Latest / Chrisman Commentary - Daily Mortgage News / 6.11.26 Pulte and the CFPB; JazzX’s Jagjit Singh on the End-to-End Mortgage; PPI Follows CPI

Transcript

6.11.26 Pulte and the CFPB; JazzX’s Jagjit Singh on the End-to-End Mortgage; PPI Follows CPI · Chrisman Commentary - Daily Mortgage News

• 0:02
  Speaker 1: Welcome to the Chrisman Commentary, Daily Mortgage News Podcast.
• 0:06
  Speaker 1: I'm your host, Robbie Chrisman.
• 0:08
  Speaker 1: Topics on today's episode include the latest, and obviously greatest, on Bill Poulty, after consumer prices yesterday, what do producer prices reveal about wholesale inflation?
• 0:20
  Speaker 1: And my interview with Jazz X's, Jadji Singh, on how close the industry truly is to the end-to-end mortgage.
• 0:29
  Speaker 1: Here, take a listen to a low preview.
• 0:31
  Speaker 1: For the last what seems like basically two decades at this point, the mortgage industry has talked about end-to-end automation, the digitalization of the process.
• 0:42
  Speaker 1: And obviously, we could have a full hour conversation about that, but but your thoughts on where we actually are with end-to-end automation and keeping in mind that we've seen some pushback in terms of maybe borrowers don't want a full automated end-to-end process, and they do like having that human in the loop to use the most cliche term and artificial intelligence currently.
• 1:04
  Speaker 1: But just your your thoughts on where the industry is with end-to-end automation and where you see improvements.
• 1:09
  Speaker: I think there's like two concepts in there that we that you just mentioned, right?
• 1:12
  Speaker: There's one is what is the customer's expectations, and second is the lender, what are they actually doing, right?
• 1:18
  Speaker: So if I hit on the customer expectations first, we've always talked about this omni-channel experience, right?
• 1:25
  Speaker: The borrower can do something on the phone, they can do something via text message, they can do something via an online portal, they can do something via chat, they can do something in in whichever method that they want to.
• 1:37
  Speaker: Right now, with AI, you can actually do that.
• 1:40
  Speaker: You're basically adding an extra one.
• 1:42
  Speaker: So that that you know that used to be chatbot or conversational thing that was just a, you know, I asked this question if I don't understand it, escalated to human.
• 1:49
  Speaker: You can start to actually have a more of a real conversation with the AI, whether that be through text, chat, or voice.
• 1:57
  Speaker: From the borrower's perspective, you still have to give them that experience that they want.
• 2:03
  Speaker: Robbie may want a different experience than Jagjeet, uh, that may want a different experience than Matt, that may want a different different experience than Sarah.
• 2:12
  Speaker: So you have to still give that, the lender still has to give that experience, and AI can only enable that.
• 2:18
  Speaker: Now, when you talk about the end-to-end from the lender's point of view, being a previous lender myself, I've seen all of those different bolt-on technologies that you can get.
• 2:29
  Speaker: You can get different uh, you know, underwriting systems, you can get different condition systems, you can get different asset calculations, you can get different income calculations, you can get different doc management or doc intelligence systems.
• 2:40
  Speaker: You can bolt-on all of these different technologies, but somebody has to manage all of those vendors.
• 2:47
  Speaker: Somebody has to make sure that all of those vendors have the same logic implemented.
• 2:53
  Speaker: Somebody has to make sure that if I change the way that I want to originate a particular loan, or if I want to add a new loan program, somebody has to make sure that that effective impact assessment is done across all of the vendors to make sure that everything can sync and harmonize to have that end experience.
• 3:13
  Speaker: So I have not personally have seen that end-to-end happen in one intelligence system.
• 3:21
  Speaker: And that's what we're trying to build at Jazz X.
• 3:23
  Speaker: How do I make sure that the user or the lender can just manage one intelligence system and Jazz can orchestrate the rest?
