JazzX AI’s Mike Romano on why mortgage AI needs to do more than automate tasks
In a conversation with HousingWire’s Allison LaForgia, Mike Romano, Director of Sales at JazzX AI, shared how artificial intelligence is moving beyond basic automation to handle more complex mortgage lending tasks. Romano explained the differences between generative AI, agentic AI and enterprise general intelligence (EGI), highlighting how whole-loan reasoning, human oversight and AI coworkers could reshape mortgage underwriting and improve productivity.
Romano began by addressing one of the industry’s biggest misconceptions about mortgage AI automation. “I think that the biggest misconception is that all AI is the same, when there’s actually really been a distinct evolution already within AI, because the space is moving so quickly,” Romano said.
From generative AI to enterprise general intelligence
Romano explained how different AI technologies perform distinct functions, beginning with generative AI. “Generative AI is your ChatGPT, for example. It’s very good at summarizing information for you or performing a simple task, but you’re still, as the user, giving it that context of what to do,” he said.
He shared how Agentic AI takes that functionality further but still has limitations. “Agentic AI can absolutely perform a task, like a waterfall set of rules to follow, but it typically lives within pretty narrow guardrails, and those guardrails become quite brittle actually if something that’s outside of the expected path is introduced,” Romano explained. He pointed to the unpredictable nature of mortgage lending, where new documentation can change an entire loan scenario.
Romano then introduced enterprise general intelligence (EGI) as the next evolution. “This is where you can have coworkers that actually think and reason across an entire loan and understand the entire scenario across the whole loan, including things like appraisal or title, not just income and assets, but to be able to think and reason and collaborate across the other assistants to actually perform meaningful amounts of the person’s work,” he said.
Managing risk and evaluating mortgage AI underwriting
For lenders considering building their own AI solutions, Romano emphasized the challenges of managing regulatory requirements. “As we know, the industry is highly regulated, and there’s a lot of risk,” he said. “To build an AI model that can keep up with all those changes and the governance of it, and then still perform those tasks, because those tasks and needs are constantly changing too, that’s a really difficult task for most lenders.”
When evaluating vendors, Romano recommended testing how AI responds to changes in loan circumstances. “You need to be able to ask what it’s reasoning across. What risk categories are included in your analysis? Because my underwriter needs to look at a bunch of different things,” he said.
He emphasized that effective mortgage AI underwriting requires understanding relationships across the entire loan. “A model that can reason will understand the downstream effects once something else comes in and when something changes,” Romano said.
Human oversight and the future of mortgage productivity
Despite advancing AI capabilities, Romano stressed that human oversight remains essential. “A human has to touch it and ultimately make that decision on the underwriting. The decision can’t be left entirely to an AI assistant or AI coworker,” he said.
He also highlighted the importance of monitoring AI performance. “A human should also be involved in overseeing how these assistants are continuing to perform,” Romano said. “Are you watching for any hallucinations or anything that changes in the wrong direction?”
Looking ahead, Romano described how AI could transform mortgage productivity without necessarily eliminating existing roles. “There’s absolutely no reason to fear AI that’s coming. I think it’s important to understand that we should all learn to work with it,” he said.
He explained how processors and underwriters could eventually oversee significantly larger pipelines. “Instead of a processor or underwriter managing 20 to 40 loans in their pipeline at any given time of the month, you might now be overseeing a team of assistants that is managing 4x that,” Romano said. “Your role as a processor, a lead processor or underwriter is more in managing those teams of assistants instead of just only focusing on your day-to-day tasks.”
For Romano, the future of mortgage AI is about helping teams work more efficiently. “I think it’ll make us a lot more productive.”
Featured in this interview
Managing Editor of the Content Studio at HousingWire
Learn how JazzX AI is advancing whole-loan reasoning and mortgage productivity.
Latest Videos
JazzX AI’s Mike Romano on why mortgage AI needs to do more than automate tasksOct 14, 2026
Spring EQ’s Joe Steffa on home equity growth, AI and the future of mortgage lendingOct 14, 2026
Checkr’s Felicia Lee on income volatility, mortgage affordability and the future of verificationOct 14, 2026
Western Alliance Bank on expanding mortgage liquidity, warehouse lending and payment innovationOct 14, 2026
Sponsored Videos
Western Alliance Bank on expanding mortgage liquidity, warehouse lending and payment innovationOct 14, 2026
Spring EQ’s Joe Steffa on home equity growth, AI and the future of mortgage lendingOct 14, 2026
Checkr’s Felicia Lee on income volatility, mortgage affordability and the future of verificationOct 14, 2026
JazzX AI’s Mike Romano on why mortgage AI needs to do more than automate tasksOct 14, 2026
