Opto Sessions – Invest in the Next Big Idea

AI in Lending: Real Innovation vs Hype

• Haydn Brain & Ed Gotham

Share episode

Leah Price, VP of the Tinman AI platform at Better.com, joins OPTO Sessions to unpack the company’s pivot to a full-stack GenAI mortgage OS and Betsy, the chat- and voice-based assistant embedded across the experience - delivering up to 75% faster cycle times, 45% lower fulfilment costs, and 10× volume, plus a first look at a pre-release avatar advisor.

-----

The content in this podcast is for informational purposes only. Opto Markets LLC does not recommend any specific securities or investment strategies. Investing involves risk & investments may lose value, including the loss of principal. Past performance does not guarantee future results. Investors should consider their investment objectives and risks carefully before investing. The information provided is not an endorsement of this product and is for information and/or educational purposes only.

soon you'll be able to speak to one of our avatars in the middle of the night. You're nervous about your mortgage application. You want to have a FaceTime call with an expert mortgage advisor. You can do that. And I'll tell you, when I go out and demo this software, it's magical. The demos speak for themselves. People's jaws drop because they just, they've never been shown something like this. So I'm going to say here that better is leveraging Gen.AI tools in a way that no other lender on the market is able to leverage. this is like the opportunity of a lifetime. So watch this space. There are amazing things to come. Hello and welcome to Opto Sessions, everyone. Today we have Leah Price on the show, Vice President of the Tinman AI platform at better.com. is the leading AI powered home finance company and Leah leads the commercialization of Tinman, bringing Better's proprietary technology platform to lenders nationwide. Tinman is an AI driven rules based decision engine that helps lenders and brokers automate underwriting, reduce cycle times, by up to 75%, lower fulfillment costs by up to 45%, and manage as much as 10x more volume, allowing them to serve more customers faster and more efficiently. Previously, Leah led the Office of FinTech and the Office of the Chief AI Officer at the Federal Housing Finance Agency and held senior roles at Figure and Fannie Mae. She brings deep expertise in both technology and regulation to her leadership of Tinman. It's great to have you on the show, Leah. How are you doing today? Really good, thank you so much. Where are you calling from? I am on the 80th floor of One World Trade in New York City. I think you can see behind me the background a little bit. How long has it been in those offices? We've been downtown for a couple years. This office, I think it's um only like one year, but it's spectacular. It's an incredible place to come to work every day. That's awesome. And yeah, like I said, it's great to have you on the show. You recently joined Better, I believe, in June. I just thought it'd be good to start, just learn about what your vision for Tinman AI platform is, specifically what areas Tinman is involved in, and any early initiatives that you might have set in motion already that are working towards that vision that you can outline. yeah, so let me take a step back. Let me kind of start at the beginning, just in case your listeners aren't familiar with the whole story. um when I, earlier in my career in the mortgage industry, I worked at Fannie Mae, and one of the most tech forward mortgage lenders at the time was Better.com. They were always participating in pilots. They were always seen as an industry leader, really pushing the envelope. They were the go-to for any innovative tool we wanted to test out in the market. So that's when Better first came on my radar. uh So fast forward a couple years, I ended up working as a federal regulator at FHFA. As you mentioned, FHFA, for those who don't know, is the regulator and conservator of Fannie Mae and Freddie Mac. And there I was leading the office of FinTech, uh basically taking a bird's eye view of everything that was going on in FinTech and AI in the mortgage industry and assessing where we were. ChatGPT had launched a year before and the mortgage industry seemed pretty far behind. So the director of the agency at the time was really curious, okay, where is this going? What are the risks? What are the benefits? Are borrowers really going to see the benefit of this technology? And so about, let's see now, it would be two years ago when I joined FHFA, uh Better again came on my radar. CEO Vishal Garg and I spoke. I was curious about what they were doing with Generative AI specifically, and it was clear they were way ahead of everyone else. So I started to bring Better in with other companies as well to show FHFA what they were working on. At FHFA, we hosted this big event called a tech sprint where we brought in industry participants to uh pitch projects and better showed up really strong along with other uh lenders and technology vendors in the space. And so that's how I really got to know Vishal and the now CTO Ziggy. And I was really impressed. And so. At the time, they were really focused on being a mortgage lender, not being a technology service provider. And I had in the back of my mind the thought, you know, these guys are so far ahead of everyone else. If they ever wanted to license out their software, these guys would kill it. But that didn't happen for a little while. uh But over time, I think Vishal started to get phone calls from other lenders. asking if he would be willing to license out the software. So at a certain point, I was interested in going back to the private sector and he was starting to build out this business and we talked and the idea was, okay, the mortgage industry is so far behind in general. Here's this lender that has invested over a billion dollars in technology and AI. They can now allow other yet lenders to use this software. And I want to lead that charge. So in terms of what is my vision coming from having been a regulator, seeing how far behind the rest of the industry is, I'm really excited about leading the market as a technology service provider and offering this technology to other lenders in the industry. So you could kind of think about the Better story with Tinman. Tinman is the platform that they built for themselves. You could think of it almost like Amazon built their software for themselves. And now we think of Amazon Web Services as the big enterprise level service provider that powers, I mean, our country, uh the world in terms of technology infrastructure. So it's a big ambition that I have and that Vishal has to penetrate. this market and we're still in early days with uh selling this software to other lenders. So we had our earnings release recently and the dominant theme of that