• 3:33
  Speaker: Jazz can orchestrate the different technologies that are there, jazz can sync with your LOS, Jazz can sync with your product pricing engines, jazz can sync with your doc management systems.
• 3:42
  Speaker: Jazz can be that layer that there's one Jazz X team to work with to implement that new loan program to make that one overlay change effectively do what you need to do in order to run your business.
• 3:59
  Speaker 1: Thanks to this week's podcast sponsor, Jazz X, the first true end-to-end AI platform built for mortgage.
• 4:05
  Speaker 1: From application to underwriting, Jazz X is a new operating model that helps you scale growth, boost productivity, and transform how your team performs.
• 4:13
  Speaker 1: To learn more, visit jazzx.ai.
• 4:16
  Speaker 1: I was asked if I'd heard about this humility pledge that the CFPB examiners read each supervised entity before conducting exams.
• 4:27
  Speaker 1: What in the world?
• 4:29
  Speaker 1: Yep, nicknamed the Humiliation Pledge.
• 4:32
  Speaker 1: It's been around about six months.
• 4:34
  Speaker 1: I don't know how it helps borrowers, but the CFPB is doing it.
• 4:37
  Speaker 1: Strange times.
• 4:39
  Speaker 1: When President Trump named Bill Poulti, the current head of the Federal Housing Finance Agency, which oversees Freddie and Fanny, as the interim director of national intelligence, a loud collective huh?
• 4:51
  Speaker 1: Could be heard all over Washington.
• 4:53
  Speaker 1: Poulte has no intelligence or national security experience as required by law.
• 4:57
  Speaker 1: He doesn't even have a security clearance.
• 5:00
  Speaker 1: But an intelligence background isn't necessary for the task Trump is apparently handing over to Poulte.
• 5:05
  Speaker 1: Poulte is certainly distracted from overseeing Fanny and Freddie.
• 5:09
  Speaker 1: President Trump said in a truth social post that Bill Poulte would take over as acting director of national intelligence on June 19th, and called for a short-term extension of a foreign surveillance program due to expire at the end of the week.
• 5:22
  Speaker 1: Changing gears slightly, May's consumer price index report largely met expectations, with headline inflation rising 0.5% month over month and 4.2% year over year, driven primarily by higher energy and shelter costs, while core CPI increased a softer than expected 0.2% on the month.
• 5:38
  Speaker 1: Although the annual core inflation rate edged higher to 2.9% from 2.8% in April, underlying price pressures remained relatively contained as core goods prices declined and services inflation moderated, despite a 23% year-over-year increase in energy prices.
• 5:53
  Speaker 1: MBS and Treasury markets initially rallied on the cooler than expected core reading, viewing the data as slightly better than feared rather than downright benign.
• 6:02
  Speaker 1: Obviously, inflation forecasting remains complicated by ongoing geopolitical tensions and elevated energy costs that are likely to persist even if the current supply disruptions ease.
• 6:11
  Speaker 1: The report reinforces the Federal Reserve's ability to remain patient in the near term, while determining whether the economy's resilience can withstand an extended period of restrictive financial conditions without a meaningful slowdown in growth or employment.
• 6:25
  Speaker 1: For today's interview, I wanted to welcome to the show Jazz X's Jajit Singh to talk about how close the industry truly is to the end-to-end mortgage.
• 6:36
  Speaker 1: He's head of product and mortgage lending at Jazz X, where he's focused on both product innovation and strategic leadership.
• 6:44
  Speaker 1: Obviously, people that have listened to this podcast for a while or are familiar with some of the shows we do on the Christmas network, or even read the commentary, will have heard of Jazz X.
• 6:53
  Speaker 1: But y'all are coming on the scene like gangbusters in a lot of ways.
• 6:56
  Speaker 1: And and I guess to some people in the industry, they hear, oh, another AI company.
• 7:02
  Speaker 1: It sounds like I want to take the time to for people to know exactly what Jazz X does.
