Mortgage Lenders Build Variable Cost Models With AI Automation

Mortgage lenders are rethinking their operating models to escape the industry's boom-and-bust hiring cycles, using AI and automation to create cost structures that flex with market conditions rather than forcing mass layoffs during downturns.

Phil Shoemaker, CEO of The Loan Store, and Francesco Paola, Chief Growth Officer at Mozaiq, outlined an approach that prioritizes sustainable scaling over rapid headcount expansion. Their strategy centers on building variable cost structures through automation while preserving the human relationships essential to mortgage lending.

Why it matters

Mortgage lenders traditionally respond to volume swings by hiring aggressively during refinance booms and cutting staff when rates rise. This pattern creates workforce instability and operational inefficiency. A variable cost model built on automation offers an alternative: stable core teams supported by technology that scales capacity up or down without the human cost of layoffs.

Foundation before AI deployment

The Loan Store began its automation journey with basic process improvements before AI became an industry buzzword. Mozaiq started by implementing document indexing and data extraction to clean loan files, enabling processors and underwriters to make faster decisions with better information.

This foundational work proved critical. When AI capabilities matured in 2023, the lender had the infrastructure and clean data necessary to deploy more sophisticated tools immediately.

Process redesign comes first

Both executives emphasized that technology purchases alone do not deliver results. Shoemaker noted that many management teams sign contracts for new tools but fail to engage deeply in implementation and process optimization.

"If you're not highly engaged in figuring out how to optimize that process before you apply the technology, you usually end up with a pretty bad result," Shoemaker said.

Shoemaker took this principle to an extreme by spending three months working as an underwriter to understand the processes he wanted to automate. He described it as the best career decision he made, providing insights impossible to gain from executive briefings.

Paola reinforced that AI implementation requires different thinking than traditional technology deployments. Organizations must reconfigure processes and manage change, not simply install software.

The accordion workforce model

The result is what Paola called an "accordion workforce"—a stable team of employees who understand the business deeply, supported by technology that expands and contracts capacity based on market conditions.

This model keeps variable costs low during slow periods while enabling rapid scaling when refinance volume returns. For employees, the benefit is job stability rather than the constant threat of layoffs that has characterized mortgage lending for decades.

Human oversight remains essential

Both leaders stressed that automation should augment rather than replace mortgage professionals. Shoemaker argued that the emotional significance of home purchases requires human involvement, even as AI makes processes more efficient.

Paola said AI should provide recommendations, alerts, and warnings while humans maintain final decision authority. "AI should never make credit decisions, not at this stage. And in fact, maybe never," he said, emphasizing the importance of maintaining a clear source of truth.

Competitive differentiation through service

Looking forward, both executives see competitive advantage going to lenders who successfully balance automation with relationship management. In a commodity market, service quality becomes the primary differentiator.

Shoemaker predicted that the strongest lenders would be those integrating AI while maintaining the human aspects of mortgage lending. Paola agreed, noting that variable cost structures cannot come at the expense of customer and broker relationships.

These details were first reported by HousingWire in a conversation with Allison LaForgia.