release was this pivot from just being a mortgage lender to now being an enterprise platform service provider and I'm really excited to be a part of it. Yeah, I'm sure that does sound exciting. so the platform you describe is democratizing access to advanced AI for lenders of all sizes. In reality, what does that actually mean for borrowers and lenders? What benefits are they getting from that? Yeah. All right. So here's where you need to know a little bit about uh the mortgage industry and the tech, uh the way the, the, the players are. And I don't want to assume that you know anything about that. So um in our industry, about 85 % of lenders are on one large technology service providers platform. That's huge market share. And this platform was built in the 1990s. And some of these people are my friends. They're really nice people. I've worked with them closely over the years, but they're really bogged down. It's a mainframe system. They're still doing their cloud migration. So our industry, it's already post a great financial crisis. Already it was uh very much penalized. There's a lot of regulation. It's a, uh there's very robust risk management frameworks from the regulators and also internalized. But when you look out to the rest of the world, there's all of this technology investment. There's so much excitement around other industries that the mortgage industry really hasn't seen any of that benefit yet. We're lucky at better over the past 10 years to have made that investment. So this big dominant player that has 85 % market share, they simply can't offer the same type of uh solution that we can. And let me explain in a little bit more detail what I mean by that. And I'll get to why this benefits lenders in a moment. uh So Better built its own technology platform. It's an end-to-end system. So typically, a mortgage lender of our size, they would use like eight different pieces of software. So one for the borrower to have their front door. So what is the UI that the borrower uses? The POS, the point of sale. Then there's the loan origination system. That's the... uh That's where all the data remains. That's where underwriters and processors and loan officers work on a mortgage application. Then there are different data providers. There are document manufacturers. There are compliance engines. Then there is a matching engine to investors in general. So there's a pricing engine because what happens when you make a mortgage, a borrower provides all of their credit information, their data, income and assets. Then you have the property information. So that's the collateral. And then you have the investors on the other side who buy these mortgages once they're funded. And so in order to do that, typically it's like eight different pieces of software that are strung together. So why that doesn't really work anymore. And the weakness there is starting to come become very clear in the world of gen AI. So let's imagine like in the world of Gen.ai, we have all of these different chat bots and chat interfaces. have chat GPT. Today, if you want to, if one vendor of those eight wants to leverage Gen.ai, let's just say put a really slack slick chat bot on one of those. It can only work on the POS. It's really hard to have that same chat bot have access to everything that's in the pricing engine and also have access to everything that's in the LOS. Like forget agentic systems, forget like context awareness across all of those systems. You can't do it better having built its own end to end platform. So Tinman includes point of sale, loan origination system, compliance engine, decision engine, and pricing engine. We are very well positioned. We were very well positioned to layer a gen AI tools on top of that. So. Chatbot, just layered on top. Voicebot, we have Betsy, which is the name of our AI assistant that is available in voice and chat. Betsy, Better that's why we named it that. So I just laid out the differences between the the paradigm is for most lenders versus the way we've built it out. So on the borrower side, You asked me what are the benefits on the borrower side and on the lender side. So the borrower benefits in this world now because they have more access to different front doors that they might not have access with fragmented platforms. So that means if I'm a borrower, um if I work nine to five, I can't use a user interface. um I want to do a voice call. I can just use voice. If I have questions that I want to use chat. I can use chat. uh If I want to, soon you'll be able to speak to one of our avatars in the middle of the night. You're nervous about your mortgage application. You want to have a FaceTime call with an expert mortgage advisor. You can do that. So that's one of the benefits of having a Gen.ai tool that's able to access end to end uh platform information about the borrower, the property and pricing options. In our system as well, I mentioned that we have our own pricing engine. So a borrower who comes into our POS has access to 40 different investor options who all offer a myriad of different products. So borrowers who come into our front door actually have more product offerings. uh There are more flavors of products that they'd be able to qualify for. So that's better for the borrower. It's also much less expensive for us to be able to manufacture a loan. In addition to those Gen.ai tools I talked about on top that our borrower facing, there are also all kinds of ways we leverage AI within our platform that cut costs that we have to pay to loan officers, underwriters, and processors. So in our system, loan officers, uh they can handle three times as much volume as they can with other lenders. processors and underwriters can handle 10X as much volume, which just makes it cheaper for us. They can handle more borrowers coming through the front door, which makes it cheaper for the borrower in the end. Those cost savings, we always pass on to the borrower. So it's a cheaper, faster, uh more delightful experience, I like to say. So I touched on the borrower benefits. The lender. also is recognizing those cost benefits, right? So uh I talked about loan officers now, they can handle three times as much volume. That's a lot more money that they're able to make. And because Betsy, our Gen. AI tool is able to follow up with borrowers on off hours, loan officers don't have to be up at all hours of the night. We can be collecting data. We can be processing data, analyzing, underwriting in an automated fashion without relying necessarily on humans to take every single action on their behalf. So it's just continuously processing the data and the borrower's documents, which makes it faster for the lender. Our defect rates, because of all of that automation, are much lower than the industry average. So we are one third the defect rate. of the industry. uh And like we've even found while we