11.6.26 Bipartisan Opportunities; Crittora's Doug Long on AI Governance; Credit Modernization의 전체 녹취록

Freddie, may be the only investors taking only a vantage score. 00:04:42.319 --> 00:04:50.959 I expect most, if not all, private investors will have a different price for vantage loans to account for the estimated 10 to 20 plus point score inflation of vantage versus FICO. 00:04:51.439 --> 00:04:54.399 But your readers should be aware of a potential industry dilemma. 00:04:54.639 --> 00:05:02.000 First, lenders will want to run all three scores when 10T comes out in January for HUD loans, and then cherry pick the best score to give the most benefit to borrowers. 00:05:02.240 --> 00:05:06.480 Second, investors will have huge pressure to pick the lowest of the three scores presented. 00:05:06.720 --> 00:05:16.480 They cannot unsee that there's a lower score in FILE, and the data will show a loan with a high vantage and a low FICO will perform worse than a loan with the same vantage, but a medium or high FICO. 00:05:16.639 --> 00:05:23.600 If lenders game the system but then rerun credit with only the higher scores, private investors will quickly build in a gaming adjustment for those lenders. 00:05:23.839 --> 00:05:28.560 I think that the future state is most lenders pick only Vantage or 10T. 00:05:28.879 --> 00:05:36.319 Vantage will be able to keep their price low since they don't need to generate any profit, and FICO will have to charge a score fee since this is their only income. 00:05:36.480 --> 00:05:41.839 The market will then decide if 10 T is more predictive, and if by enough to justify paying a higher price. 00:05:49.360 --> 00:05:49.680 U.S. 00:05:49.759 --> 00:06:05.040 Treasuries, and to a lesser extent, Agency MBS rallied on Thursday as escalating volatility in AI-related equities triggered a broad flight to safety, despite stronger than expected economic data, including low initial jobless claims, robust productivity growth, and declining unit labor costs. 00:06:05.360 --> 00:06:17.920 The resulting demand for safe haven assets pushed treasury yields to their lowest levels since mid-September, with the tenure yield extending its decline by more than 50 basis points from its October 7th peak, as equity market weakness continues to weigh on investor sentiment. 00:06:19.920 --> 00:06:32.720 For today's interview, we wanted to welcome to the show Kratora's Doug Long to talk about enabling secure, controlled AI adoption that helps organizations manage regulatory requirements, mitigate risks from unauthorized AI use, and operationalize AI strategies. 00:06:33.199 --> 00:06:45.040 He's CEO at Kratora, which secures AI agents by enforcing execution time authorization before agents can use tools, call APIs, or change systems of record. 00:06:45.360 --> 00:06:53.759 Kratora is the execution time authorization layer for AI agents, a cryptographic enforcement layer that sits between agents and the tools, APIs, and systems they're trying to use. 00:06:54.160 --> 00:06:57.680 Kratora is mortgage AI governance. 00:06:58.160 --> 00:07:05.279 It adds

action-level control and reviewable evidence to help mortgage organizations move consequential AI workflows toward governed production. 00:07:05.519 --> 00:07:08.240 You've been in the industry a long time. 00:07:08.480 --> 00:07:15.920 How did you get to this point of deciding that mortgage AI governance was the opportunity you wanted to pursue? 00:07:16.240 --> 00:07:24.000 One of the things that uh, you know, like I said, I've been in the mortgage banking business for four decades and have been through a lot of different cycles. 00:07:24.160 --> 00:07:31.680 And it kind of reminds me back in the late 1990s where we had this thing called the internet that was going. 00:07:31.759 --> 00:07:35.759 You know, it was all all things about dot com and people getting websites. 00:07:35.839 --> 00:07:49.040 And I know that kind of sounds juvenile today, but back then that was a reflection point that was going on in the industry that it was like, wow, okay, there's there is uh there's different ways of looking at the business today. 00:07:49.279 --> 00:07:55.839 And I remember a lot of mortgage companies back then were not, you know, they were like anti even putting a website up. 00:07:56.079 --> 00:08:07.839 And as as the people understanding the power of uh digital advertising, the power of search, the power of being able to put intranets up for communication protocols in your company. 00:08:08.000 --> 00:08:14.800 The people who adopted that early on seemed to have a substantial advantage over those who didn't. 00:08:14.959 --> 00:08:25.519 In fact, if you kind of look at what where Rocket Mortgage is today, they were the first guys to really kind of become the first internet lender and they sold into it, and the rest is history. 