AI Unpacked: Leveraging AI to Transform Mortgage Operations | Optimal Insights | May 19, 2025

00:02

Welcome to Optimal Insights, your weekly source for real-time rate data and expert capital markets commentary brought to you by Optimal Blue. Let's dive in and help you maximize your profitability this week.

00:18

Welcome to Optimal Insights, your weekly source for timely market analysis and expert commentary from Optimal Blue. I'm your host, Jim Glennon, Vice President of Hedging and Trading Client Services at Optimal Blue. Our clients and industry partners have long relied on Optimal Blue for trusted insights and commentary, and these podcasts are an evolution of our commitment to keeping the industry informed. Let's dive into today's episode. Welcome everybody. Happy Friday as we're recording this, but it'll be coming out on Monday. A big group of us will be at the MBA secondary.

00:47

Hope to see a bunch of you there. Got a great show for you today. We'll of course have a market update with James Cahill here in a couple of minutes. And then we'll have our main segment is a long overdue discussion about AI. So we're talking about artificial intelligence. We will be joined by Kevin Foley, who you all know. And we will also have Shawn Chandwani, senior dev manager at Optimal Blue. So if you have any interest in the current state of AI development, how AI works,

01:17

why it's better today than it was just a few days ago and how you should be employing it in your daily life in any industry or even at home. I feel like that should be all of us. You should listen through this episode. We'll also touch on some features that we've built into Optimal Blue to help lone originators specifically to do their jobs more effectively. So an overview of AI in general, where it is, where it came from, what it's used for, but also a little bit about what we're doing with it at Optimal Blue. But first, just some high level.

01:46

% of: 2019

02:15

There are purchases going on despite what you read out there. So there's a little bit of market info for you. And let's go talk with James about what's going on in the broader market. All right, James, welcome. Thanks for being here again. Thank you. It's nice to be back. So CPI, PPI, right? Inflation week. What are we seeing? The numbers all seem to be jumbled and somehow tarnished and manipulated by

02:45

by tariffs and consumer sentiment at this point, but where do we fall in terms of expectations? Definitely. It feels everything this month comes with bit of an asterisk after it, but CPI came in at 2.3 % year over year. The expectation was 2.4, so we're actually a little bit better. And PPI, the producers, they came in at a shock. It was negative 0.5 while we were expecting 0.7. So it seems on the producer front,

03:13

we saw a decrease in inflation. To take a second just to harken back to last week, one of Powell's big points was the American economy is very robust. We've got good employment, inflation looks pretty good. There's a lot we can absorb. And these numbers do reflect a bit of that. It's, we've been working towards a soft landing for half of a decade now. This is kind of progressing in the right direction. This is what they want to see.

03:42

All right. So yeah, we're resilient if nothing else economically. Even if we post negative GDP numbers, we're not considering it recessionary because there's asterisks. Like you said, there's this huge flood of imports that happened all of a sudden because everyone was trying to buy stuff before tariffs come into play. Now, tariffs are in play and we're still not seeing inflation, which is somewhat encouraging, but there's also probably some competitive factors here too, right? Because you've got kind of a souring consumer.

04:12

as we see in those numbers. And some of that is because of tariff, like it's kind of circular. Folks are being, I think, more selective about where they spend their money and they're probably gearing up for more of that, right? If I know that Nikes are going to cost 30%, 40 % more in the next year, I'm only buying two pairs instead of three. And I might be selective about what store or what website I shop at for that. You go through shoes a little bit faster than me, then I run mine to death. But definitely, there's definitely some competitive factor going on in there.

