- 2 months ago
In this exclusive executive interview, Steve Butler focuses on not just the future of AI in mortgage but practical applications that True is implementing today. This conversation centers on background AI, a core differentiator for True. Unlike assistant-based tools or traditional automation, background AI functions continuously on the backend, offering a more seamless, scalable approach to enhancing mortgage workflows.
Butler also address one of the key challenges facing leaders in the industry, how and where to begin implementing AI in their processes. This interview explores implementation and highlights strategic considerations for lenders, emphasizing the importance on focusing on pain points that yield meaningful operational improvement.
Butler also address one of the key challenges facing leaders in the industry, how and where to begin implementing AI in their processes. This interview explores implementation and highlights strategic considerations for lenders, emphasizing the importance on focusing on pain points that yield meaningful operational improvement.
Category
🗞
NewsTranscript
00:00From HousingWire in Dallas, Texas, I'm Alison LaForgia, managing editor of HousingWire's
00:09content studio. And today I'm joined by Steve Butler, the CEO of True. Steve, thank you for
00:14joining me today. Alison, thanks for having me. Steve, let's start with a exciting announcement
00:21that recently came from True. You recently stepped into the role of CEO after serving as the company's
00:28president and COO. You've been with the organization since some of its earliest days, and you have a
00:35background that's pretty impressive, spanning fintech, B2B SaaS, and AI innovation. How has
00:42your journey shaped your vision for what's next at True? That's a great question. I think it's a
00:47great way to start, Alison. You know, in looking at it, I would say that the common theme in my
00:53career, and it's a long career now, it's over 40 years, is business process optimization. And it's
00:59been in, you know, only the last 10 years I've been in mortgage, but it's been in equally complex
01:04areas. I mean, for example, I was in semiconductor manufacturing for many years. You know, I was in
01:09software manufacturing for many years, and then a number of financial services transactions at the
01:16enterprise level. And so that perspectives of kind of outside mortgage and some really complex things
01:22had to get solved. I mean, these are all mission critical, like mortgages and complex problems.
01:28And the other theme that I found, and I see this in mortgage a lot, is that usually you can't find a
01:35single person that understands the process from end to end. And that was similar in those other
01:40industries, trying to get somebody to understand how this whole thing's supposed to work. And so I bring
01:45that to mortgage. And at the same time, you know, I'm still a newbie. I've been here 10 years. And so
01:53I'm not going to parachute in and say, we're going to change everything. I think many people have tried
01:57that before me. So I have a deep appreciation of mortgage and how it works and how people want to
02:03get it to work and how passionate they are about it. But I do bring in some outside ideas from
02:10manufacturing that actually did solve the problem in a complex way. So that's how I'm looking at it.
02:15So let's talk a little bit about that business process optimization and one of the ways that
02:22True's approaching AI. I've seen there's been a lot of talk at the forefront of True's work,
02:30specifically with background AI. How does it differ from traditional automation or assistant-based
02:37tools? And why does that matter for lenders today? Yeah, no, that's a great question. It matters a ton
02:44because in the way that you get a lift in ROI from AI is that you can give it to that AI worker and it
02:53can fully handle the task. You know, there's a lot. I mean, AI is everywhere now. We're using
02:58ChatGPT for everything. We all know, though, it only takes you partway there. No one's going to turn in
03:03a paper to their professor without reviewing it and updating it. You know, you're not going to do a
03:09legal brief from a lawyer without actually changing a lot of things. So it's the same way with AI. If
03:15you're going to require a human to sort of deal with the output of it in mortgage, then are you
03:21really getting a lift, right? And that's the problem with assistance. And by the way, I talk to the C
03:26Suite all the time and they actually are frustrated that a lot of their software investments, they end up
03:31with a bigger team. They end up with specialists on that tool. All of a sudden, we have this whole
03:37team we didn't have before just because you have to run that tool. So background AI workers, if you
03:43can figure out a manufacturing process that they can handle fully, you basically get 100% gain for
03:49that manufacturing process. So it's really critical that you figure out the AI, what it can do, its level
03:56of intelligence, and then the manufacturing stage and mortgage that it can handle fully. You get a
04:01big difference when that happens. That is so interesting talking about the places that AI
04:07appropriately fits in this actual process. And I want to dig a little bit more into one of the
04:15challenges, because you mentioned you talk to the C Suite a lot. So despite a lot of the hype around AI,
04:23one of the challenges that executives face is identifying where to begin with AI implementation.
