According to HousingWire, mortgage rates have continued their upward trend, with the 30-year fixed-rate mortgage hitting 7.03% as of late September, marking a steady climb over recent weeks and a significant jump compared to a year earlier. The Mortgage Bankers Association has responded by lowering its forecast for single-family mortgage originations as Treasury yields remain elevated and inflation stays stubbornly high.
Technology vendors are making bold claims about their ability to cut costs, with some companies reporting origination expenses below $150 per loan or per-loan profit increases in the hundreds of dollars range. However, across the broader mortgage industry, the actual cost to originate a loan remains stuck around $11,000, suggesting that technology investments have not yet translated into widespread, industry-wide cost reductions despite significant spending in this area.
According to the analysis, compensation makes up roughly two-thirds of the direct cost to originate a loan, while technology accounts for only about 4% of that cost. This ratio has remained virtually unchanged over more than a dozen quarters, leading industry observers to question where the promised technology savings are actually materializing.
The real culprit, according to HousingWire's reporting, appears to be a lack of disciplined business processes. Mortgage lending operates with significant variation in how different firms and even different employees perform their work, and many lenders fail to fully adopt the automation tools they're already paying for. This inconsistency means employees often work around the technology rather than integrating it into standardized processes.
The reporting highlights emerging AI systems that are beginning to show real results in production deployments. These platforms don't just move documents around faster—they actively read files, balance closing documents, clear conditions, and identify exceptions for human review. One major lender's implementation resulted in quality control review time dropping dramatically and closing team throughput nearly doubling, all without hiring additional staff. If these gains were extended across the entire industry with an AI-native operating model, the analysis suggests origination costs could fall to around $8,090 per loan, though sales compensation would represent a structural floor that technology alone cannot address.
What I am seeing locally in the Bay Area and East Bay is that lenders who commit to both technology and disciplined processes are positioning themselves much better for the tough lending environment ahead. The challenge isn't that the tools don't work—it's that many firms haven't figured out how to make their people actually use them consistently. For buyers and sellers, this matters because lenders who get serious about operational discipline and AI adoption will be able to compete more effectively on rates and fees, which should eventually improve pricing in our market.