still like humans are completely in the loop, uh even we as one of the most tech forward mortgage originators, all of our loans are touched by human beings. uh And one of the things we've implemented in September, which has been pretty magical to realize is before a human being, before a human underwriter is allowed to decline a file, We require that they also put it through our Betsy underwriting process to see if the AI tool can underwrite and figure out how the borrower could possibly qualify for a given loan to find them a product that might meet their needs. And so this is really revolutionary. I've never heard of this happening, actually having the AI do a double check and be able to potentially there were gray areas. that the human didn't know of, or potentially there was an investor product that the human wasn't aware of, or potentially there was some debt that the borrower could just pay off uh through their process that uh this becomes almost like a debt doctor that uh the underwriter just didn't have the time uh or necessarily the knowledge to be able to figure that on out on their own. So this really multiplies uh the ability and capabilities and strengthens the skill of all of the people uh that are working on the file. So again, cheaper, faster, higher quality is what we end up with at the end of the day. Yeah, thanks, Liam. I mean, it's incredible to see how many layers you touch with the AI and what the amount of opportunities there are to reduce costs and prove efficiency. Really interesting to see that in a space because, you know, it's obviously leading in that area. How big is this? Can you give us a sense of the size of the market that Tin Man is targeting in the mortgage space? Because people can picture it. Okay. So there are a lot of different ways to think about this and you have to, um like my perspective is really from uh the Fannie Mae, Freddie Mac mortgage space. So I'm really, uh I could talk about it in terms of home equity, but I'm really going to talk about uh what is the biggest segment of our consumer market. So if we think about uh originations and refinances. We're talking about two to three trillion dollars worth of originations and refis ah in a given year. And so we are the infrastructure that will power that going forward. uh If you think about how many loans that is, that's like four million loans a year. And I told you that 85 % of the market is on that old technology stack. And I'll tell you that that 85 % of the market is really frustrated with their provider. So they are anxious to get off of that tech stack and move on to something more novel. So it's a pretty significant market for us to penetrate. And I know having come from Federal Housing Finance Agency, that what we have in terms of AI and automation is best in class versus everything else out there. So it's a pretty massive opportunity, the US mortgage market in itself. And over the next one to two years, in terms of which areas have the biggest growth opportunity for Better.com, do believe it is this new partner segment where you're licensing the software for these other providers? Yeah. Okay. So think about Better.com on the recent earnings release, Vishal Garg, he talked about this portion of better.com as really being the driver for earnings growth in the future. And the thing is, is we've already built the technology. We've already ah done the investment. So now the investment and the cost is really associated with building up a team. that can sell and scale this software. uh So uh in terms of the pipeline, we've already shared publicly that our approach is to go with the biggest provider in a given vertical. Vishal says, land and expand. So we go for the big guy, we prove ourselves out with that vertical, and then we expand for there. So what you could imagine, and what we're already demonstrating, as talked about in the earnings call. We've gone ahead and we've secured a partnership with one of the country's biggest personal finance companies. We've got Tinman working there. We've landed a top five independent mortgage bank who will be deploying Tinman for their HELOC and HELOANS and we will expand from there. And then we've also got the largest the country's largest reverse mortgage provider, Finance of America, that we've worked with. So in our pipeline, you could also expect to see us working with the largest servicer in the country. We'll aim for the largest bank in the country, and that's how we'll prove it out. And I'll tell you, when I get onto sales calls there, and when I go out and demo this software, it's magical. I mean, the demo... The demos speak for themselves. People's jaws drop because they just, they've never been shown something like this. In a way, there's a lot of inertia. One of the challenges with this is that people are just resigned to using old software. Nobody comes out and says, Hey guys, there's something better. There's something that is actually like fun to use and is like, uh will make you much more efficient. So uh the pitches have gone a little bit like that. ah And so that is our expansion plan. Pick the biggest player in a given vertical, land that, prove it out, and then expand. So it's an exciting way to go about it. Yeah, so sounds like just at the start of this new growth phase, which is really, really exciting. You also mentioned the earnings call. So I thought we could touch on the market more generally at the moment in terms of sentiment towards AI stocks. How do you distinguish genuine innovation from hype or AI washing and what signals should investors look for to tell whether a company is truly leveraging AI versus just using it as a buzzword? Okay, yeah, so I'll tell you while I was at FHFA, this became very clear. AI washing is something that the SEC has been hyper aware of. Basically any company can go out there and make claims, oh, we have all this AI, it's really driving our efficiency and it's really cool and show off some chat bot that they've slapped on top. And there they are, they're an AI company. They should trade at a higher multiple. Um, and they can get a lot of attention. It's remarkable how many companies would come in and demo their AI solution to us at FHFA. Uh, and I would see it and I would be like, wow. So actually that looks like the same chatbots that have been around for the last 10 years. Can you explain to me exactly like. what is special about this right now. And a lot of times they just couldn't get very deep into it. So what I'm gonna say to all of the investors out there, you need to do the homework yourself. And it's not hard in the financial services space. So it's one thing for me to talk to you here and say, Betsy is amazing. Borrowers can talk to her 24 hours a day. The good thing is you can just go to our website. Dial the phone number, call Betsy yourself. See if she'll take your application. I want you to