land Gorilla on a $500 credit card line a few years after the 2008 financial crisis — during which period he lost both his personal residence and business and moved back in with his parents with a baby on the way.
Faries says success at finding near-term efficiencies and savings will be shaped by how deliberately companies deploy investment into new AI solutions. He attributes the centralized design of Land Gorilla to the demands of his early customers, who clamored for a single loading dock to handle inbound loan applications, inspection documents and funding requests.
“You hear so many CEOs talk about listening to your customer,” says Faries. “That’s all we did was listen to our customer and they really helped shape the product.” The current technological revolution provides a similar opportunity for companies to pursue targeted and collaborative growth that responds to market demands, he believes.
“I think AI is the closest thing to magic that we’ve ever seen,” he adds, noting that the generalized nature of today’s demands on AI is one of the most interesting features of its evolution.
“Everybody wants AI, but they don’t know how to articulate really what they want the AI to do. I actually love that for a lot of selfish reasons,” says Faries, as it provides the opportunity for Land Gorilla to build lenders more tailored solutions.
Most of the “battles being fought” with innovation involve peeling workflows into ever smaller “micro tasks” for automation, Faries observes. More specifically, he says Land Gorilla has been on the “AI treadmill” for two full years developing an agentic AI, now in beta testing.
“We’re getting ready to release it to select customers now for them to experiment,” Faries says, in a potential sign of things to come for AI in mortgage lending.
