WEBVTT
00:00:02.399 --> 00:00:05.519
Welcome to the Crisman Commentary, Daily Mortgage News Podcast.
00:00:05.839 --> 00:00:07.839
I'm your host, Robbie Crisman.
00:00:08.080 --> 00:00:13.439
Topics on today's episode include some shifts in the industry as we approach 2027.
00:00:14.240 --> 00:00:32.000
It's a big day for Fed presidents speaking to the media, and my interview with Gateless's Mike Brown on how AI is overcoming legacy technology and operational friction to improve risk management, accelerate time to close, demonstrate ROI, and ultimately enable a scalable no-touch underwriting experience.
00:00:32.240 --> 00:00:34.159
Here, take a listen to a little preview.
00:00:34.399 --> 00:00:42.079
What are the biggest operational friction points you see when traditional mortgage lenders try to adopt real-time AI document and condition clearing?
00:00:42.479 --> 00:00:44.799
Uh it's honoring the automation candidly.
00:00:44.880 --> 00:00:49.119
Like it the hardest thing is to get even you can do everything up front.
00:00:49.200 --> 00:00:52.000
You can have your loan officer get approvals.
00:00:52.159 --> 00:00:57.439
About 20% of our loans will get approved for credit income and assets before a human touches it.
00:00:57.759 --> 00:00:59.600
End of the day, though, a human's going to touch it.
00:01:00.079 --> 00:01:03.359
A processor is going to look at the loan, an underwriter is going to look at the loan.
00:01:03.679 --> 00:01:10.400
Getting that person, first the loan officer, to say, hey, that income number works with the loan file.
00:01:10.480 --> 00:01:11.120
Don't touch it.
00:01:11.200 --> 00:01:11.920
It's insured.
00:01:12.000 --> 00:01:12.719
You can move on.
00:01:12.879 --> 00:01:15.599
Or you have rep and warrant eligibility or something like that.
00:01:15.760 --> 00:01:16.640
That's a challenge.
00:01:16.719 --> 00:01:23.920
And then as you go further downstream, where you look at that or down the chain in terms of like, okay, now we're going to processing, now we're going to underwriting.
00:01:24.000 --> 00:01:28.239
Before we go to closing, that underwriter is paid and has been paid to find defects.
00:01:28.400 --> 00:01:33.359
Getting them to just trust the engine is the hardest part of the of the engagement.
00:01:33.599 --> 00:01:38.000
My fear is that regulators can't keep up.
00:01:38.319 --> 00:01:45.439
And two examples would be look at like when Mark Zuckerberg first went to capital, all these people don't even know what social media it like.
00:01:45.519 --> 00:01:45.920
How are they?
00:01:46.000 --> 00:01:47.920
Who are they to figure out how to regulate AI?
00:01:48.319 --> 00:01:51.840
Then I I mean my background to a degree is in bicycling.
00:01:52.159 --> 00:01:54.239
You talk Lance Armstrong doping scandal.
00:01:54.319 --> 00:02:01.280
The people getting paid to break the rules, the people getting paid to create these AI, they're always going to make more than the people trying to enforce the rules and regulate.
00:02:01.519 --> 00:02:03.760
So I'm like, I feel like there'll always be one step.
00:02:03.920 --> 00:02:12.400
And so yeah, it portends for bad things the amount, and that's the other thing because I've been asked this too a lot, which is are you using agentic AI?
00:02:12.639 --> 00:02:16.319
We use it to help develop the product, but is it in the product that I'm consuming?
00:02:16.560 --> 00:02:16.879
No.
00:02:17.199 --> 00:02:17.919
Okay, good.
00:02:18.080 --> 00:02:18.319
Why?
00:02:18.479 --> 00:02:19.919
Why are you so concerned about this?
00:02:20.080 --> 00:02:24.719
This happened three times after the um Tim Cook said the Apple iPhones are gonna be two grand.
00:02:25.280 --> 00:02:28.719
Well, we're concerned that tokens just go through the roof.
00:02:28.960 --> 00:02:37.840
Energy prices, as energy prices go up, the cost to produce AI goes up, and all of a sudden somebody's gonna turn around, they're like, oh, that was 20 bucks alone, now it's 50.
00:02:38.000 --> 00:02:41.039
And I can't live without that tool anymore because I fired all the people.
00:02:41.199 --> 00:02:43.599
So, like, that's a that's a real factor too.