00:08:25.680 --> 00:08:27.519 You get, you know, what happened there. 00:08:27.759 --> 00:08:36.639 But you kind of look at today, you fast forward today, and I see what's happening in the mindset of what's going on similar. 00:08:36.799 --> 00:08:38.559 It's a very similar scenario. 00:08:38.799 --> 00:08:54.639 Because this is a time where I think a lot of mortgage companies and banks and credit unions, et cetera, could be sitting down saying, I can redefine, I can rethink, I can reimagine what my company can be with utilizing this technology. 00:08:54.879 --> 00:09:01.440 You know, at the end of the day, the mortgage business always comes down back to relationships and doing the right thing to the for the consumer. 00:09:01.759 --> 00:09:07.360 But uh the question is, you know, what could AI do to actually maybe drive down the cost of the consumer? 00:09:07.519 --> 00:09:10.480 What could it do to create more efficiencies within my company? 00:09:10.639 --> 00:09:11.679 What can it do? 00:09:11.840 --> 00:09:24.480 And I think a lot of the the mortgage folks out there and people I know, a lot of them are kind of having that same thing that happened 20-something years ago where they're thinking, wow, is this really something I've got to do? 00:09:24.559 --> 00:09:28.320 I'm I know it's coming, but should I embrace it? 00:09:28.480 --> 00:09:32.559 You know, I know I've got you know a lot of people using AI outside my company. 00:09:32.720 --> 00:09:34.639 I don't even know what's really going on. 00:09:35.600 --> 00:09:41.600 And it's it's to the point where people know that AI is going to change our industry. 00:09:41.840 --> 00:09:44.000 The question is, how is it going to change it? 00:09:44.559 --> 00:09:47.679 And how's it going to change your mortgage strategy? 00:09:47.840 --> 00:09:50.080 How's it going to change the way I do business? 00:09:50.240 --> 00:09:52.240 How, you know, how, how, how, how. 00:09:52.399 --> 00:09:53.919 So it's all about how right now. 00:09:54.080 --> 00:09:59.279 And one of the things that that came into play is uh being in this industry for so long. 00:09:59.440 --> 00:10:08.960 I was one of the companies, uh I was uh president of sales for Prospect Mortgage back when uh it had its first CFPV audit. 00:10:09.120 --> 00:10:17.519 And, you know, understanding the need for being audit ready was something that companies really didn't have in play. 00:10:17.679 --> 00:10:33.279 So I I kind of looked at, you know, everything from the beginning of like 1990, the birthing of the internet when internet started getting strong to the regulatory environments that we went through back after uh Todd Frank and after the Great Recession. 00:10:33.519 --> 00:10:38.080 I was like, man, we know that there's going to be a change. 00:10:38.240 --> 00:10:43.519 And so the question where is where we started at playing is like, where would we fit? 00:10:43.600 --> 00:10:54.080 And how could we actually get to a position to where we could play in the game of uh making AI operational, being able to be agnostic to what AI you use. 00:10:54.240 --> 00:10:58.320 We're not gonna pick an AI, you pick your own AI, you reimagine what you want to do. 00:10:58.639 --> 00:11:11.840 At the same time, we said you've got to be able to do this efficiently, you got to be able to do it securely, and you've got to be able to make sure that whatever you do on your AI, you've got to have an audit-ready trail. 00:11:12.080 --> 00:11:15.200 Because I knew eventually regulators were would get involved. 00:11:15.279 --> 00:11:17.039 So that's when we started building it. 00:11:17.279 --> 00:11:19.759 We started building it with those things in mind. 00:11:20.159 --> 00:11:28.159 We didn't want to be you had to be this kind of vendor or that kind of, you know, we are vendor agnostic and we're also AI agnostic. 00:11:28.240 --> 00:11:38.480 But what we do is we kind of create a command and control center for you to reimagine your company and be able to implement AI in a safe, secure method and allowing humans to stay in control. 00:11:39.279 --> 00:11:59.039 And at the same time, make sure that there's an audit trail of every every one of the AI actions that occurred

