According to HousingWire reporting on a recent STRATMOR Group study, mortgage lenders have made real strides in using artificial intelligence throughout the lending process. The research shows that nearly seven out of ten lenders are now using AI to organize and categorize documents, while around six out of ten use it to read those documents, and approximately half rely on it to evaluate borrower income during the underwriting phase. These numbers tell us that lenders understand where AI can add value and are actively putting it to work.
Here's where things get interesting, though. Having AI read a document at one stage of the process doesn't automatically make your entire loan file flow smoothly from start to finish. If that classified and analyzed information can't move seamlessly into the next step, you end up with lenders manually re-entering or double-checking the very data that AI just processed. A document might be correctly sorted when the borrower first applies, but then someone has to type that information in again when the file reaches underwriting. That kind of disconnect defeats much of what the technology was supposed to accomplish.
The reality is that many lending operations still run on systems that don't talk to each other very well. Your point-of-sale system, origination platform, underwriting tools, and document management software often exist as separate islands of technology. This fragmentation creates real problems down the line. By the time a file reaches closing, you might have two different versions of the same data floating around, one processed by AI and one entered manually. If there's an error, it might not get caught until after closing, when regulators or investors review the file and discover a data problem.
The manual checking at these system handoff points doesn't disappear just because technology improves. Staff still need to verify information where systems stop connecting, and they have to do it because compliance requirements like TRID, RESPA, and HMDA rules demand accurate, traceable documentation. When borrower information keeps getting re-entered between different platforms, maintaining a clear audit trail becomes harder and compliance risk goes up. One lender, Dearborn Bank, showed what's possible by connecting its point-of-sale, origination, and document processing systems through integrated platforms. The result was faster applications and the ability to handle more loans without hiring additional staff.
The path forward isn't about ripping out everything and starting over. According to the reporting, lenders should be looking at how to connect and modernize their workflows piece by piece so that intelligence created at one stage actually carries through to closing and beyond. A borrower's information should move consistently through the process without being re-entered manually at every transition point. The investment lenders have already made in AI for documents and income analysis is real, but that value only fully materializes when the systems themselves are connected.
What I am seeing locally is that Bay Area lenders who are investing in these connected systems are positioning themselves well for the future. When you can take information captured early in the process and move it cleanly through underwriting to closing without manual intervention, you're not just saving time, you're reducing errors and compliance headaches. That efficiency translates to faster closings for our buyers and sellers, and that matters in a market like ours where speed and accuracy both carry real value.