04:42

I would say looking at these numbers, that CPI is the front end. That's what the consumer is getting. It actually decreased a little bit. And then you look at the PPI, the wholesale level, there's decreased significantly more. So there's a question there. What happened? Why is it that they would see, as we're expecting this inflationary effect of the tariffs, why would on a wholesale level they be seeing a decrease? The answer that people have kind of been working through is

05:11

GDP last month was negative and that was caused by a rush to import to get ahead of the tariffs or to get them as low as they could. So a lot of wholesalers and retailers are sitting on a large inventory right now. They have six, eight weeks to try and draw down that inventory and wait out to see what happens. So, you know, if you're a wholesaler and you're sitting on this huge inventory,

05:40

and someone's coming to shop from you, you have two options. You can push that inflationary price onto them, or you can eat a little bit of the cost yourself. In the long run, eating that cost is something that you can't do. But in the short term, looking at, hey, will these tariffs even exist in six weeks? And if you think the answer is no, why would you be the first mover? Why would you drop

06:10

30%, 25%, 50 % on your consumer, they're going to flee from you. They're going to start looking for better deals. They're going to look at who else they can go to. And then six weeks later, the tariffs are gone. You can drop your prices back down, but everyone's going to remember what your prices were. Right. You're going leave that bad taste in their mouth. And it seems like this is a strategy that's being employed by a lot of businesses right now. mean, we've incurred inflation in things like food and energy.

06:40

throughout the years and those are very volatile items, right? But a lot of these items that come up in the tariff discussion, there doesn't seem to be any significant changes in actual consumer prices there. And as you can see in the data, right, the 2.3, which is almost at the Fed's target. So yeah, it's interesting that that's how it's playing out. It seems to be the consensus then that most producers believe and most retailers believe that we will end up in a spot where

07:06

there's not significant tariffs, but when there are, then that's when they raise prices. Like who's going to step first, right? Who's going to jump first? But no one's trying to get ahead of that train apparently. Definitely. It doesn't pay here to be the first mover. What I will say is, you know, looking at that CPI number, it is slightly down from expectation from last month. But if you start to drill into it a little bit, you'll see that energy is very much down from last year. And the administration was, you know, drill baby drill, it very forward.

07:35

we're going to try and unlock energy, bring that cost down. That is a huge driver of lowering inflation. But if you start looking at goods, and the one that really stuck out to me was electronics, audio, speaker equipment, all of which are imports, that's Samsung, that's Korea, that's Japan, that's China, that is up about 10 % on the year, which is what the universal tariff that the United States has on all incoming goods is right now.

08:04

when you start to look more line by line, we are seeing goods start to get this inflationary effect. It's just the overall, there's still a big driver of energy downwards, which is helping to compress that inflationary effect. That's interesting. because I mean, stereo equipment, that's probably one of those areas where you would say it's not volatile at all. There's been inflation over 20 years, but you don't see it like the price of eggs.

08:32

Yeah, the counterbalances energy and the administration has been pretty clear even during the campaign that one of their main priorities was to drive energy prices lower. And that does tend to happen in a Republican White House because there's going to potentially be less regulation. So potentially more drilling, less regulation in general on those activities and less fees and less cost to do it. Definitely. On the kind of the same vein here is so yesterday Walmart released their earnings and their CFO

09:02

had a lot to say about tariffs and inflation. And it's always a shock to me, but Walmart is, of course, it's the largest retailer in the United States. It's the largest grocer in the United States. It's actually the largest private employer in the world. So they are, you know, maybe this is a dirty word in a way, but they are too big to fail. They are so strong, so resilient, such a, in a great view.

09:30

into what's happening in the American economy. They're selling stereo equipment. They're selling tomatoes and onions. You go get a shirt from Walmart. They've got a great view into all of it. And their CFO was saying, they've got, until the end of May, you're going to start seeing prices rising. And you're going to see it more in June and you'll see items start to come off the shelves. He was celebrating that the tariffs on China were down to 30%. But the quote was, that's still too high.

10:00

So when I kind of look at that and I think Walmart is this behemoth, it's this giant, it's in a way very representative of the American economy. And they're saying, yeah, we are going to see shortages. We are going to see inflation. If they're saying that, what does that mean for Main Street? Like what do the small family owned businesses, what does it look like for them? And the story has to be repeating and it has to be a little bit more dangerous. Yeah. I mean, like you said, they are

10:30

They're the biggest business in the world basically that's not a government and their margins are extremely thin and they rely very heavily, I know, on China and some of the other countries that produce goods very cheaply. Yeah, they're putting the math out there on the board and saying this is what obviously needs to happen or is going to happen. Yeah, 30 % might even be conservative. keep reading about effectively a lot of these.