• 7:07
  Speaker 1: So can you speak about the product and what y'all are trying to accomplish?
• 7:11
  Speaker: Yes, there are a lot of new AI companies coming out here, and I'll uh you know, we we should hit on that.
• 7:16
  Speaker: What's what's different about Jazz X versus some of the new new folks that you see?
• 7:20
  Speaker: Jazz X, we're an AI native uh mortgage application that sits above your existing LOS.
• 7:26
  Speaker: So we're not trying to rip and replace any existing technology.
• 7:28
  Speaker: We're we're sitting above there.
• 7:30
  Speaker: We have purpose-built AI assistants that sit across the entire mortgage lifecycle.
• 7:36
  Speaker: So there's not this bolting on of different applications or different technologies.
• 7:40
  Speaker: We are sitting across your entire mortgage lifecycle from your borrower's first interaction through underwriting, closing, post-close, et cetera.
• 7:49
  Speaker: Every assistant that we have, it automates the day in the life of the persona that it is emulating.
• 7:55
  Speaker: So we have a loan processor assistant, we have a loan officer assistant, we have an underwriting assistant.
• 8:00
  Speaker: So we're emulating their day in the life.
• 8:02
  Speaker: And we'll use you know full file cross-document reasoning and uh produce audit-ready outputs.
• 8:09
  Speaker: We'll make sure that every finding that we're doing is tracing back to guidelines, we'll make sure that every action that a person is supposed to take, Jazz can either take that for them or Jazz can enable or augment them to do it more efficiently and more effectively.
• 8:24
  Speaker: And so, you know, that that that's I think you know what we're trying to do at Jazz X is build that end-to-end ecosystem sitting on top of your existing tech stack so you can get that return on day one.
• 8:36
  Speaker: When you think about the new AI players that are coming out, you know, I was talking to somebody, I won't name names, but I was talking to somebody uh not not a competitor, but I was talking to somebody in the industry that's very well known.
• 8:48
  Speaker: But the he, you know, we were talking about what makes JazX different than these new entrants.
• 8:53
  Speaker: And when you look at the new entrants, they're they're all maybe fresh grads out of Stanford or you know, one of the Ivy League schools.
• 9:00
  Speaker: Uh, they don't have mortgage and uh mortgage industry experience, and they're trying to raise capital.
• 9:05
  Speaker: And those three things, kind of as a combination, is a recipe for not very good.
• 9:12
  Speaker: So just put it that way.
• 9:13
  Speaker: With Jazz X, we have mortgage experience.
• 9:16
  Speaker: I myself, leading product for our mortgage application, have you know, over a decade or so of experience in mortgage operations and mortgage technology in the mortgage business.
• 9:26
  Speaker: We have folks that have 30 plus years, we have staffs of underwriters, we have staffs of different types of people that have lived mortgage or living mortgage today.
• 9:37
  Speaker: Uh, number two, when you talk about funding, we're a company underneath the SAI group, where SAI Group has several different portfolio companies, including like Symfony AI and different different uh different portfolio companies.
• 9:52
  Speaker: And we have a billion dollars backed within SAI group of committed equity.
• 9:56
  Speaker: And so when you look at you know the amount of money that it takes to create this type of technology, it's it's insane.
• 10:03
  Speaker: So I think you know those are those are a few things that stand out in terms of why is JazzX different than your other folks that are you know coming out new in the industry and things like that.
• 10:14
  Speaker 1: Well, since since we talk differentiators, let's talk impact.
• 10:18
  Speaker 1: What is the impact you hope that Jazz X has or that you're working towards?
• 10:24
  Speaker: Yeah, the impact, uh I won't say that I hope, but the impact is where we can automate a significant portion of what users are doing.
• 10:35
  Speaker: So when you look at processing as an example, processing, what are we doing in processing?
• 10:40
  Speaker: We're gathering documents.
• 10:42
  Speaker: We're getting borrower docs, we're getting third-party docs, we're setting up that loan file for underwriting.