go to our website, better.com, start to fill out an app, and you'll see Betsy pop up. Push her. Push our AI hard. Ask her hard questions. Drill her. Why do you need my phone number? Why do you need my social security number? I already gave you my phone number. Why do you need it again? Push her. Then I want you to go to some other company websites. I'm going to give you an example. So Rocket is the second biggest uh mortgage lender, independent mortgage lender in our country. They go on their earnings calls. They talk all about AI. They master AI. They are putting a lot of investment there and they go out and talk about it. I want your investors to also go to the Rocket webpage, create an application, try their Gen.ai tool. Do a side-by-side comparison. Say, hey, I already gave you my phone number. Why do you need this? Ask them about your application. See how far you can push it yourself. You will see how far ahead Betsy is and how much uh benefit can come to borrowers versus a company like Rocket. And I'm just mentioning two companies that are out in the world talking about their AI. You could do this with any company, any financial services company that has some kind of consumer facing aspect. That's how you really need to do your homework. Do not only listen to an earnings call. I'll say on our, on our recent earnings call, Vishal insisted, our CEO insisted on webcasting and he included the demo. We want to start challenging our competitors to also do live demos, really show. what you have and how that's going to benefit borrowers and the industry. That's what I want to see. If I can get my competitors out there uh feeling like it's a little bit uncomfortable, that's when I know I've done my job and I think they're starting to feel it. So that's my advice to get through the hype versus what is real. Just do it yourself. Push, push. Yeah, that's the best kind of research is just go do it yourself rather than just reading about something and assuming it's correct. So yeah, everyone should go out there and test it themselves and really get a feel for what it's like. A good segue here is just to talk about, there's obviously a difference between rule-based automation and machine learning AI in any software product. Can we just talk about how each of these plays in Tin Man architecture? Yes. um And I've got like, I've got some, ah let's see, I've got some numbers here that I wanted to pull you up some concrete details. um But basically I want to talk about the different modes of AI that we're talking about. And I can talk about both the industry and what better is doing a little bit differently. So I'm going to say here that better is leveraging Gen.AI tools in a way that no other lender on the market is able to leverage. uh Many lenders in the industry are leveraging machine learning models for underwriting for the matching engine. An additional way that we are using machine learning is we have been able to feed our underwriting tools data that better.com has on their borrowers. in order to drive better outcomes for borrowers. So we have all of this data. We're able to train our tools to be able to really push that automation and get better outcomes. Other vendors on the market who may be launching for the first time. So let's say a newbie is like, I'm going to build a completely AI native platform that leverages all these same tools that Better has, but they're coming out of the gate. Let's say they're able to raise a bunch of capital and it's people who have uh significant experience already. What they do not have, what no other vendor on the market has is this data in-house to train their models. OK, that's good. Yeah, that's one of the key USPs of better is the data. You can't make it up yourself. yeah, I'll see when I can do it. of the selling points here is Better's platform is battle tested on $100 billion worth of loans. And we're just getting started offering that software to other lenders. So anyone else who comes on the market now does not have that history, does not have that know how, doesn't have the battle scars really uh from the experience. When I go out and pitch with my colleagues here at Better. What's amazing is that I can actually bring along an underwriter who's been embedded in the software themselves. That other big provider that I mentioned that has 85 % market share, it's just people who work on software. They're not people that are themselves improving the software uh driving forward. So it's a really cool thing to be able to go out there and pitch. Yeah. And as soon as you mentioned AI, a lot of people tend to, especially with finance, think about safety. How do we make sure that the AI is making the right decisions? Can you just touch on, you know, model governance, and compliance, especially with your regulatory background, how has that been built into? Yes. Okay. So I want to like nerd out on the regulatory stuff. Um, just for a moment with you. So there's something called the Collingridge dilemma. And I thought about this a lot while I was at FHFA, while I was a regulator. So I was going in to work at FHFA at a moment when Gen AI and was huge and the mortgage industry was really fearful about how the regulators are going to come in and regulate this. And so the mortgage industry has been in general, very paralyzed because the federal government hasn't come out clearly stating exactly what the guardrails are going to be. And if you think about Europe for a moment, it was a very different approach once once chat, chat, GPT launched in Europe, there's all these frameworks that have been put into place, like uh evaluating AI based on the risk, there's a perception that there's a lot of guardrails being put up in place before the use cases are fully understood. And so what the Collingridge dilemma talks about is regulation as a pacing problem. And just like, bear with me for a moment, because it's interesting to think about. because it's easy to criticize regulators. They're moving too fast, they're too onerous, or they're moving too slow. And it's really tricky. if a regulator, there's an arc. And it's uh awesome to think about uh Gen.ai as part of this arc. So a technology launches. At first, nobody is really sure exactly how it's going to be used. There's all this fear. I mean, in the Gen.ai case, people are afraid it's going to kill us. um But there's all this fear, uh you know, how big are deepfakes going to be? uh Is AI going to take over the world? There's all this like negative and positive hype, but use cases over time get tested and the successful ones are end up being broadly adopted. So a regulator's role, a regulator can jump in at the very beginning and say, okay, you know what? We're going to ban blanket. ban on AI that's used for X. And the truth is it's never even been tried in a sandbox environment to be used for X. So if a regulator waits too long