11.10.26 Talent Movement; Figure’s Alysse Guitar and Michael Tannenbaum on AI Assistants; Bonds Preempt Fed

Robbie Chrisman Welcome to the Chrisman Commentary, Daily Mortgage News Podcast. I'm your host, Robbie Chrisman. Topics on today's episode include migrational shifts within the mortgage industry, why the bond market is ahead of the Fed, as usual, and my interview with Figure's Alysse Guitar and Michael Tannenbaum on using AI loan officer assistance to reduce abandoned applications and help loan officers focus on sales and closing. Here, take a listen to a little preview. Philosophically, I mean disruption's kind of in your blood, Michael. People have been trying to disrupt the mortgage process for decades now. The people have been trying to disrupt the real estate agent from the process, that hasn't quite worked. They've tried to disintermediate the LO from the process that hasn't quite worked. Why is your the philosophy now at figure we're here to supercharge LOs rather than trying to replace them? You speak to that. Alysse Guitar One thing we found that we were not necessarily surprised by, but we were happy about is in our own testing, the best performance was with the loan officer and the AI working together. So I'm the first to tell you if the LO didn't perform, we'd have to rethink what we were doing. But truthfully, adding an LO makes it a little bit better. Adding the LO and AI together makes it a lot better. So we're just referencing our own data and the performance. And we've been able to see that the two together just supercharges things. And so we want to take this as an opportunity to empower loan officers to focus on sales and not waste their time telling somebody how a HELOC works, you know, leave that to the AI. And then you have the opportunity to find a business, uh, actually sell and find the best opportunities for your clients. Michael Tannenbaum I honestly wasn't surprised by that. If you look at similar financial markets where there are professionals providing either sales or advice like insurance brokerage or wealth management, you've been able to get an insurance an insurance policy without a broker for years. You've been able to use something like a robo advisor Wealthfront for years. So there's always been options for at least the since the beginning of the internet, there have been options to remove a salesperson and yet the vast majority of people want a wealth advisor, a stock broker, they want an insurance broker. And those roles haven't changed despite the technology. So it doesn't surprise me that in the next wave of technology, people are still looking for some level of advice and guidance, especially in what is almost always the top or among the top financial transactions of their lives. Robbie Chrisman Thanks to this week's podcast sponsor, Truework, the checker company. The one verification solution to replace in-house waterfalls. Verify any borrower with a VOIEA solution that automates the entire process to quickly deliver the most accurate and complete reports with broad GSE coverage. To learn more, visit Truework.com. I found a 95-year-old painting in my attic. I asked him to leave. Humor aside, many people have been asked to leave places like Fannie Mae, Freddie Mac, Go Mortgage, Honity, and others. It isn't surprising, and the reductions are in fact a sign of the times given the time of year and the rate environment. That said, finding success elsewhere might just be what happens for those cut, and other lenders are using sources like Redder, MLOop, MMI, and Model Match, that list is by no means exhaustive, to take granular looks at LOs and their fundings, extensions, price concessions, and profitability. A broad brush statement like, I'm calling on real estate agents to find business may not be the best path forward. One experienced LO wrote to me saying, quote, any originator needs to find out which agents are actually doing business and focus on those. NARP puts out regular stats about its members, that could be a good place to start. End quote. Broad brush is not the current trend in residential lending where precision is now possible for many facets. Switching years slightly, let's go through a couple of the main factors driving mortgage rates. Despite stronger than expected employment growth in October, worsening long-term unemployment, and historically low labor market turnover point to underlying economic weakness. Investor sentiment remains heavily influenced by AI-driven equity volatility and potential regulatory changes following the Democratic midterm sweep. Persistent housing affordability challenges affect first-time and repeat buyers alike. Treasury yields remain relatively stable, with markets currently pricing in about an 80% chance of a December rate increase from the Fed. Ultimately, expectations for further tightening remain driven more by the Fed's long-term commitment to restoring price stability than by any single inflation report. Economic uncertainty and safe haven demand may continue to limit upward pressure on long-term treasury yields. U.S. Treasuries posted mixed results to open the week, stronger equities in weak three-year auction demand pressured shorter maturities, while longer-term bonds remained resilient and received a modest late boost from reports of a potential U.S. Iran agreement, easing geopolitical tensions. For the first time since February, declining real yields, a narrowing Treasury two-year to three-month funding spread, and widening credit spread simultaneously signal less favorable mortgage servicing rights valuation conditions in October, with these trends continuing into November. Meanwhile, limited refinancing incentives and declining purchase activity are constraining mortgage production, increasing the likelihood that monthly gross MBS issuance will remain below $100 billion. Despite these headwinds, the US MBS market last week posted its strongest weekly excess returns since early August, as lower volatility and improving investor demand brought year-to-date returns back nearly to breakeven. 