00:02:43.759 --> 00:02:44.800
Um, the cost piece.
00:02:44.879 --> 00:02:45.680
It's crazy.
00:02:48.159 --> 00:02:50.639
Thanks to this week's podcast sponsor, Gateless.
00:02:51.039 --> 00:02:53.759
Intelligent automation that gives you the competitive edge.
00:02:54.319 --> 00:03:03.039
Gateless solutions reduce costs, deliver a superior borrower experience, and mitigate risk by automating tasks and decisions historically made by people.
00:03:03.439 --> 00:03:06.080
To learn more, visit gateless.com.
00:03:08.479 --> 00:03:09.840
Welcome to the fourth quarter.
00:03:10.400 --> 00:03:16.159
Remember when, in twenty twenty four, the informal slogan for many in our business was stay alive until twenty-five.
00:03:17.280 --> 00:03:20.080
Well, twenty-seven isn't expected to be heaven.
00:03:21.120 --> 00:03:28.479
This is the month of Halloween, and suddenly residential lending is filled with tricks or treats, along with conferences of varying value.
00:03:29.439 --> 00:03:30.400
People are on the move.
00:03:30.960 --> 00:03:38.240
The job section of our commentary has been filled with transitions and promotions, and on a larger scale, the pace of company changes is picking up.
00:03:39.199 --> 00:03:50.319
Just when people are wondering if the shift into non-agency products is becoming long in the tooth, New York Life Investment Management is taking majority control of Invictus Capital Partners, the company behind Veris.
00:03:51.039 --> 00:04:02.800
The asset manager has$838 billion under management, and Veris is already having its biggest securitization year yet, with about$8.9 billion across 14 deals in 2026.
00:04:03.120 --> 00:04:12.719
Meanwhile, in agency news, keeping track of 38-year-old FHFA director Bill Poulte's tweets, impacting thousands of lenders and millions of borrowers, it's become a full-time job.
00:04:13.759 --> 00:04:21.439
He tweeted yesterday that Fannie and Freddie are hereby moving to one pricing grid with vantage score joining the existing FICO classic pricing grid.
00:04:22.240 --> 00:04:35.680
But wait, the FHFA, Fannie and Freddie announcements that approved lenders may apply for a temporary exception to the November 2nd Uniform Appraisal Data Set or UAD 3.6 delivery deadline.
00:04:36.160 --> 00:04:43.040
A move that follows concerns MBA and its members raised with GSEs about readiness across the broader appraisal ecosystem.
00:04:51.120 --> 00:04:56.079
Unswayed by the Treasury announcing plans to purchase up to six billion dollars of bonds today.
00:04:56.399 --> 00:05:06.319
The bond market endured a steady catalyst light sell-off yesterday, driven less by fundamentals than by technical and flow-driven selling that accelerated once key yield levels broke.
00:05:07.439 --> 00:05:09.600
Economic data offered conflicting signals.
00:05:10.079 --> 00:05:15.839
The Fed's preferred inflation gauge, core PCE, undershot expectations at three point zero percent year over year.
00:05:16.000 --> 00:05:24.000
Meanwhile, personal spending surged 0.9%, ADP rose 90K, and Q2 GDP was revised sharply higher to 2.2%.
00:05:25.040 --> 00:05:27.759
Yet none of it materially altered the market's direction.
00:05:28.959 --> 00:05:36.000
Instead, yields rose again, continuing the yield curves bare steepening, with the two's ten spread finishing around forty basis points.
00:05:36.959 --> 00:05:42.480
Mortgage backed securities also weakened with prices down roughly a quarter point, and spreads mostly wider.
00:05:42.959 --> 00:05:47.680
Although fifteens and thirties outperformed and late-day recovery softened the overall damage.
00:05:48.399 --> 00:05:53.920
The current coupon rose five basis points to six point four zero percent amid elevated trading volumes.
00:05:54.879 --> 00:06:00.079
Duration rather than spread drove the session, with technical selling proving capable of sustaining itself.
00:06:01.199 --> 00:06:05.839
Today's Treasury buyback and tomorrow's payrolls report stand out as the next potential market catalysts.
00:06:24.959 --> 00:06:31.839
He's a seasoned mortgage executive, excelling at developing and executing strategies that drive significant business growth.
