At this point, pretty much anyone can put together a voice AI that says something over the phone. Building the agent is the easy part. The only question that matters is whether it delivers a better, faster, and more inexpensive experience for the borrower. At Sela, we build voice agents purely for mortgage origination, and over the last 12 to 18 months we have been able to prove that at scale. We run our agents head to head against human loan originators and against other voice AI systems, measure lead-to-origination rates on the same kinds of leads, and come out ahead. That test, not a polished demo, is what should decide who gets your business.
The advantages show up in specific places. The most obvious is persistence. AI is better at dialing, better at call timing, and better at follow-up, whether you are a single broker or a team of 800 loan officers. We have also seen surprisingly large gains from models that predict the best next time to call each borrower. A second advantage surprised me, which is there is a set of borrowers who respond extremely well to AI: the chatty ones who ask a lot of questions and want to take their time. The average loan officer does not have infinite patience for that, but an agent does, and those borrowers get better outcomes at their own pace.
The third advantage is control. Our customers can see what is actually happening on every call, which helps with compliance and with performance. If you want to run a promotion, it just happens, whereas getting every loan officer to pitch it the right way is nearly impossible. Experiments are easier too. You give the agent two sets of instructions and see which one performs better, which is a very different process from retraining a team of people. Our agents now go well beyond top of funnel. They pull credit through an API, trigger HELOC offer selection steps, walk borrowers through parts of the application, and check in when someone falls off.
The lesson I have learned about adoption is that AI makes great demos, and it did even before voice was ready for prime time. What separates successful deployments is a clear, quantitative definition of success before we start. Where will the return come from? How many extra originations can we expect based on prior deployments? How does that compare to the current process, or to working on a different use case first? When both sides agree on those answers, we set ourselves up to win. When a lender simply feels they need AI and cannot say how it will add to the top line, the work gets harder. We have grown to six of the ten largest independent mortgage banks, with the sixth going live just a couple of weeks ago.
My view of where this goes is not that loan officers disappear. I would be surprised if the winning model in three or five years contains no loan officers. I expect much of origination and file building to be automated, but I think the realtor relationship in distributed retail will stay human, because realtors are not going to hand their clients to a process that never touches a person. Buying a home is also a scary and stressful process, and our data shows places where a loan officer joining at the right moment adds reassurance and lifts conversion. Top loan officers will use the AI as a backbone that runs the process patiently, then step in to build trust and keep things on track.
The surprising consequence is that this technology could level the playing field instead of tilting it toward the biggest companies. We make over 600,000 calls a day and see almost 10,000 new LendingTree leads daily, a scale only the largest consumer direct operations could reach before. We have effectively built a call center in a box. That puts the power of those operations in the hands of a much smaller team, and perhaps even an individual loan officer.
The future belongs to the loan officer who treats AI as horsepower, not a replacement. Let the agent handle the dialing, the follow-up, the endless patience, and the first stretch of the process. Spend your own time on what an agent cannot do as well: the trust, the reassurance, and the realtor relationships. A great loan officer with that leverage can originate ten or twenty times as many loans, and spend every hour on work that creates real value.
Nate Becker, CEO at Sela
80,000+ mortgage professionals get this every weekday morning.
By submitting this form, you are consenting to receive marketing emails from: Chrisman LLC, 1930 Village Center Circle, #3-3932, Las Vegas, NV, 89134, https://www.chrismancommentary.com/. You can revoke your consent to receive emails at any time by using the SafeUnsubscribe® link, found at the bottom of every email. Emails are serviced by Constant Contact