to regulate, all these crazy use cases can come out like you lose control. But if a regulator waits, starts too soon, it can damper all growth. basically destroy value, destroy innovation for a country. And right now, the tone at the federal level is really about using AI to be competitive and to thrive. And so uh the tone, I think, is it's shifted a lot. But as a former regulator, I really think of this give and take. How do you try a little bit to explore just come to the point where you realize what some of the risks are and then you mitigate those risks. And so the way that I um lived that at FHFA was when we held a generative AI tech sprint, we brought the industry together to actually build gen AI use cases. I made sure that not only when they did the pitch, they pitched the benefits and the upside, but they also at the exact same time, pitched the guard rails. so um what is lucky about the way that Better I look, they're lucky. I'll just say they're lucky and smart. uh The way they grew up is that mortgage industry is very heavily regulated by the states and the federal government. And so everything that they've built, unlike a vendor offering, was strictly subject to all of the regulations. so that now by the time we're offering this software to the other lenders, we've been in their shoes regulated as heavily as they are. So compliance is just part of the blood. That's part of the engine. That's part of what was envisioned as a key component of this platform. Our general counsel is former CFPB, which is one of the big regulators in the mortgage industry. And that's actually what attracted her. to come to Better because of the way that they were planning on automating the compliance. She knew that that would get her ahead. So she's been here for 10 years and that dream has been realized. um But specifically, what's really cool about some of our Gen.ai tools. So for example, Betsy, the underwriter, we have, um it's almost like chat GPT kind of screen chat bot. embedded in each loan file. So specifically an underwriter can go through and say, hey, Betsy, can you help me figure out the debt to income ratio? uh Is it going to work for this borrower? Can you help me figure out what debts the borrower could help pay off? And so in terms of transparency, Betsy doesn't just say, oh, he should pay off his BMW. You can see what she's looking at. and you can see what she's processing to make that recommendation. So there's a whole different level of transparency. um I'll also say, we're going to talk about the avatar in a little bit. But while we've been testing the avatar, I've had one person talk to the avatar and then our CTO on the other side going through the code to see what's going on on the background. So we can see exactly what service the avatar is calling when she. is looking up pricing for a borrower, for example. So there's this, uh in terms of transparency, like there's next level transparency that we've, because of that requirement of better as a lender, Tinman now offers that same transparency to other lenders that vendors, just, it's not in their blood to need to offer. Yeah, thanks, Lou. It's a good overview of the regulatory environment and how it's embraced, I suppose, the AI. It would better to move things forward in a positive way with good outcomes for the end users, which is great to see. You touched on the partners and the go-to-market plan for them earlier. And I just wanted to dig into that a little bit more because you've highlighted it as a key area for better. When onboarding a new lender, What typically do they have to change? What processes to help them succeed with Tinman? And are there certain risk averse lenders that you have to approach in a particular way to help them overcome the hesitation to use AI and the software that you're providing? Yes. So the biggest hurdle, I will say, it is not internal to Tinman or Better luckily, it's really the inertia of the industry. And imagine an industry that one player has 85 % market share with very long sales cycles. When you sign a contract with them, it's for multiple years. And that also means that all of the people that are driving this industry, they've been using the same software for their entire careers. And so the biggest hurdle there, I will say, is really, it's getting the people comfortable with using slightly different tools from what they're used to. um The biggest hurdle is not the technology implementation. It's basically getting people off of what they're used to. um And so I think our strategy of going for the biggest player in a given vertical that gets a lot of attention in the industry and proving it out with the highest performer is the way that we get, we convince the rest of the industry that look, this is achievable. You're the envy of your segment is using it. You can too. makes a lot of sense. Yeah, it's just follow the leader sort of approach, isn't it? suppose everyone falls in line once the person that everyone will able to look up to starts using something. I'll say like, I sense FOMO in our industry. That's the funnest part. When I get to go out and like people are in awe of the demo and they're asking, okay, how long will it take for me to get started? The longest poll uh in the tent right now for implementation is compliance and signing the contract. Just because in most cases, uh our customers are used to signing multi-year Contracts that are very onerous and they're locked in so one um One thing I'll point out is that right now our pricing model is very disruptive No one else in the industry is doing what we're doing in terms of pricing which is we are pricing in terms of We think of it as outcomes as a service. There is no upfront cost Right now we are not charging any implementation fees. It's not a per seat cost, and it's not a monthly subscription. So those are the standard costs for any type of enterprise SaaS model. The way we price is, and I'm going to explain it really simple terms, like when a borrower comes in to our website or talk to Betsy, whatever front door the borrower gets through, a loan is underwritten and that loan funds, that is the revenue event for us. So we only charge at that revenue event for our customer. So we are a hundred percent aligned with other lenders making money. So it's a lot of investment for us to get a lender stood up. And then our goal is really to help them make money using our software. What's incredible about that is that that 100 % alignment will benefit both the partner and us. Both of our earnings will go up. And so that model is very disruptive in that 85 % market share giant that I told you. There's very little switch cost. I can go into a pitch and I can say, you know what, you're on the other guy, totally fine. You want to test this out? We'll just pick five loan officers. We can pick one product. Let's just do a home equity line of