10:58

tariffs on things, especially on clothing and shoes and things like this are actually higher if you start to compound the existing trade war tariffs in addition to the favorite nation tariffs and then the fentanyl tariff and just the baseline tariffs that the president has put in during the 90 day negotiations. Yeah, it's kind of inevitable. So yes, where goes Walmart? There goes everybody else. So yeah, the mom and pop shops are going to

11:26

They're going have to lower their margins. Yeah. You're going to have to act the same way as at the beginning we were saying. You're going to have to eat some of the cost upfront and hope that this changes before it becomes a little bit too heavy for you to bear. Then you're going to have to start putting it on the consumer and hoping that they can take it. I'll give a brighter side to that. Since January 22nd, the tariff policy in the United States has actually changed 50 times. Yes. I believe it.

11:57

So if you look at it that way, the likelihood that it will be different in six weeks seems very high. We've actually changed it every other day. So there's a likelihood that it will be different, hopefully for the better. And with the negotiations ongoing, hopefully we'll come to a resolution. But I think that 30 % might be here for a while. A lot of optimism out there, but change is inevitable.

12:26

Yeah. Where does that leave us? mean, next week, probably headline dependent, probably tariffs, but really no major economic releases or data to speak of, Agreed. This was definitely a little bit more of the market week with CPI and PPI, but nothing of major interest next week. we'll be, everyone will be glued to the headlines and excited to see what the world produces next. Beautiful. All right. Thanks so much.

12:55

Thank you for having me. All right. Welcome Kevin Foley, Shawn Chandwani. We're to have a little conversation today about something that is just out there everywhere and every piece of news that you read over the past year for sure. And then, you know, even preceding that, and that's AI, artificial intelligence. We hear a lot about Nvidia or a lot of the other big players out there chat GPT. you know, I think a lot of us at home, you know, we've, we've used some of these applications on our phones sometimes to do something very simple.

13:25

Like last night I was using it to check my oldest daughter's homework. So literally took a picture and said, is this right? And it went through in detail, checked every question and gave us either a correct or an incorrect and why it was anyway, it's just amazing stuff. Uh, but you hear a lot about it. Everybody claims that they're doing it. We thought it'd be good to have a conversation here on the podcast. Uh, so we have looped in our senior dev manager, Shawn. And of course you all know Kevin, uh, just to talk about.

13:55

Generally AI, what's out there? What are the practical uses for it? And then at Optimal Blue, how do we think about it? What have we already rolled out for AI and then maybe what do we have in terms of plans to continue down that road? So welcome, welcome Shawn, welcome Kevin. Yeah, good to be here. Good to be here. Thanks for having me. Yeah. Thanks for taking a little time here on a Friday. So yeah, let's get right into it. kind of, I don't know, started out, Kevin, what kind of questions do you have for Shawn and what would be good for this, you know, our listeners to hear?

14:25

Shawn. So Shawn is one of the: 2025

14:53

For our listeners, what does that really mean, AI, and why is that so important and seems to be proliferating across not just our industry, but everywhere? Yeah, right. In general, AI and machine learning is not necessarily a new concept. It's been around for a long time. Even when I was in college 10 plus years ago, I took classes on artificial intelligence and machine learning.

15:22

But I think the big breakthrough that happened was with specifically with LLMs, right? Large language models. And in previous to that, right? You still had AI, but it required very specific instructions to be able to use rather than just general language. Like, you know, all the AI that people are using today where anybody like, you know, what we just mentioned can pull it up on their phone and use it how they wanted. Right. So really when

15:51

LLM started to become a thing a few years ago is when it really took off in this whole concept of AI. That's what it became. Because people would refer to AI prior to that as something like Amazon Alexa or something, right? People would think about that as AI. So it's kind of been this evolving term, but now with large language models, we truly start to see it because the way that you interact with it has completely changed. And one of the jokes I like to tell...

16:19

from being a very technical person, I'm always diving into new technologies, whatever they are. But for when I know something is starting to get big or go mainstream is my wife who is not technical at all. When I see her using chat GPT and asking questions to it versus going on Google to look up something, then I'm like, okay, this is starting to get mass adoptions. know what I mean? Yeah.