• 10:47
  Speaker: And what Jazz is doing is able to, you know, get those docs from the borrowers, get those docs from third parties, review the documents, understand what conditions, additional conditions are needed, understand what conditions can get cleared, understand what underwriting might need, effectively doing 80 to 90% of that role.
• 11:05
  Speaker: Where now you have people, the processors that are there, they're focused on the exceptions, right?
• 11:11
  Speaker: They're focused on those harder use cases, those judgment calls that that come up about 5, 10, 15% of the time.
• 11:18
  Speaker: And so when you talk about the impact, the impact is you get this really big step change in productivity, step change in cycle time, because jazz is doing it 24-7.
• 11:28
  Speaker: I'm not waiting for a human to do it.
• 11:29
  Speaker: And a step change in quality because jazz comes pre-trained with industry guidelines and your overlays across a conventional all-agency guidelines, you know, non-QM, uh DSCR, etc.
• 11:40
  Speaker: You get that step change in quality because jazz knows your underwriting criteria before it gets to underwriting.
• 11:47
  Speaker: The impact is across many different areas, including product productivity, cycle time, quality, uh, borrower experience, et cetera.
• 11:56
  Speaker 1: So when in our conversations, and we've had we've had several outside of this podcast, obviously, you you kind of brought up this term institutionalizing knowledge.
• 12:04
  Speaker 1: And I'd like you to explain what that means to the audience and why it's so important for the mortgage industry.
• 12:12
  Speaker: So when you think about how many loans everybody is processing, right?
• 12:16
  Speaker: There's uh you know, many, you know, several millions of loans that are originating every year, right?
• 12:22
  Speaker: Uh, there's about, I think last number I saw was like 11 or 12 trillion dollars worth of outstanding mortgage debt.
• 12:28
  Speaker: That's a lot of information.
• 12:30
  Speaker: There's a lot of data that is out there across many different types of loans, many different types of borrowers, many different types of properties, many different types of situations and scenarios that real life people go through.
• 12:43
  Speaker: Now, imagine if you can take all of that and put it in a in a memory bank where the next loan that I'm originating, I'm doing it on the backs of understanding what happened in previous loans.
• 12:56
  Speaker: I understand what kind of buybacks were there, what kind of issues I ran into, what kind of delinquency trends I would see.
• 13:04
  Speaker: If you think about you can institutionalize that knowledge, you're effectively getting your best underwriter's knowledge institutionalized into jazz and using that to underwrite your next loan.
• 13:18
  Speaker: And by the way, that's only getting better and better after every single underwrite that's happening.
• 13:24
  Speaker 1: Which is as a quick follow-up, what is the the nirvana at the light at the end of the tunnel when it comes to getting better and better?
• 13:30
  Speaker 1: What's the end game?
• 13:33
  Speaker: The the end game is how do I make sure that I don't have any uh any buybacks?
• 13:36
  Speaker: How do I make sure that all my you know any any loan that I'm originating, I can I can originate it quickly, I can make sure that the borrower gets the funds that they need to purchase that house or to refinance or get that get cash out that they need for that home improvement.
• 13:48
  Speaker: They can get it quickly, they can get it uh more affordably, and the lender can ensure that they're originating a loan with high confidence that's not going to get bought back by the investor.
• 13:58
  Speaker 1: So, this next question either matters incredibly much or zero at all.
• 14:03
  Speaker 1: And that is how is AI actually designed and improved, like taking inside the black box?
• 14:09
  Speaker 1: Because on one hand, I've heard people say, I don't really care about how the algorithm or whatever for Uber Eats works as long as the food gets delivered to my dual.
• 14:18
  Speaker 1: I don't care what what Jazz X or anybody else does, as long as it automates the process or you know unifies tax tasks across the origination chain, whatever it might be.
• 14:27
  Speaker 1: And so I'm I but I would like to know, kind of from a personal understanding, how do we train the AI models?
• 14:35
  Speaker 1: How do uh we we create actual value?