30-year MBS, particularly UMBS and Ginnie Mae securities, with 5% to 6% coupons, led the gains, while seasoned Ginnie Mae pools and select higher balance securities continue to offer attractive relative value. In the broader fixed income market, declining oil prices, easing inflation expectations, concerns over AI-driven economic growth, and post-election political gridlock have supported safe haven demand for treasuries, helping contain yields despite persistent inflation concerns and the Fed's hawkish stance. For today's interview, I wanted to welcome back to the show Figure's Alysse Guitar and Michael Tannenbaum to talk about using AI loan officer assistance to reduce abandoned applications and help loan officers focus on sales and closing. So while some companies are trying to use AI to remove the loan officer from the lending process entirely, Figure is taking a different approach that actually supercharges them. As a lot of people know, U.S. homeowners are sitting on over $35 trillion in home equity, yet across the industry, fewer than half of HELOC applications ever cross the finish line, costing lenders upwards of $10 billion a year in abandoned revenue. It's not from a lack of interest, it's the sheer operational headache salespeople face while getting borrowers through the pipeline. And traditionally, LOs haven't targeted stalled customers because it can take hours to unblock them. LOs aren't meant for unblocking, they're meant for sales and closing. So forcing a LO to spend an hour explaining to an applicant how to photograph their ID clearly, find their tax returns, or reset a password all for a loan not to close represents losing economics. Because of that, most lenders simply leave their stalled applications alone because the economics don't work, whereas AI has infinite time and patience to help a borrower take a better photo of their ID. So that's where Figure's new AI tooling comes in. I want to bring in Michael and Alysse. You at Figure recently launched an AI tool that targets abandoned loans. Why did you choose loan abandonment specifically? Michael Tannenbaum Thanks, Robbie, for having us both here. Excited to be with you as always. For Figure, we're always looking to invest in our partner success. We're a company that's built on the 500 or so relationships that we have with IMBs, with depositories like banks and credit unions, with fintechs. And these companies use our technology to originate loans. As you mentioned in your preamble, there's a bit of a quiet crisis within mortgage of loan abandonment. Leads are precious right now. People are increasingly wanting to make sure, originators especially, that uh leads that that come in are closed because it's a more difficult environment. And in HELOC is one of the growth areas. And so for us as a company that specializes in HELOC, we thought, well, there's this technology with AI, this big change. How can we use this trend and this new technology to add the most value to our partners? And that was the framework through which we started this endeavor. Alysse Guitar And I also saw a big opportunity in the differences across partners and across channels in conversion. And so I thought that it was an opportunity to level set across these different groups and test into some new opportunities. And we've seen some really great growth. Robbie Chrisman I know you both actually wanted to give listeners a little chance to hear what this sounds like. So let's play a preview clip for them. Sam, Figure's AI assistant You've made a ton of progress today. You selected your offer, got through the application steps, and uploaded your income documents. You're really close now. Just wait for that income verification to process and then you'll be able to keep moving forward. If you run into any issues, just use that portal chat or call figure support at 888-196388. Is there anything else I can help you with right now? Yeah. Are you a real person or are you an AI assistant? I'm Sam, a virtual assistant with Figure. Not a human, but I'm here to help you through the process. If you'd rather have a licensed loan officer call you back, I can arrange that. But it sounds like you're in really good shape and almost done. Is there anything else you need help with? Wow, I'm impressed with you. That means a lot. You did the hard work today. You got through a lot of steps. Just hang tight for that income verification to process and you'll be on your way. If anything comes up, don't hesitate to reach out. Robbie Chrisman Thank you, bye-bye. What have the initial results been? Do you share any of those? Alysse Guitar The big one that we're really excited about is uh when a loan officer and the AI work together, we see a 143 percent increase in conversion from approval to fund. Michael Tannenbaum The results are huge. I think from the perspective of our 500 partners, again, they're really concerned with how do I actually use AI to make a difference in my business? There's so many vendors out there saying that, oh, we have this AI, we have that bot. And as Alysse mentioned, we've specifically focused this initiative on conversion because that's the optimization that we're trying to drive from our partners. Whereas a lot of AI, it's about volume. And as we know, there's this concept within AI called slop, right? We see it all over LinkedIn, we see it everywhere, which is that when the cost to produce, quote, intelligence, words, voice, anything goes down because it's the AI's time, not yours, the quantity goes up. But that quantity isn't necessarily quality. We didn't focus on quantity here. We focused on conversion. And we were able to use the significant amount of figure data to train this loan officer assistant, right? We've done almost 30 billion of loans through our platform since we've been uh founded in 2018. And we were able to leverage that to actually train the assistant to be a better loan officer assistant with our partners. And one of the reasons we were able to do that is because of how fast and efficient our own process is, right? People probably on the audience are familiar with the figure platform and they know we're known for speed and efficiency. We don't have the same traditional level of contingencies that a we don't have underwriters, right? We have a much smoother, integrated capital market and loan origination system. And the bot was trained on that process, which is just inherently easier, faster, and quicker. And therefore, the intersection between an already fast process and an AI trained on conversion has led to this great result. Robbie Chrisman Yeah, at least you threw out the 143% gain. Michael, you threw out the tens of billions of dollars that have been originated on the platform. What do the results tell you about