16:47

I hear you on that. It's also like when you hear your neighbors start to buy Bitcoin, it's like that's when you know it's really gotten out there. Maybe it's time to sell. Time to sell, No, but that's super interesting. Shawn, you and I have worked pretty closely on some of our AI initiatives and also be remiss without giving a major shout out to our senior product manager, Ravi Desai, who's been a huge integral part of

17:15

helping us on our AI journey as well as Chris Tu and Eric Shirley who have worked on additional capabilities on the product to pricing engine side. But I thought it would be cool to give like an inside scoop at the story of how some of the things that we've been working on here at Optimal Blue have come together over the past year plus and also be able to share some insights in terms of

17:42

what we can offer at Optimal Blue for lenders who are thinking about starting their AI journey. I would think about us here at Optimal Blue, we've sort of been through it. We started out like anybody else as AI newbies, right? And then we've learned the technology, we've grown the expertise and we've now deployed capabilities that are helping lenders at scale. it's something where

18:08

we're kind of like technological ambassadors for the industry. Here are the things that we've learned and here are the things that if you're looking to build something in-house or outside of pricing or if you're just looking to get a better understanding of the technology, there's a lot that we can offer. But I figured it would be good to kind of share the inside scoop of at least start out with that, how we started off. I know we've worked

18:37

pretty closely together on this, but I would probably peg it maybe around a year ago or so. We started really diving into the technology. know on the product side, we had kind of crowdsourcing of idea formation. We all got together and wanted to tackle some big problems that really have been big, long-standing persistent problems within the industry that

19:03

haven't been solved before, but we thought had the potential to be solved using artificial intelligence. And so the way that this all happened is we had this big list of things and we got together with Shawn, you guys and the dev team and we're just like, all right, here's what we want to do. How do we help us figure this out, help us get there? And some of the things that we narrowed down ultimately turned into originator assistant, which we rolled out back in April.

19:33

And Ask OB, which we rolled out this weekend is now going to be available to all of our customers. Both capabilities, by the way, at no additional cost to our lenders. This is part of our initiative to really drive new innovation at no additional cost. We'll talk a little bit more about Originator Sys and Ask OB, but Shawn, maybe you could pick it up from here. So, you know, we've got this big list over to you guys and you're like, all right, help us figure out what to do.

20:01

How did you guys pick that up from there? Yeah. So, you know, we had kind of already been exploring AI a little bit with the GPT models about a year ago, right? And now we're at GPT 4.0.4.1, but it was GPT 3.5 was kind of where things were at at that time. It wasn't even as robust as what it is today when we started. And we really just started experimenting with it a little bit on like a small scale of how do these things work.

20:30

You know, how could we potentially leverage them? We had a lot of brainstorming sessions, as you remember, Kevin, about, you know, different ideas and ways to implement this. And when we were going through those ideas, there were some that just made sense to open up that weren't available prior to this technology, right? And so we had to look at the ones that would have the most impact, but were also the best use cases, right? For something like working with an LLM. So I think.

20:58

That's kind of how the decision making process started, right? And from the technical side of things, to peel back the curtain a little bit, we really just started with very organic learning, right? Like it was a lot of just reading through documentation from OpenAI and Microsoft and other places to kind of get a grasp on how a lot of this stuff works. And then we were able to actually engage Microsoft and we have a great partnership with them.

21:25

And I remember specifically early on too, when we were kind of throwing some of these ideas out there and running it by them and, you know, kind of starting to develop some early prototypes, we would talk to them about it. They were really helpful in guiding us. mean, we would have, you know, calls multiple times a week. There'd be multiple email chains going back and forth with them. And they have like some really heavy, you know, machine learning and AI experts there that helped us with this. So it wasn't just like, you know, us doing this fully.

21:53

amateur style, if you will, just diving into this new technology, right? So that was a big part of it. And then once we really started to get up to speed on it and we saw that we could create tangible products, it really was an organizational thing that we came together. We brought people over from multiple different teams who were interested in working on this to come in and share their ideas and kind of learn and work on both of these projects that we had going.