credit with those top loan officers who want to be tech savvy and tech forward. And they've seen the demo and they think they can drive the Ferrari that we call Tin Man. Just give it to them. see what they can do with this. em that is very alluring for some of these big players. So you think of that combined with our uh targeting of the biggest player in every vertical. And I think that's really where the magic will come for us. Yes, you massively reduced the risk for any new lender to get them over that barrier. Do you see that change in the future? that pricing model likely, is it probably hard to say, but is this the pricing model you foresee for the foreseeable future or will it change over time? that we'll see TBD. Like I'm so excited about the opportunity and the possibility to disrupt a market that wants to be disrupted. I don't know that we'll change that model. I love that model. I love that we're going to be a hundred percent aligned with our customers. And another thing to think about is like these big players, the other software providers, they also charge pass through fees to all of the different vendors that are integrated. So they take a cut from everything. They take a cut from, they take a big chunk and lock in their lender customers, but then they're also scooping up benefit from all of these vendors. So, and it's a lot of vendors in this process, like data providers, document providers. um So right now, I don't want to do that. I want to be totally aligned with my customers. I want to be totally aligned with the other vendors that are a part of the process so that they also see benefit from all of this volume coming towards us. I'm totally okay with my competitors hating me. Totally. hope they're driven to give better deals to lenders and to give better deals to the vendors. That helps everyone. More competition is going to ultimately benefit borrowers. em I think that'll drive better outcomes for everyone. It's interesting to see, obviously, what you're describing is this player is almost a monopoly in the space. And often, they get greedy when they're able to have the power to become greedy and people have to still go through them. But then it often becomes their biggest weakness because that's where the new players can come in and undercut them massively so that it makes sense to switch. em Thank you for that. we've also, yeah, we've had a, you've mentioned it already, the opportunity to demo a new product, a better, which you're releasing soon, which is an AI mortgage advisor. And I've tried it myself, which is really fun. Thank you for that opportunity, but also useful in helping myself find refinancing opportunities by mortgage and things like this. it was a good process to go through. We're actually going to have just a quick demo for everyone here. We're going to cut it in. So have a quick look at this. name is Ryan Grant. How are you doing today? I'm doing great. As you can see, it's a beautiful day in New York City. Absolutely. That panoramic city view with the river and bridges looks stunning. Perfect backdrop for talking about your home plans. What's your goal today? Are you thinking about buying, refinancing, or something else? I'm thinking about refinancing. I have a mortgage. It's 7 % interest. And the balance is about $500,000. I was wondering um how much money I could save if I were to refinance. Great. Refinancing from a 7 % rate could definitely save you some money. To get a clearer picture, could you tell me how many years you have left on your current mortgage? Also, if you know your current monthly principal and interest payment, that would help too. I don't really know off the top of my head the monthly payment, but there are 27 years left. Your current monthly payment is about $3,439. I have two interesting refinance options for you. One with the lowest monthly payment that involves paying points upfront, and another with a par rate that has lower closing costs. Which one would you like me to explain first? Tell me about the par rate. par rate option has a rate of 5.99 % with a new monthly payment of about $2,995, saving you around $445 each month. Would you like to hear about the lowest monthly payment option next? No, that actually sounds really good. I'm gonna let you go now, because I'm at work, but let me think about this and then get back to you. Is that all right? Absolutely. Take all the time you need to think it over. If you have any questions later or want to explore your options more, just reach out. Have a great day at work. All right, thank you, Ryan. Hello, my name is Leah and I am the Vice President of Tin Man AI at Better Mortgage. Thank you for being a part of our pre-release and I am excited to learn a little more about your experience with Ryan. Let's start with a basic question. What did you think? Wow, okay, well this is pretty wild. He was pretty amazing. um I was really impressed by how quickly he could compute um basically a math problem versus how long it would take a human to go look up uh how much I could save with a refinance. That's awesome to hear. Ryan does have a knack for lightning fast number crunching. I sometimes wonder if he's secretly part calculator. How comfortable did you feel sharing your personal or financial info with Ryan during your chat? You know, I guess I didn't think about that. um When I share my personal financial information with a human, sometimes I feel a little bit embarrassed. But I guess I didn't feel that at all talking to a bot. That's really interesting. And I totally get that sometimes it's easier to open up to a bot who's not judging you for your coffee budget or impulse buys. Did Ryan understand your questions clearly and respond in a way that made sense to you? Totally. Very understandable. um Made me feel really comfortable. I'm glad to hear that Ryan made things feel comfortable for you. That's a big win in my book. How natural or human did Ryan feel during your conversation? Did he come across more like... I mean, I wouldn't have... I knew that he was a bot. I knew that he wasn't a human, but he was pretty close. Thanks for sharing that. It's great to hear Ryan was pretty close to feeling human, even though you knew he was a bot. That balance is tricky to get right. Can I ask, how easy or difficult was it to complete your mortgage application with Ryan's help? Well, I didn't complete one, but actually I'm at work. um We're very busy here working on an AI platform. um So I've got to go, but it's been really fun chatting with you. I totally get it. Busy days call for quick chats. Thank you so much for sharing your thoughts about Ryan. I'll make sure the team hears it and Ryan's ego can handle it. If you ever want to pick up where we left off, I'll be here. Have a great day at work and good luck with your AI platform. Thanks. Yeah, it's great to see uh it in the flesh, if you like it. And