within your company so that when the board members or your executive committee or your internal audit team, or if any maybe auditor or somebody from the state shows up and says, Can you prove what that AI did? 00:11:59.360 --> 00:12:02.159 We have an audit log that says this is what happened. 00:12:02.399 --> 00:12:13.840 So that's how we kind of started thinking about Kratora is we know that there's going to be a game-changing wave that of how AI is going to play in the mortgage industry. 00:12:14.399 --> 00:12:31.600 We just thought, how do we make sure that we sit down and create a secure methodology for the mortgage company, crediting or bank that wants to implement AI in a way to be able to do it very efficiently, very securely, and make sure they're proof prove what happened. 00:12:31.919 --> 00:12:33.440 So that's how we started, Robbie. 00:12:34.080 --> 00:12:42.080 Why governance, considering all the possibilities that AI presents in terms of the type of company you could create? 00:12:42.159 --> 00:12:43.600 Why do you choose governance? 00:12:44.000 --> 00:12:56.000 Well, if you look at uh the governance side of things, that kind of if you if you look at what's really driving all mortgage companies, we have a boss that's called Fannie Mae or Freddie Mac, that's where our liquidity is, right? 00:12:56.080 --> 00:12:59.440 And those guys provide most of the liquidity to the mortgage industry. 00:12:59.759 --> 00:13:15.840 And if you kind of look at what Fannie Mae has done, they've said and Freddie Mac, they both have said, we need to make sure that if you're doing something in AI, you need to have a governance policy around it. 00:13:16.080 --> 00:13:20.399 And so we're not the guys that are going to create your government's governance policy. 00:13:20.559 --> 00:13:25.279 There are great people out there like James Brody and others that can help you do that. 00:13:25.919 --> 00:13:32.799 What we decided to do was we need to be able to actually help you execute on that policy. 00:13:32.960 --> 00:13:47.759 Whatever policy you decide to put in place, and you've got to document that policy, you got to make sure that everybody in your company, you've got to have designate who's involved in that policy, and you got to make sure that you're you're the person that is responsible for it. 00:13:47.840 --> 00:13:51.039 So if you got designated, somebody's gonna be responsible in your company for that. 00:13:51.279 --> 00:14:04.000 So we know that we're not if you look at the kind of the top 10 things that Fannie Mae's requiring of at this January 18th uh requirement that's coming out soon. 00:14:04.320 --> 00:14:07.759 We fit in like six, seven, and eight. 00:14:07.840 --> 00:14:22.000 Okay, we're we're the you know, we're we operationalize your AI, we secure your AI, we can we prove what your AI did, but you need to be able to put the whole team together when you're putting together your AI governance strategy or AI strategy. 00:14:22.159 --> 00:14:23.840 You need a whole team that's gonna do that. 00:14:24.000 --> 00:14:28.720 And what we're finding is that's kind of why we're partnering with other people that do other things. 00:14:28.879 --> 00:14:31.440 We're we're not gonna set your AI governance policy. 00:14:31.679 --> 00:14:39.120 What we can do is we can document it, help you document it, we can help you operationalize it, and we can uh and we can audit it for you. 00:14:40.080 --> 00:14:45.840 Let's talk best practices for mortgage, AI, governance. 00:14:46.480 --> 00:14:49.039 Why do people need something like Kratora? 00:14:49.200 --> 00:14:51.120 Why should people trust Kratora? 00:14:51.679 --> 00:14:52.799 Let's start with the why. 00:14:52.879 --> 00:15:02.559 All right, so you you've got a lot of a lot of lenders I've talked to are just waiting for their vendors to kind of create their own AI strategy for me. 00:15:02.720 --> 00:15:08.240 I'm waiting for Encompass or I'm waiting for my LOS to create an AI for me. 00:15:08.559 --> 00:15:09.840 I'm waiting around. 00:15:10.080 --> 00:15:20.159 In the meantime, they've got their loan officers or processors and other people who are using Cloud or they're using OpenAI or they're using other other things, and we call that shadow AI. 00:15:20.960 --> 00:15:30.799 So things are going on that you probably don't even work you're not even aware about, and it's maybe outside your network, but that doesn't mean that you're not responsible for it. 00:15:30.960 --> 00:15:40.480 And that so at the end of the day, you're kind of responsible for any of your employees' use of AI, you're responsible for your third-party vendors' AI, what they're doing. 00:15:40.720 --> 00:15:51.919 And so, you know, when you start realizing that, you know, what the governance requirements are, most people I don't believe actually realize that they need that they're responsible for all this. 00:15:52.080 --> 00:15:55.440 And so that's kind of when you were asking originally why governance. 