22:23

And then we even had sessions with Microsoft themselves here at their Los Colinas campus, where we went in and they helped us out too with some of these. they were super interested in the use cases that we were working on too. remember talking to them and, you know, them almost wanting to, you know, use some of our stuff as kind of a study on how things could be implemented, you know, for other use cases for, you know, other companies that they might work with.

22:52

So that was really interesting to see because we were on the bleeding edge of it to where, you know, they would come in and sometimes say like, oh, this doesn't work right now, but next week or in a month, we're going to release something that will work. Right. So that's how much on the bleeding edge that we were developing this because there really wasn't a lot. You're starting to see some of them emerge now, but even still, there isn't a lot of these very targeted.

23:20

kind of corporate applications, right? You have a lot of general LLMs, but for what we're doing, it's leveraging those general LLMs for a specific product that we want to build. And that's kind of where we were really on the bleeding edge. So was very exciting. Yeah, that's wild. I didn't know all this was going on behind the scenes. I mean, knew we were at the forefront, you know, but I'm more on the client facing side. And this sounds to me like

23:46

you best practices for a business that is or was trying to be at the forefront of getting AI products out there to the point where Microsoft was looking at us and saying, Oh, like that's an interesting use case we hadn't thought of, or you're already getting ahead of what the technology is capable of. But we have that in mind and we'll have something available for you in the next few weeks, which is kind of wild. So it is wild. even just from my perspective, the amount that the AI landscape has changed from where we started.

24:16

like things like AI agents weren't really talked about very much. And now they've, you know, they're, sort of entered the fold more kind of like around the beginning, like end of last year, beginning of this year, as we were working on a lot of this stuff. I think you're right also to call out the partnership with Microsoft, you know, that it's, obviously they have their own partnership with open AI who owns the chat GPT model. And they've got a ton of in-house expertise and they were a great partner.

24:45

for Optimal Blue to work with to help accelerate a lot of the development and getting to ultimately where we wanted to be, which ultimately ended with us rolling out Originator Assistant and then ASCO B as well. so maybe just to, we can hone in on those capabilities for anyone who hasn't heard us talking about them before, maybe has, but hasn't heard exactly what they are and what they do. So Originator Assistant,

25:13

we put together, helps identify the best possible terms for a borrower using small changes in the loan scenario. we can help offer information around breakpoints for LLPAs and helps, you know, if there are small adjustments in the loan scenario, how can you leverage some of those adjustments to ultimately end up with the best possible terms for the borrower? And so we're using artificial intelligence to help.

25:38

evaluate the rules that are configured in optimal blue and drive that towards getting the best possible terms for the borrower. ultimately we, you know, we've demoed this, we demoed this at our user conference back in February. We've talked a lot about it since, and just the ability to run through not only originator system, but how it sits on top of some of the features that we built in prep for rolling that out. Like our scenario optimizer capability also our built in what if capability.

26:07

You can run through over a dozen loan scenarios, different possible versions of a loan scenario, really within a matter of seconds. So a big industry problem, finding the best terms for the borrower, we were pretty bent on trying to leverage artificial intelligence to help solve that problem. We rolled that out successfully back at the beginning of April. We had over 10,000 AI-based recommendations just in our beta period before we rolled that out.

26:36

to production. it's really great story about how all that came together. And the second feature which we rolled out by the time this airs on Monday, we'll have rolled that out, is Ask OBI. And this is really an AI-powered executive assistant for your Optimal Blue account. previously, if you're looking for information around

27:02

you loans that you're locking around, whether that's concessions, your margin information, your overall loan officer performance. A lot of that in the old world is handled by analysts who need to go and they got a poll reports. They got to run through the data, crunch the numbers. They got to figure out exactly what they're looking for and then ultimately get to the answer. And that answer might be just simply give me the LOs who have the

27:29

the top five LOs with the worst performance and concessions over the past month. Now, all of that capability using AI is available just within a chat interface. You can ask that, get an immediate response. Then all that time that an analyst was crunching the numbers, they can free that up to look at.

27:50

Maybe it's extensions. Okay, I used to have enough time to really dig into concessions. Now I have enough time to dig into both concessions and extensions. Are my originators locking for long enough? So we're really excited about rolling that out. And that one was kind of turning our whole data strategy on its head, right, Shawn? So maybe you could talk a little bit about from your perspective how that was.