yeah, it's nice to see how companies like Better exploring practical uses of AI beyond chat GPT, which is what most people are used to. I mean, I'd be surprised if anyone can name anything outside of chat GPT where AI has actually given them value in something. How and when will you be rolling this out to prospective customers? Yeah, all right. So the avatar, no one has anything like this. uh I'm so excited. uh We are so excited to offer it in a pre-release format. So pre-release means we've done a lot of testing internally. We've demoed it externally, and now we want to get feedback. It is not broadly available to borrowers yet. So for right now, we have, you have to apply to participate in the pre-release. My team is very strict about who we will allow to use it. So keep an eye on it as we figure out when we'd want to broad launch. uh Because feedback is so important at this stage, we specifically not only offer access to the AI mortgage advisor bot, but then we also have the similar experience for the feedback bot. So. All of those transcripts from the feedback bot are maybe a little bit more important than what each person says about the actual mortgage advisor bot. So I can't say exactly when we release it. We'll be very careful. uh I'm really curious about what the value ends up being. So on the one hand, you could say, this is kind of a gimmick. who would really wanna use this? Like what exactly is the benefit um versus speaking on the phone to a bot versus just talking to a human being versus doing a chat bot versus the website? Like what, um is this silly? And so my personal thought uh is twofold. So one, in terms of like this as a front door for borrowers, I suspect that first time home buyers who have stupid questions that they're embarrassed to ask loan officers like a real human being about, like maybe they're insecure about their income. Maybe they just have no idea how the process works and they feel embarrassed that they don't know. Like uh maybe you're in your 40s and you've never had a mortgage and you feel like, my God, I can't believe I've never done this before. This is so embarrassing. You can just ask your chat bot, the chat bot, the avatar. It has unlimited time. It's not judgmental. You can ask questions over and over again. It doesn't get frustrated. So for first time home buyers, I think that that's a great use case. Otherwise, I actually sent it to my uh dad, who's a baby boomer in his seventies. And he pointed out like, you know, he has all the time in the world. And sometimes he wants to really dig in to questions. Like it's nice to have somebody just listen to you. And like that's actually rare in the world this day to have someone deeply listen to every word that you're saying, think about it and come back with a really rational response. So I'm also thinking in addition to first time home buyers like the baby boomer or the aging population, which by the way is sitting on significant amount home equity, they're also available, like they're thinking about reverse mortgages. uh Like that is a huge segment that's probably underserved by all of these gen AI tools that are out there right now. uh They're also the ones, so I suspect that the avatar works best on a screen. We'll see if people want to FaceTime with an avatar and if that's valuable. Like my personal preference would probably be just like typing into a UI or on my phone, getting my own mortgage, it was all on my phone. And that was just the easiest for me. So we'll see. So we'll see what the feedback says. But I'm really curious about the use cases on both of those ends of the spectrum. um I'll also say like, feedback, the feedback avatar, Leah, my avatar, we kept her funny and we kept her a little loose. so she can make jokes. And I think there's something to that, like getting a mortgage is a pretty joyless experience. I wanna get feedback on what types of personalities um create a more positive experience for people. There's no reason why it has to be so dry. um Maybe there's new stickiness. One of the things I wanted to tell you that... for our lenders, we allow them to rebrand Betsy to meet their needs. So we could rebrand Betsy Ed, if you were a lender, Ed's Lending Shop. And it could be, you know, ask Ed. And then we could have Betsy voice, use Ed's voice. We could just, we need a couple minutes worth of your voice and we can go ahead and do that quickly. We can also change the tone. So the tone of Leah, is very different from the tone of Ryan. And so we can allow, we've actually, have on my team, I have someone who's from Australia. And so we asked the internal GenAI assistant, hey, can you have text as if you were from Australia? And it started, good day, good day mate. like, so the sky the limit in terms of the ability to customize this tool. And so that for our lender customers. offers a totally different type of branding experience from their software that forget it, that company that has 85 % market share, like no way. You call it what you call it and you don't get to rebrand it. We'll even rebrand Tinman to call it whatever the customer wants. So we've got a customer that's a bank right now. We don't care, call it Kirkland for Costco. Origination software, like whatever, it's yours now. Yeah. definitely something to be said for that because even in like we obviously we make a lot of content. Opto make it in a huge amount of different formats because different people like consuming it in different ways. If they want the podcast, they want the YouTube, they want an email, an article, know, things like that. So it's exactly the same thing. There'll be a certain proportion of people that will prefer to interact in that way. It'd be really interesting to see who those segments are. and which ones get the most benefit from it. think, you know, personalizing it makes a lot of sense and probably unlocks new segments that didn't particularly engage with before. Also early on in this technology, think it can get, it can improve, it can get better. You know, the speed of which it can interact with people and, you know, the variety of different things it can talk about will obviously improve. you know, there's a lot of areas where it could expand into other markets as that happens. When you were building it, what were some of early challenges or teething problems you faced, if any, developing this sort of product? Sure. Do you mean the avatar or do you mean the platform? Okay, cool. Yeah. So I've been here for most of the building of the avatar. So one thing to think about here is that I talked about the platform that we're able to really easily layer on Gen.ai tools. So we already had in place a voice bot. Like that's been live since a year ago. And that voice bot we just announced in our earnings call over the quarter took, had 700,000 borrower interactions and we're learning from those interactions. So