• 14:38
  Speaker 1: What is the feedback process?
• 14:40
  Speaker 1: How where would someone even start if they're trying to design an AI mortgage product?
• 14:45
  Speaker: Yeah, it's uh it's a loaded question, Robbie.
• 14:48
  Speaker: Let me answer it as simply as I as I can.
• 14:53
  Speaker 1: We have a we have a cohort of older listeners to the podcast, but yeah, thank you.
• 14:58
  Speaker: You know, the end user may not care when you're ordering something on Uber Eats.
• 15:03
  Speaker: You may not care how I get the food delivered to me.
• 15:08
  Speaker: But I guarantee you the restaurants that you're ordering the food from do care.
• 15:13
  Speaker: Right?
• 15:14
  Speaker: They want to know when are they gonna get the order uh in.
• 15:17
  Speaker: They're gonna want to know is that order getting confirmed and is that notification going back to the customer?
• 15:23
  Speaker: They're gonna want to know does the customer know that is being delivered, so I don't get a phone call from Robbie saying, Hey, where's my uh where's my pad tie at?
• 15:30
  Speaker: So you may not care what what is happening with the Uber Eats app, but I promise you there's many people that do.
• 15:36
  Speaker: Now with mortgage, it's different, right?
• 15:39
  Speaker: With mortgage, it's even more it's even more.
• 15:41
  Speaker: Mortgage is one of the most regulated industries that we have.
• 15:46
  Speaker: And also, by the way, you Robbie, you might care about how your loan is being underwritten because that is a big purchase for you.
• 15:54
  Speaker: That is your probably the biggest purchase you're gonna make of your life.
• 15:57
  Speaker: So you want to know how is it gonna, how is it gonna come about?
• 16:00
  Speaker: How is it going to uh affect my monthly payment?
• 16:03
  Speaker: How is it gonna affect my cash to close?
• 16:05
  Speaker: You're gonna you're gonna care about all of that.
• 16:07
  Speaker: And on top of that, the lenders are gonna care too, because you have all of these different regulating bodies, whether that be at a state, local, or federal level, that are gonna come in and say, How did you make that decision?
• 16:19
  Speaker: How did you make sure that you didn't make a different decision for Robbie that you made for Jijeet, even though their credit profiles are the same?
• 16:27
  Speaker: And so I think it's super critical to make sure that you have at least two things in there, right?
• 16:35
  Speaker: One is can the AI system that you're using be fully transparent about how the decision is being made, what decision is being made, and can it explain that to you?
• 16:45
  Speaker: And and number two is how is your AI being as deterministic as possible?
• 16:54
  Speaker: Because we know that AI is a probabilistic system, but you need it to be deterministic so you're not making that disparate impact.
• 17:01
  Speaker: And at Jazz X, you know, we have that technology, and I'm happy to show show any lender or anybody that's out there how we are making sure that that decision is deterministic and not probabilistic where you may run into issues.
• 17:15
  Speaker: And that's it's gonna be based off of that, how we orchestrate the LLMs, how we orchestrate the technology, and how we make sure that that is fully transparent to the lender and the lender has full control.
• 17:30
  Speaker 1: We started the interview by talking about all these AI companies that are popping up, for lack of a better term.
• 17:34
  Speaker 1: Obviously, I I do think I personally believe in Jazz X, so so I'm not I'm not lumping you into that.
• 17:40
  Speaker 1: But there's all these AI companies popping up in the mortgage industry.
• 17:44
  Speaker 1: How does an AI company get from idea to viable product?
• 17:49
  Speaker 1: What is the coding that's happening?
• 17:50
  Speaker 1: What is the actual like how do you train a model?
• 17:52
  Speaker 1: Where do you are you going to Chat GPT?
• 17:54
  Speaker 1: And I mean, obviously, no, that's I'm I'll edit that part out, but but how do you get from idea to product in if you're making an AI mortgage company?