04:29How should leaders think about getting started with this process when you're looking at such
04:35high optimization rates?
04:38Yeah, yeah. No, in fact, you know, there's so many places you can jump in. It's a complex process, as you
04:45know, many, many handoffs along the way. We're really strong advocates that you got to get your data
04:50right first. It's all about data, quality data. And so applying AI so that you can get clean data. When I say
04:58clean, it means that it's full, it's complete, it's consistent. You know, if you've got a borrower
05:05address in the LOS, it should be present on all the documents that have that borrower address.
05:10It can't be, you know, right in one place and wrong in another. It's got to be consistent. So
05:15getting clean data, and why that's so important is because all the decisioning that happens after that.
05:21So if you get clean data, you know, you know, several weeks down the way, then guess what,
05:28that's when your decision is going to start. But if you can get clean data within a few days
05:32of borrower supply documents, provider documents, if your data gets clean, then, then your decisioning
05:38can start, like income calculations and asset calculations and pre approvals, they can start
05:43as soon as you have clean data. So we're huge advocates of get your data right first. And then some
05:49really nice things happen downstream. So I'm guessing one of those nice things that happens
05:54downstream is the elimination of stare and compare tasks, right? Yeah, yeah, yeah, exactly. Yeah,
06:01that's, that is one of the only weapons today available to lenders for how they get clean data
06:08there. So you can just picture they got the document on one screen, right? And then they got the
06:13extracted values or the, or the, or the, or the, the information that's going into the LOS and
06:19another, and they're trying to sort of compare them. Are they correct? I mean, that's such a, such a
06:24cumbersome, slow process. And the poor underwriter who gets the file, and finds the inconsistencies has
06:32to close the file and kick it back down to, to the processor or to the LO and say, clean this up,
06:39get it right. And so stare and compare is, is the way it's done today. And that's got to be eliminated
06:44and AI can do it if you have the right AI. So let's talk about the implications of streamlining
06:50those stare and compare steps. Yeah. Well, the, the, the biggest thing of all is that if you can,
06:57if you can eliminate stare and compare, in fact, if you can take it away from the processors, you don't
07:03have to worry about that. And the LO, their productivity goes up significantly because you've
07:09got a background AI worker that's doing all that work. So imagine the processors double or triple
07:15their productivity. They're doing other things about in that loan, doing some initial, you know,
07:20pre, pre, uh, pre underwriting calculations, right. And, and, and looking at different ratios and so
07:26forth. They're collecting other things about that loan that they need to get collected. They're not
07:30having to do any of the document related work. So productivity is a big win for that. But the
07:35other thing that happens is you can start to decision early so you can do much faster pre-approval.
07:41So when you get, when you get applications that come in and, and borrowers are shopping around,
07:47if that application can be, you know, pulled right away and, and the borrower can get some feedback
07:52almost instantly about, Hey, I, I, I, I, you, I like your calculations. They look good to me.
07:58Your documents look good to me. You've probably, you know, reined in that customer. That's probably
08:03a new customer for you. It's all about transparency and instantaneous feedback back to the borrower.
