Previous Non-QM lending remains a complex and occasionally fragmented area of mortgage finance, and it’s exactly where technology has the opportunity to make an impact. In this conversation, Max Klein, Founder and CEO of Loanlight, sits down with Allison LaForgia to discuss how the company is rethinking underwriting for the non-agency market.
Klein explains how Loanlight approaches underwriting, and why thoughtful automation, not blind AI adoption, is the key to improving speed, consistency, and compliance. He also shares his perspective on where AI can truly deliver value in mortgage operations and how non-agency lending underwriting may evolve as technology matures.
“Loanlight was born from the need to serve the growing economy of non-W-2 employees better,” Klein said. “It was built to serve lenders who want to serve these borrowers and really just maximize the impact that an underwriter can actually have on the decision-making process, instead of spending time poring over documents and, you know, staring and comparing.”
Klein said the non-agency market has changed dramatically in recent years. “The non-agency space has gone through what I believe like a tremendous transformation from a bit of an outside misnomer, to a mainstream one of the critical drivers of the modern mortgage landscape,” he said.
With the expansion of non-QM, he said, “the challenges have become so apparent,” creating the need for “new tooling” to “drive innovation” and “transform this into a scalable, verifiable asset class, akin to agency lending.”
Unlike agency lending, Klein noted, non-QM lacks a standard decisioning backbone. “Agency lending relies on the backbone of DU and LPA for decisioning,” he said, but “non-QM has no equivalency.”
Loanlight’s answer is to build “an intelligence layer that standardizes quality, eligibility and liquidity,” with an initial focus on “the quality and eligibility layer first.” The goal, he said, is to create “a shared underwriting language” that both lenders and buyers in the non-QM ecosystem can rely on.
For Klein, the biggest value of automation is not removing the underwriter but removing the least efficient parts of the job. “We are not replacing the underwriter,” he said. “We think the underwriter still plays a critical role in this process.”
In his opinion, what needs to change is the amount of time underwriters spend on manual review. “They spend way too much time on inefficient work that honestly is best served by a machine,” Klein said. “What we’re trying to do is empower the underwriters by better augmenting their workflow and accelerating the path to decisioning, but ultimately, they’re the ones who make the decision.”
That balance, he said, is central to how Loanlight approaches AI. “I personally don’t believe that a full automated AI system is ready for a mortgage underwrite,” Klein said.
Currently, the company’s tools are focused on “extracting information from documents, synthesizing them, running them against rules, presenting facts and findings to the underwriter,” who then uses that information to move the file forward. “We are building our solution around the underwriter knowledge and an underwriter’s workflow,” he said.
Klein also pushed back on the industry’s tendency to confuse flashy demos with production-ready systems. “You could build something really slick in a couple of days,” he said, “but ultimately, when rubber meets the road, when you actually put your product into production, everything breaks.”
The real challenge, he said, is “productionizing them in the messy world, specifically around non-agency.” That is why Loanlight spent months with design partners using production data before going live.
Looking ahead, Klein said AI will continue to reshape underwriting and lending operations, but in stages. “I think we’re going to see a transformation from humans doing 90% of the job to humans doing the last mile,” he said.
Full end-to-end automation may eventually come, but for now, he said, the priority is understanding where AI, automation and traditional software each fit best. “What a lot of folks in the ecosystem want is they want better tools to help them do their work better,” Klein said. “Whether that’s AI, whether that’s automation, whether that’s good old-fashioned software.”