the avatar is just the next, it's the extension of that same AI. So there's nothing really new about it, about the model itself. All it is is video and audio. So specifically, My avatar that I just showed you, that was three minutes of my voice and me on a screen just like that, speaking to my laptop. em So in terms of technical challenges, there really was no new model development. It's just a different skin on something that's already widely released and widely used. And so that's how people should be thinking about that's. an ideal way that people should be thinking about deploying new technology. If they've built it to scale, it should be easy. I'll tell you the next generation of, for us, could be just a chat GPT like interface. So the front door for a borrower could just be you go to Chat GPT and you're like, I want to buy a house. And you don't have to do anything. It knows who you are. It has all the context from all of your conversations. It knows where you want to live. um It knows what kind of mortgage, it knows your job, it knows how much money you have. It has access to your Dropbox where you have all your income statements on all of that. Just through a chat, you can, with a chat CPT, you can get your mortgage. So that is also very easy. If we get to a point where we think that that's the interface that consumers want, super easy for us to turn it on. But I'll say, um Like there are working with the avatar, it's really fun of all of the bots, like all kinds of fun decisions that are more like uh moviemaking in a way. So for example, these are fun decisions, like who's likeness do we want to represent the company in this pre-release stage? Who would even want to have their likeness used? What is the legal contract look like with that person? What's the compensation model? What are they worried about? um What could go wrong with the brand? What kinds of prompts do we need to put in place? What if somebody wants to do something dirty with my avatar? And that's me out there, right? That's my likeness. um So I have to think about myself. What does that mean for me, for my brand? What if she says something crazy? um One of the fun things is how loose do we want her to be. Like you can change the spiciness on hers, which we've played with because she's not actually collecting a mortgage so we can have her be a little bit more silly. um For demo purposes, she was really loose at one point. I could talk to her about like all of my life problems and she would give me all kinds of life advice, which is really fun for a demo. uh So it's really more about, okay, let's think of all the things that could go wrong. and let's make sure that we've um prompted them and put guardrails in place um so that we can mitigate any of those uh potential negatives to happen. um The other thing we put in place is I'm really restrictive right now about who can even access the avatar. So it is not broadly launched. You are one of the select few um who gets to access early. Yeah, it's really interesting to see and really, we'll be following closely to see how it develops. I hope we get the opportunity to use it again. I had a couple of other questions before we wrap up and the first one was, I was interested to know, and I'm sure a lot of other people are as well, and maybe it's a simple answer, but what are the most important factors that influence a borrower's available mortgage rates? em And how does Tin Man sort of optimize around that? Yes. So very straightforward question. So typically your credit history, that's a big driver. um Your income is also the other big driver. So when I said that we in our end to end platform, we have more than 40 different investors. They have each of them has a myriad different uh amount of products that are available. So a borrower coming into our better.com front door. They can have a wide range of credit scores, a wide range of different incomes. They could not have good documentation for their income. They could have very lumpy kind of income. They'll probably find some kind of product that meets their needs through Better's platform. So that's really what drives that. Thank you for that. finally, talking about the future, obviously you're in quite a sort of, not a niche space, but it's a particular space in mortgages. What other product areas do see Tinman expanding into or other innovations in the mortgage space in the coming years? If you can touch on anything, that'd be really interesting to know. Well, uh you know, one of the things we talk about is being considered. We want to be the infrastructure that powers all loans everywhere. So our CEO will talk about we want to be Stripe for loans. Mortgages are the most highly regulated and the most complex. So we are masters at the hardest loan type. So we've climbed Mount Everest. We've proved it out with 100 billion of our own production. And now we can offer all of that pain that we have behind us, the benefits of that to other lenders. So we've positioned ourselves to have the hardest segment first. After that, personal loans, auto loans, like piece of cake compared to everything else we've already done. And that's only considering the US. There are other markets as well that we could go after. So right now we're very much, let's prove it out in mortgage land strategically. We'll work with other fintechs, but our bread and butter, our expertise is in the hardest piece. New fintechs coming in, AI geniuses who've got a lot of money backing. they're just, it's going to take years. It would take years for them to build what we've built for the mortgage industry. So we're going to prove it out. with our bread and butter and as our CEO says, land and expand from there. yeah, I suppose. As you mentioned, the difficulty that you're facing in the mortgage industry, getting these new providers to use it as also a moat for you once they actually switch over because they don't switch often and you get those relationships that take years to develop. Leah it's been great to have you on the show. I'm sure everyone's going to benefit greatly from listening to this and they'll be very interested to see. um I mean they can't see the avatar product right now, I suppose, but they can go on the website and use the chatbot etc that exists today. Yes, go to better.com, check it out. And yeah, is there anything else you'd like to say before we wrap up? No, I mean, for me, this is like the opportunity of a lifetime. So watch this space. There are amazing things to come. Yeah, thank you. Yes, it's incredible when the loan industry is just gigantic as a as a market and in particular America. I didn't even know that as a useful stat that is incredible really. So, um, very interesting space and, you know, definitely a company to keep on on top of. Thanks again, Leah. Have a great day and hope to catch up and again in the future. Yep. Thank you. See ya.