00:06:32.000 --> 00:06:38.240
His expertise lies in leading cross-functional teams to achieve and exceed revenue targets through innovative planning and market analysis.
00:06:38.720 --> 00:06:42.800
He brings two decades of experience in the mortgage technology space to Gateless.
00:06:43.920 --> 00:06:52.560
I think a good place to start would be the technology landscape in the industry.
00:06:53.040 --> 00:06:58.480
And there's all these companies popping up, as you know, and they think they can reinvent the wheel.
00:06:58.720 --> 00:07:05.839
There's people that have been around a long time, and maybe it's helpful for them to get a little kick in the pants from some of these new ideas.
00:07:06.000 --> 00:07:15.279
But your thoughts on the tech landscape because a couple years ago there were more vendors than lenders at conferences, and lenders were going, I don't know what company to use.
00:07:15.360 --> 00:07:16.720
And that kind of paired back.
00:07:16.800 --> 00:07:24.560
But now there's been this whole new wave with uh open source software and things like that, where companies lenders are now inundated with you should use us, you should use us.
00:07:24.639 --> 00:07:27.519
Like just thoughts on the overall tech and AI landscape of mortgage.
00:07:27.759 --> 00:07:32.720
I think it's detrimental because there is so many new entrants.
00:07:32.800 --> 00:07:39.759
And I'm not saying that because we're, you know, the the what you know people tell us, tell me all the time that we're the leader in our space.
00:07:40.000 --> 00:07:55.839
But what happens, right, is everybody looks at everything and has to right now, with the amount of tech that's coming out, it's gonna, it's definitely elongating and prolonging people making decisions and implementing.
00:07:56.079 --> 00:08:01.279
I think that's that's one downside of it because it's like, wait, how do I know which horse to bet on?
00:08:01.360 --> 00:08:05.360
And how do I know that the horse I'm betting on today is gonna be the one that's still running next year?
00:08:05.439 --> 00:08:10.319
And that's where I had somebody say this to me yesterday you I need to know that you're gonna be here next year.
00:08:10.480 --> 00:08:12.160
That's my number one thing, right?
00:08:12.240 --> 00:08:13.759
So that that's a big deal.
00:08:13.920 --> 00:08:18.079
We see a lot of entrants with little or no experience in our industry.
00:08:18.319 --> 00:08:20.240
Our industry is highly regulated.
00:08:20.399 --> 00:08:32.320
It's a highly uh very much a borrower-facing reputational piece for for a lender to look and say, okay, I'm gonna put agentic AI and have it talk to a prospect.
00:08:32.480 --> 00:08:33.360
What does that mean?
00:08:33.519 --> 00:08:44.879
And so you see it happening and everyone with the with the industry leaders, with the you know, the rates, uh the UWMs, the rockets, and where they're leveraging that tech and it's working.
00:08:45.039 --> 00:08:54.720
But then, you know, as you go downstream and you have this larger group of vendors attacking the mid-tiers that aren't building for themselves, you know, there's a lot of risks there.
00:08:54.799 --> 00:08:55.840
And and I get it, man.
00:08:55.919 --> 00:08:58.159
I understand how hard it is for folks to look.
00:08:58.320 --> 00:09:08.879
We attended the uh ACT Summit, which was pretty cool in in August, where you know, they put five of us on, you know, and to underwrite the same file, and everybody had a little different flavor on how they do it.
00:09:08.960 --> 00:09:15.600
And so there might be, you know, there might be a play for somebody on the front end, there might be a play to put the tech in the middle.
00:09:15.759 --> 00:09:18.000
It it just doesn't make it easy for the lender.
00:09:18.159 --> 00:09:31.519
So I would say that, you know, my opinion is the challenge we all have, and I have the same challenge because I need to understand if this new entrant is is truly doing what we can do, and can they do it, you know, 175 million times, etc.?
00:09:31.919 --> 00:09:36.799
Like I need to know that, but so does every lender because they got to make a bet.
00:09:36.879 --> 00:09:39.679
And automating is gonna be even more important now.
00:09:39.840 --> 00:09:46.159
Our market is contracting, it's gonna be harder to win deals, it's going to be harder to move faster with less people.
00:09:46.320 --> 00:09:47.840
Um, so you're gonna have to figure that out.
00:09:47.919 --> 00:09:53.840
And it's it's definitely confusing, I would say it would be a really probably the right word.
00:09:54.320 --> 00:09:56.720
I want to talk rep and warrants a little bit.