the potential here? Alysse Guitar I think we're just scratching the surface at the opportunity. I have other stats. We actually increased average loan size, we increased the uh number of dollars originated per funded loan, the whole process at every stage. We saw gains, we saw improvements. And this is just on our original test. You know, we're continuing to grow the number of use cases we implement this tool on. Uh, as Michael mentioned, we're rolling it out to wholesale and private label partners. So there's a huge opportunity here to not just improve conversion and improve the economics, but actually make loan officers more effective. Robbie Chrisman So then it begs the question: what are the plans for rolling it out among your partners? Over 500 already. Holy smokes, that's a lot of partners. Nice job. Thank you. Alysse Guitar So we're live with all of our wholesale partners. We're really excited uh to see the results of this and share that data out broadly. Um, we're starting to have conversations with our private label partners as well to get that scoped and built. Michael Tannenbaum Yeah, this was actually something that we, in hindsight, probably should have given to our private label partners first. We thought that historically with Figure, we've been careful and treat our private label partnerships as the valuable relationships, which they are. And so we're always looking something like AI is new. And our thought was let's test this in the wholesale channel, right? On in the broker channel. But what we've heard from existing partners is they've seen what we're doing. And it's probably worth mentioning in this context that we are partnered with OpenAI's uh chairman, Bret Taylor. He has a company called Sierra, and that's who we worked with to build this. And because of the excitement around that company, him, as well as the results that we've been sharing, we've had a lot of partners come to us and say, I'd love to get this right now. So we are rapidly rolling this out, as Lise mentioned. It's live today in the wholesaler broker space, and we're going to be rolling it out among our private label and white label partners very quickly. It's it's something that everybody is clamoring for. And again, it comes back to where we began this conversation, which is how do I one deal with the loan abandonment? And two, I'm looking for figure. We sort of call we sometimes use this phrase the figure factor, which is how can we at figure help our partners grow their business? And as an example, in 2025, looking at HMDA data, partners that adopted Figure grew their home equity business more than two and a half times those that didn't work with Figure. And you can see that this figure factor is already manifest in the very technologically savvy process that we bring. But then adding AI to that is only going to increase that disparity. So we're really encouraged by what we see and excited about working with more and more partners to roll this out. Robbie Chrisman Like if I go buy a car, I don't want my salesperson from the car going and wrenching on it and changing the oil. You know, it's just exactly right. Alysse Guitar Um, these agents have infinite patience. And so they're I've listened to an agent have a 40-minute conversation where it's mostly dead air with the borrower just digging, looking for their ID. We don't want to pay a person to do that. A person doesn't want to do that. That is just maddening, but the AI agent is happy to do that. Robbie Chrisman I'm sure there are LOs and organizations listening to this right now, and they're still ambivalent, skeptical, sick of AI slopper, overwhelmed with the amount of AI chatter there is right now. What would you say to those folks? Michael Tannenbaum I'd say that look at how much of an innovative experience Figure brought to the home equity. It's a absolutely a wow moment anytime someone uses our platform. And that's what you can come to expect from Figure. It's valuable technology that saves you time and saves you money. And we built AI, and in particular, this AI loan officer assistant from that framework. So you should expect more from us there and ignore whatever slop you see elsewhere, because that's what Figure stands for. Alysse Guitar And I'll add that, you know, from a boots on the ground perspective, figure loan officers are only talking to borrowers that are warm and ready to go. They're not leaving voicemails, they're not having to make outbound calls, but they're still getting their commissions and doing their business. And so that's the opportunity for this AI. Michael Tannenbaum Thank you both. Thank you. Robbie Chrisman Today's economic calendar is already underway with October and FIB small business optimism that declined slightly to 97.8. We've received CPI, which came in at 0.4% month over month, slightly above forecast, but still below the previous reading. Core CPI came in at the previous readings level of 0.3% when it was expected to be at 0.2%. Nothing earth shattering here. In addition to the CPI data, Fed commentary and continued equity market volatility will be critical in determining whether the recent Treasury rally and improving mortgage performance can be sustained. And after yesterday's $58 billion three-year treasury auction delivered mixed demand with a higher than average yield and weaker bid-to-cover ratio offset by strong non-dealer participation and elevated direct bidding, today brings $42 billion of 10-year Treasury note auction results. We will also receive Fed remarks from Waller, Logan, and Schmid. After the latest inflation data, agency MBS prices are slightly worse than Monday's close, the two years yielding 4.61 and the 10 years yielding 4.89 after closing yesterday at 4.86%. Let's wrap up with a joke and some housekeeping. A wife tells her shiftless husband to get a farmbread turkey for Thanksgiving. He brings home a turkey, but she finds a shot in it. I told you to get a farmbread turkey, not a wild one. That is farmbread. Well, where'd this shot come from? I think the farmer was aiming at me. Thanks again to Truework, a checker company, for sponsoring this week's podcast. Truework is the one verification solution to replace in-house waterfalls. You can verify any borrower with a VOIEA solution that automates the entire process to quickly deliver the most accurate and complete reports with broad GSE coverage. To learn more, visit Truework.com.