Lenders like Rocket Mortgage and UWM have reshaped the mortgage industry. Their massive tech stacks and national footprints allow them to manufacture loans at incredible speed and scale. The rest of the industry has responded by layering automation onto existing workflows in an attempt to keep pace. But the challenge isn’t just about speed. It’s about structure.
Most lenders are still running a race on a linear track, moving loan files step by step through application, document collection, processing, underwriting, and conditions. No matter how fast each step gets, the model remains fundamentally sequential, and that means friction.
Smaller lenders and regional players can’t win that race by running faster. But the good news is: they don’t have to. AI opens a different path: a nonlinear one.
Why AI enables a nonlinear process
In traditional workflows, underwriting is too time-consuming to happen continuously. It can take a human underwriter six hours to complete a full review, often spread over two or three days due to competing priorities. Compounding the problem is that, per STRATMOR’s 2024 Peer Group Roundtable, the average retail lender’s underwriters touch files 3.8 times between application and close.
The alternative is to hold the file, gather every possible document, and aim for a “one-touch underwrite” to avoid doing the same work twice. That means the borrower waits for days, too.
AI changes that. Because underwriting with AI can take six minutes, not six hours, you can afford to do it repeatedly. Lenders can underwrite as they go, every time a document is uploaded, a fact changes, or a question arises. That instant feedback loop keeps the process moving without delay and without frustrating the borrower. It doesn’t matter if it takes 100 touches; you still close faster.
Just as important, AI enables true parallel processing. While one part of the system calculates income, another simultaneously checks its impact on asset thresholds, appraisal logic, or large deposit flags. Tasks that once required multiple specialists across days now happen together, in seconds. Every time the system receives a new piece of information, it runs hundreds of validations automatically. Income, employment, assets, credit, appraisal conditions, and more are all checked at once.
This is what makes a nonlinear process possible: decisions don’t have to happen eventually. They can happen instantly, repeatedly, and holistically. And when they do, the structure of mortgage manufacturing changes.
The productivity paradox
Mortgage technology has never been more advanced. Digital portals, e-signatures, robotic process automation, and machine learning are all in play. Yet productivity is moving in the wrong direction.
According to the MBA’s Q1 2025 Mortgage Bankers Performance Report, retail lenders now close just 3.85 loans per fulfillment employee per month, a 54 percent drop from pandemic-era highs. Meanwhile, production costs have risen to $12,579 per loan, with personnel costs making up more than three-fifths of that amount. Net production has fallen to -$28 per loan, the tenth quarter of negative profitability since the beginning of 2022.
It is a paradox: more tools and automation, yet slower throughput and lower profitability. That’s because most automation has focused on speeding up individual steps without changing the structure of the process itself. Bottlenecks like underwriting and document remediation persist, and the industry is still trying to go faster on the same assembly line. Meanwhile, human underwriters continue touching files more often, not less.
Step off the assembly line
The real opportunity lies in collapsing that line. When underwriting becomes a continuous, real-time capability instead of a fixed phase, everything changes. Errors get resolved upstream. Files stay clean. Borrowers move forward without long gaps in communication or repeated document requests.
This isn’t just better for operations, it is transformative for sales. When originators receive underwriting feedback in real time, they don’t need ten years of experience to structure a deal. They can engage a broader range of borrowers, from first-time buyers to self-employed applicants to those seeking niche or government loans.
For smaller lenders, this shift is existential. You can’t outscale the giants, but you can outmaneuver them. Instead of waiting to fix problems, you can prevent them. And instead of pushing files forward, you can work in all directions at once.
That’s how mortgage lending moves forward.
Theo Ellis is co-founder and CEO of Friday Harbor.
This column does not necessarily reflect the opinion of HousingWire’s editorial department and its owners.
To contact the editor responsible for this piece: [email protected].
English (US) · 