28:16

That was a big change in how we were thinking about data, how we were about aggregating it. And I know that was something that you guys had a lot of really great work on. Yeah, absolutely. And that's kind of one of the things I like to take a step back and look at it and how it's really shifting the paradigm of how we think about developing some of our data and analytics tools.

28:41

Because, know, in layman's terms, if you think about it, right, like, let's say you're looking for your lock volume by product or whatever it is, right? Like there has to be a UI interface or somewhere that you can go and get that data that has to be built, right? And let's say, okay, now I want to filter it by products or by investor. it needs to be a new dropdown on there where you can do that filtering and get your data. Or, you know, you have to reach out to somebody to ask them for this data so that you can get it back and then.

29:11

You can work on it in the turnaround time on lot of developing those things or getting that data can be days, maybe even weeks in case if you're developing something brand new into an application, maybe months. So it really changes that when you can go in to an interface where you ask about the data that you want and it returns it to you right then and there. And that if you truly think about it, cutting that down from days or weeks or months to

29:41

seconds or minutes is insane. And being able to do that on the fly just gives you so much more efficiency and flexibility. And when we were developing this, we wanted to be very intentional with it, right? We're not developing a general LLM that is going to talk to you about solving homework problems or ordering a pizza or anything like that. It's about your data, right? Which is what our focus was. So.

30:08

making sure that that's what it's focused on putting in the guardrails so that that's all that it's about and expanding on it over time from there, obviously, right. But to begin with being able to just access your data in a way that's going to be very convenient and very flexible so that you can, you know, have it in any way that you want was really our goal there in developing it. So we, we took a very intentional approach with that product and what it

30:38

was intended to do. I think so, you know, a couple of things that I've learned to appreciate much more as we've gone through our journey is how data is sort of like, like the currency that you pay the AI model to perform better. And the more data that you can feed it around whatever tasks you're trying to, you know, get it to learn or get it to do effectively, the better it actually performs in the ways in which sort of early on we

31:07

you know, just things as simple as, all right, if you're trying to look for what your lock volume was, well, you could refer to that as your lock volume, your loans, your production. And those are all things that end up being fairly niche secondary marketing type language that hasn't really proliferated enough to be understood by general LLM models. Those are things you need to be intentional about teaching it, which I thought was, you know, that was kind of a light bulb moment, right? Like,

31:37

Oh yeah, of course. This makes a lot of sense. Yeah, no, absolutely. And in addition to that, we already had data and analytics applications and we already had built out a data warehouse to store a lot of that data because regardless of how good the LLM is, if your data is not cleaned and ready to be used by an LLM, it's not going to be very efficient. It's not going to be very performant, right?

32:02

So we were able to start on it quickly because we already had a lot of this data in a form where it was conducive for an LLM to be able to use. And on top of that, we're all in Azure, right? We're in the cloud. And like you mentioned, Microsoft has a partnership with OpenAI. So those OpenAI models are also available in Azure to integrate with. So it was all able to kind of come together more quickly for us because we were already prepared for it with our data.

32:32

Yeah, makes a lot of sense. So I guess to wrap things up, in sum, I mentioned at the beginning, really do feel like Optimal Blue now, we're in a position to be technological ambassadors for the rest of the industry. So folks who haven't gone through this level of development, education, everything that comes along with trying to make the most of AI within your organization.

32:58

We have lot of lessons learned that we can help share and we're happy to share with lenders who are thinking of or just trying to get a better understanding or better handle on the opportunity for AI within their organization, within other applications. So I'm curious, Shawn, what are some things that you might call out for lenders that they should be thinking about if they're getting started on this AI journey?

33:23

Yeah, absolutely. think, you know, to begin with, think number one is not to be fearful of it. I think, you know, with any new technological shift, there's always a lot of fear and rightfully so, right? Because it's just kind of unknown. But I think even if you begin with kind of basic implementation and, you know, putting proper policies around that and, you know, giving your people access to be able to interact.