08:10Now that's really exciting. When we talk about the level of transparency that we're talking about
08:16and having the ability to get an earlier decision, especially around something like a pre-approval
08:21when you're working with someone who's eager to get started with the process. Let's talk a little
08:26bit about some of the early adopters that you've seen over at True. I'm going to ask you for a
08:32preview, Steve, without naming names, can you share about how some of the leading lenders are
08:39approaching the shift towards AI implementation? Yeah, they're following the blueprint we recommend,
08:46which is to start off with your data, getting your data right. And so they're rolling it out to
08:52their network of LOs, you know, across the country. And the background workers are doing all the
08:58indexing and all the extraction. And so you can imagine the lift they're getting at the field level
09:03because those LOs now have more time to provide white glove treatment to their customers, right? And
09:09the processors have a lot more time to get the loan through quicker. So you're seeing, we're starting
09:14to see those pre-approvals come way down and we're starting to see that productivity up. And I'm talking
09:19big national lenders. I'm not talking small. I'm talking some significant lenders that are
09:24really leveraging this technology the way it should be. No one's trying to boil the ocean.
09:29That's, you know, where you basically say AI is going to do everything. They're doing it in stage
09:33by stage in a very thoughtful approach. And I think that's the way. And down the road,
09:39they're going to be very automated. It sounds like True has really been a partner with these early
09:44adopters and has helped shape their philosophy towards integrating that into their processes.
09:50Yeah, I think we like, we've been doing AI for 10 years. So we understand it really well. We were
09:55fortunate to be running. We've probably run tens of millions of loans through the system now. So
10:01we've got very, very high accuracy, which is why we can deliver it as a background AI worker. I think
10:05we're the only ones that can deliver it as a background AI worker. So that gives us the ability
10:09to help lenders transform. The whole mortgage insurance industry, True, most of the mortgage
10:16insurance, by the way, runs through True today. We have almost all the vendors in that. So the loan
10:20file goes to them and we turn it around in under 10 minutes now, where it used to be several hours.
10:26That's the kind of transformation. So we talk to originators now and correspondents and services
10:32and say, look, we can help you transform at that level. So we have a lot of credibility as a result of
10:39the work we've done over the last, you know, last 10 years.
10:42So True isn't looking at AI as anything new. I mean, the speed of innovation is rapid,
10:48but you guys have a depth of knowledge here. So what I want to wrap with today, Steve, is
10:53as you look ahead for the next, let's say, 12, 18 months, what are you most excited about for True's
11:00future?
11:01Yeah. So we have, yeah, it's a great, it's a great, great question. I love these kinds of questions.
11:06I mean, we have brought a lot of generative AI now into the product. And so we're seeing accuracy go
11:13up even higher because we have this 10-year machine learning model, which is so intelligent.
11:18Now we have this other model, which is an LLM model. And those two models work in concert
11:23to get to the right answer. And so you're seeing accuracies that are better than human accuracy
11:28using machines only. So our solution doesn't need to go to India for review, for example. So I think
11:34we're going to see tremendous gains in accuracy and automation as a result of that. We're integrating
11:40our solution now into underwriting solutions and other tools. And so you're going to see some
11:45announcements from us, which have us plugged into some of the mainline big underwriting income and
11:51automated underwriting solutions in the industry. And you're going to see other partners joining into
11:58this platform that we have called the MOS, Mortgage Operations Service, because it gives them a common
12:04data layer that they can then launch from. So all the tools out of the box are interoperable.
12:08So I think this idea of manufacturing the way physical products are manufactured, which is all
12:15downhill, instead of this back and forth, I think we're going to nail that. And when we nail that,
12:20we're going to see a lot faster manufacturing happening in the mortgage space.
12:23Well, it sounds like a bright future for TRUE and how TRUE is approaching improving the tech
12:29process for mortgage. Steve, thank you so much for joining me and for giving me this insight
12:34on what's going on over at TRUE.
12:37Allison, I really enjoyed your questions and thanks for really having me. Look forward to the next steps.
Recommended
2:51
|
Up next
23:26
28:28
21:15
18:52
36:23
24:50
19:13
22:24
12:58
16:03
Be the first to comment