• 18:04
  Speaker: You have the best talent in the world.
• 18:08
  Speaker: I don't I can't even count on, I don't even know.
• 18:12
  Speaker: I don't even think I can count on two hands how many PhDs in just AI and learning and knowledge and reasoning that we have in Jazz X.
• 18:20
  Speaker: We have assembled a team of engineers, an engineering team that come from the largest company, largest tech companies, you know, Google, Oracle, ServiceNow, uh, Amazon, Facebook, Microsoft.
• 18:36
  Speaker: We also have the same people plus others that come from that startup environment that understand how they can leverage the technology to build better, faster, and with higher quality.
• 18:47
  Speaker: Uh so you know, without sparing you those those technical details, I think it's you have to have the right engineering team and you have to have the right funding in order to be able to hire that team and use these LLMs and use those tokens to actually build what you need to build and test and validate and and ensure you're driving value.
• 19:07
  Speaker 1: It's an interesting time, certainly, because for a while those in Silicon Valley have said, I can disrupt mortgage, it's a multi-trillion dollar year industry, it's ripe for disruption.
• 19:15
  Speaker 1: Look at these antiquated processes.
• 19:17
  Speaker 1: And for a while, and I know this firsthand at SoFi in the 2010s when we were trying to, it wasn't as easy to disrupt as one would think.
• 19:25
  Speaker 1: And so when you're answering that question, I'm going, well, don't forget the institutional knowledge about mortgage that can help.
• 19:29
  Speaker 1: But at the same time, I do think that it takes this kind of from left field, from the side approach of I understand engineering and I'm I'm very smart in my ways.
• 19:39
  Speaker 1: Let me apply, let me just think about the problem removed from it rather than being so inside of it that I can't really see the forest through the trees.
• 19:48
  Speaker 1: And so I'm I'm excited about what the mortgage industry is evolving into.
• 19:51
  Speaker 1: Obviously, you are a huge part of that.
• 19:53
  Speaker 1: Obviously, Jazz X is going to be a huge part of that.
• 19:56
  Speaker 1: I really appreciate the time, man.
• 19:56
  Speaker 1: I enjoyed this a lot.
• 19:58
  Speaker: This is great, Robbie.
• 19:58
  Speaker: I appreciate you uh having me on here and always happy to have a chat about Jazz X and and about mortgage in general.
• 20:08
  Speaker 1: After receiving consumer prices yesterday, today's economic calendar kicked off with producer prices.
• 20:13
  Speaker 1: May PPI was up 1.1%, stronger than expected.
• 20:17
  Speaker 1: Year over year was up 6.5%, course CPI was up 0.4% as expected, but up 4.9% year over year.
• 20:24
  Speaker 1: We've also received weekly jobs claims in at $229,000, about as expected, and later today brings a $22 billion 30-year treasury bond auction.
• 20:33
  Speaker 1: Additionally, the European Central Bank raised interest rates today for the first time since 2023 by a quarter point to 2.25%.
• 20:41
  Speaker 1: In doing so, it became the first major central bank to hike in response to the Iran conflict.
• 20:47
  Speaker 1: After the producer inflation numbers, Agency MBS prices are a shade better than Wednesday's close, the two years yielding 4.15, and the ten years yielding 4.55 after closing yesterday at 4.54%.
• 20:58
  Speaker 1: Let's wrap up with a joke and some housekeeping.
• 21:04
  Speaker 1: What do you get if you ask a politician to tell the truth, the whole truth, and nothing but the truth?
• 21:09
  Speaker 1: Three different answers.
• 21:11
  Speaker 1: Thanks again to Jazz X, the first true end-to-end AI platform built for mortgage.
• 21:19
  Speaker 1: From application underwriting, Jazz X is a new operating model that helps you scale broke.
• 21:24
  Speaker 1: Jazz X is a new operating model that helps you scale growth, boost productivity, and transform how your team performs.
• 21:30
  Speaker 1: To learn more, visit jazzx.ai.