33:50

with LLM, whether it's just for basic things like writing an email or let's say you have a bunch of HR documents that people don't want to read, but you can put them somewhere, index them, put an AI on top of it and for it to be able to interact and say, hey, what is our PTO policy or whatever, right? To get people kind of started and thinking about the different ways that they could utilize it within their business. And the other thing is education.

34:20

There is a little bit of a learning curve, right? That comes with it with anything new. And obviously it's much more simplified with LLMs because it's just a natural way that we already interact with things, but that's going to be more of the norm in the future. Right? Like if you think about it today, you know, kids growing up or even people coming out of college, they're not really thinking about this as something that's unique or different, but it's already a part of their life, right? They're already using it. They're already leveraging it.

34:50

So as time moves on, this will just be a natural scale that people will have as they come into the workforce. So if you can leverage that, continues to put you ahead of the trend. Same thing with the internet, right? In the nineties, when you think about that, right? If you were able to think about it in a way of, maybe should we be developing a website? Should we be doing these things? Right? It kind of puts you ahead of the curve when things start taking off more and more. And the other thing is your data, right?

35:20

It's all about your data with AI. if you start to go, let's say you implement those initial things and then you have ideas where you want to go even deeper. It's about making sure that you have your data sets in cleansed and in a place where they're ready to use for AI. And that will also help you, you know, iterate on these things faster as you move forward. Yeah, it makes a lot of sense. The other thing too, I'll throw out there. So last.

35:48

Last week I was in Huntington Beach, had the opportunity to speak at the Mortgage Innovators Conference on a panel about generative AI. And it was mostly framed around how to build customer loyalty using AI. There are a lot of really great speakers. Obviously AI was a major topic at a conference like that. But the keynote speaker, Neil Sahota, he advises the UN on AI and he had a really great talk. One of the things that he mentioned was AI is

36:17

the only revolutionary tool. So we talked about the internet, you mentioned, mobile, APIs, but it's the only new revolutionary tool that can actually teach you how to use it. So the very first step that you can take is asking it, how can I use AI to solve my problems? They can teach you how to do that. So, and obviously that's obvious, right? But it's not necessarily something unless if you're not working with the tool regularly,

36:46

that you can, think about it like that necessarily. And I thought that was a really good point for folks who are just starting off on their journey. But anyways, this was a great conversation, Shawn, really appreciate you popping on and yeah, just really wanted an opportunity to kind of tell that behind the scenes story of how everything came together. Yeah, this has been amazing. I've been mostly a listener for this segment, but this segment was more than I thought it was going to be. So I appreciate the two of you. was, you know,

37:15

I think a great primer for anybody who's interested in how AI is being utilized anywhere today and what not to be afraid of and how to ask the right questions and how to implement it. Then we dabbled a little bit into how Optimal Blue is utilizing AI to give some ideas and to give you an idea of what we're doing with AI. This has been super fun for me. If you're an originator, we have a lot of originator listeners, a broker, a loan originator.

37:44

Definitely rewind a little bit back to a couple of those tools that we're using at OB. And if you're just interested in AI, which I think we all are, listen, you know, listen again to the whole segment. Cause that was, that helped me out with a lot of gaps I had in my understanding of this. and, you know, I'm my team and James's team, more on the capital markets side of things. We also have built some tools in Compass Edge around managing and monitoring for P and L changes and

38:10

and position changes and making trade recommendations and things like that. So again, just really practical use for large language models. And we might do a whole podcast on that at some point. We don't have time today, but this originator facing stuff is just super exciting. And we hope to continue fine tuning it as we go down the road and hopefully talk to some of you all this week at MBA. All right, Shawn, Kevin, thank you so much and we'll see you again soon. All right, let's wrap this thing up.

38:40

James, thanks so much for the market update. Kevin and Shawn, such a great discussion about AI. Lots more to come there and hope to see many of you in New York this week. That's it for today. Join us next week for another episode of Optimal Insights where we'll continue to provide you with the latest market analysis and insights to help you stay ahead. Don't forget to follow us on LinkedIn for more updates and access our latest video episodes. You can also find each episode on all major podcast platforms. Thank you again for tuning into Optimal Insights.