00:15:55.600 --> 00:15:58.000 Well, it's a requirement, number one. 00:15:58.159 --> 00:16:01.840 And number two, the question is, do I get do I need governance? 00:16:01.919 --> 00:16:02.240 Yes. 00:16:02.480 --> 00:16:03.919 Do I have an AI strategy? 00:16:04.080 --> 00:16:04.480 Maybe. 00:16:04.639 --> 00:16:06.159 I don't know if you do or not. 00:16:06.399 --> 00:16:11.039 And the same time, the question is, do I want to implement an AI strategy? 00:16:11.200 --> 00:16:16.639 And so what Kratora does is we actually help in terms of implementing your AI strategy. 00:16:17.039 --> 00:16:27.679 We utilize our encryption technology to actually make sure that uh the whole technology is locked down in terms of your security for your AI agents that you're using. 00:16:28.000 --> 00:16:41.519 In fact, we we're able to be able to make sure that every AI agent you decide to put in place, we actually just don't allow for them to go wandering around because you don't even give

them the tools to do that. 00:16:41.679 --> 00:16:50.720 So you've heard AI agents and horror stories of where they and maybe they started in a uh CRM, and next thing you know, they're in the HR database and they're just trying to be helpful. 00:16:50.879 --> 00:17:01.919 Well, by the way, what we use at Cratora, we don't even allow that to happen because you you can sit down and give access to the agent, the AI agent what it can see and what it can't see. 00:17:02.000 --> 00:17:05.440 It's almost like having blinders on an AI, so it's very focused AI. 00:17:05.839 --> 00:17:10.400 And that doesn't allow for them to go wandering around in rooms they're not supposed to be in. 00:17:10.640 --> 00:17:13.039 And so that's one of the reasons we created it. 00:17:13.200 --> 00:17:19.519 And the second thing is that we have probably on staff with you know over 100 years of expertise in the mortgage industry. 00:17:19.680 --> 00:17:28.079 So we know mortgage, we understand mortgage, and that's another reason I think when you said white trust Kratora, uh, we kind of know the business. 00:17:28.960 --> 00:17:35.039 People in the Chrisman LLC ecosystem have certainly seen Kratora everywhere. 00:17:35.200 --> 00:17:36.400 You burst onto the scene. 00:17:36.480 --> 00:17:41.359 Can you talk about your plans for the next couple of months, six months, year? 00:17:41.920 --> 00:18:00.480 I can tell you one of the things that I've found is that we want to, number one, we want to be able to partner with any company who wants to have an AI strategy that has a desire to make sure that that's uh how can I operationalize it fast and secure and provide auditable trail. 00:18:00.640 --> 00:18:01.920 That's that's where we play. 00:18:02.160 --> 00:18:16.160 But we also know that we need to be able to put together partners, and when I say partners, like strategic alliances with other like-minded people who have the same desire that play in different spots within the OAI strategy. 00:18:16.240 --> 00:18:31.039 And so we're putting together, you know, groups of people that are involved in MISMO or involved in uh the frame project with MISMO, which we're one of the proof consultants of MISMO, in order to be able to have a place where people can go, where do I find? 00:18:31.440 --> 00:18:35.200 I want to start an AI, I want to get into AI, but how do I do it? 00:18:35.359 --> 00:18:38.400 There's so much noise around it, there's so many people around it. 00:18:38.559 --> 00:18:50.960 And so what we're trying to find is the experts in maybe every position that you need in order to actually decide, I want to get into AI, or I decide I want kind of AI strategy I want to get into. 00:18:51.359 --> 00:19:13.279 So everything from consultants to start out from strategic planning perspective of like how do I get into AI to the position of how do I make sure that as different rules change or different municipality or state changes the rules on AI, or there's some legal documentation, how do I get in, how do I make sure I have access to that? 00:19:13.599 --> 00:19:16.799 To the point of being able to say, how do I operationalize it? 00:19:16.880 --> 00:19:17.599 That would be us. 00:19:17.680 --> 00:19:25.839 You know, how do we help you operationalize the AI and how do we document it uh for with governance packs to be able to provide any auditor? 00:19:26.319 --> 00:19:32.880 And then we're also looking at a couple of AI agent builders that as people like need help maybe in terms of building out their AI. 00:19:33.039 --> 00:19:45.680 So so we're in the middle trying to put together uh over the next next 60 days, we're putting together a team of different people that fill that different spots within a strategy session that you would you would want to implement. 00:19:45.920 --> 00:19:48.160 And so that's what we're doing now. 00:19:48.319 --> 00:20:08.880 And over the next 12 months, we want to be able to uh be known as the people to help secure down AI agents and provide uh provide audible tract for any uh anybody that needs to make sure that what those AI agents did for the mortgage industry, any any anybody doing mortgage lending, whether it's mortgage lenders, banks, or credit unions. 00:20:09.599 --> 00:20:10.319 Very good. 00:20:10.880 --> 00:20:12.559 Anything else you want to touch on before we finish? 00:20:12.720 --> 00:20:14.240 Do you feel like that was comprehensive enough? 00:20:14.640 --> 00:20:29.359 I will tell you uh one of the biggest things we're also doing, just uh FYI, we've come up with a quantum ready encryption product so that when quantum computers hit, we're in the middle talking to uh a lot of the federal agencies about it. 00:20:29.519 --> 00:20:36.880 When quantum computers come, the the current security loads are gonna not be close to being able to keep out the bad actors. 00:20:37.119 --> 00:20:42.640 We will eventually add that quantum security into the product we're just talking about as well. 00:20:42.960 --> 00:20:44.240 Doug, ton of good stuff, man. 00:20:44.400 --> 00:20:46.480 Wish you the best of luck, and I'm sure we'll talk soon. 00:20:46.559 --> 00:20:46.880 Thank you. 00:20:47.119 --> 00:20:48.960 Okay, buddy, appreciate it. 00:20:51.279 --> 00:21:02.400 Mortgage prepayments slowed in October, with Fannie Mae and Jenny Mae speeds declining 9.2% and 14.8% month over month, respectively, as elevated mortgage rates continued to suppress purchase and refinance activity. 00:21:02.559 --> 00:21:16.400 Although 30-year mortgage rates have declined for three consecutive weeks to 7.18%, they remained significantly above year-ago levels, suggesting prepayments will remain subdued into the seasonally slower year end, though refinancing declines may begin to moderate. 00:21:16.640 --> 00:21:27.440 With approximately 0.5% of mortgages currently offering sufficient refinancing incentives, there's potentially attractive risk reward in servicing heavy originators such as Rocket and Rhythm. 00:21:27.839 --> 00:21:38.319 Today's economic calendar kicked off with October non-infarm payrolls, which were up 165,000 versus 65,000 expectations and a prior reading of 29,000. 00:21:38.559 --> 00:21:47.119 The unemployment rate declined to 4.1% from 4.2%, and average hourly wages were up 2.0%. 00:21:47.839 --> 00:21:56.160 Despite a relatively tight labor market, wage growth has slowed to 3.1% year over year as surging employer health care costs increasingly crowd out salary increases. 00:21:56.319 --> 00:22:01.680 Later today brings preliminary November University of Michigan consumer sentiment and September consumer credit. 00:22:01.839 --> 00:22:02.480 With the U.S. 00:22:02.559 --> 00:22:12.319 equity sell-off appearing to be easing, we begin the day with agency MBS prices, little change from Thursday's close, the two-year yielding 4.49, and the 10-year yielding 4.86% after closing yesterday at 4.84%. 00:22:14.640 --> 00:22:17.039 Let's wrap up with a joke and some housekeeping. 00:22:17.279 --> 00:22:20.640 Three brewery presidents walk into a bar on a Friday afternoon in November. 00:22:20.799 --> 00:22:23.759 One from Corona, one from Corus, one from Guinness. 00:22:24.000 --> 00:22:26.640 The president from Corona says to the bartender, give me a Corona. 00:22:26.799 --> 00:22:28.160 The best Mexico has to offer. 00:22:28.319 --> 00:22:29.680 The bartender hands one. 00:22:29.920 --> 00:22:36.319 Next, the Corus president orders a coors saying, Hand me the only beer in the world made with water fresh from the Colorado Rockies. 00:22:36.559 --> 00:22:39.599 Then the Guinness president walks up to the bar and orders a Coca-Cola. 00:22:39.839 --> 00:22:42.319 The bartender, a bit taken aback, hands him what he orders. 00:22:42.880 --> 00:22:45.839 The other brewery presidents turn to him and ask, Why didn't you order a Guinness? 00:22:46.160 --> 00:22:49.599 He replies, Well, I figured if you guys weren't ordering beers, why should I? 00:22:52.640 --> 00:22:55.839 Thanks again to First Close for sponsoring this week's podcasts. 00:22:56.000 --> 00:23:07.599 First Close helps financial institutions grow a successful home equity business with innovative solutions that minimize application drop-off, automate time-consuming processes, and reduce time to close from weeks to days. 00:23:07.759 --> 00:23:10.240 To learn more, visit firstclose.com.