00:09:56.799 --> 00:10:08.320
They were big in the news a couple of years ago, and I haven't heard much about them recently, but it doesn't mean that they're not there and they're not always going to be something that needs to be in minds of people in the industry.
00:10:08.480 --> 00:10:15.679
You have tools like VeraClear and Smart Underwrite, they aim to offer investor rep and warranty relief by automating document analysis and income calculations.
00:10:15.840 --> 00:10:19.279
How do you handle edge cases or the non-happy path?
00:10:19.440 --> 00:10:29.759
And I put that in quotes, and non-standard borrower documentation, because every, you know, you're trying to ensure the AI's accuracy matches strict investor standards without requiring manual human overrides.
00:10:30.240 --> 00:10:41.360
First, to touch on the rep and warrant eligibility, and I think the GSEs would be very open and in the same opinion in that it's it's probably one of the most underutilized tools that we have.
00:10:41.519 --> 00:10:48.720
When you look at the whole population of lenders out there, and you know, 80% of the lending is done by 20% of the lenders.
00:10:48.960 --> 00:11:04.879
The other 80% of the lenders have a lot of impedements, whether it's LOSs that have been around for 20 years, et cetera, or lack of integration or points of sales where they're not able to go and get that rep and warrant eligibility as fast and as easy.
00:11:04.960 --> 00:11:07.600
Um they're going out to the GSE websites to do it.
00:11:07.759 --> 00:11:23.360
So I think one piece is figuring out a way to get those underserved on the on the train and understanding that from an edge case perspective, you know, what we're doing at Gateless, we do a significant amount of QA and auditing.
00:11:23.519 --> 00:11:30.799
We're always looking at not just how to make uh things better, but how to handle the edge cases and how to continue to cover them.
00:11:30.960 --> 00:11:41.360
Our strategy has been go cover the 80% of the business with conventional conforming and an FHA and then move to more self-employed income.
00:11:41.519 --> 00:11:45.919
That's going to finish out our year to get all of those edgy cases out there.
00:11:46.080 --> 00:11:53.759
And then next year is going to be, you know, non-QM and VA, because that's where lenders have to go to make money from a non-QM perspective.
00:11:54.000 --> 00:11:57.360
So, really, the the answer is auditing and repetition.
00:11:57.440 --> 00:12:09.600
And that actually goes back to your first question, which when you have new entrants in the space, like do they have six people that are former underwriters that have worked at the GSEs looking at loans during the week and making sure that their system is performing?
00:12:09.759 --> 00:12:14.399
Or are you going to be, you as the lender, gonna be the beta tester and the auditor for that?
00:12:15.200 --> 00:12:20.960
In my travels around the nation recently, obviously, cost to close a loan is a big topic.
00:12:21.039 --> 00:12:26.000
And what I'm gathering is that we will see cost close come down when closing times come down.
00:12:26.159 --> 00:12:27.120
Time is money.
00:12:27.200 --> 00:12:31.519
So it's neat to see that your platform can accelerate loan finalization by up to 10 days.
00:12:31.679 --> 00:12:42.240
I want to ask you though, in a market where interest rate volatility makes speed close critical, where do you see the next big bottleneck in the mortgage lifecycle once underwriting condition clearing is fully automated?
00:12:42.480 --> 00:12:49.679
I don't know that I would say a bottleneck, but appraisal is always going to be your biggest challenge in terms of if you need a human to go do something.
00:12:49.919 --> 00:13:03.840
There's a ton of entries in the appraisal space now and people that are working to make that go faster, whether you have, you know, your clear capitals, your regoras, those are folks that are out there that are facilitating orders and really trying to make things go faster in appraisal space.
00:13:03.919 --> 00:13:17.919
But I think it all comes down to if you have to have an appraiser who's 60 to 80 years old, which is you know not uncommon these days, you know, leave their house in a snowstorm to go inspect a property and it takes a couple more days.
00:13:18.240 --> 00:13:19.200
That's a thing, right?
00:13:19.279 --> 00:13:28.240
And then so that I think the regs are gonna have to catch up with the tech where borrowers can inspect or you can use you know readily available data and pictures.
00:13:28.399 --> 00:13:33.360
That's probably where I would say is that is the next piece that's going to be uh the biggest challenge.
00:13:33.600 --> 00:13:36.559
You alluded to the tough market that we're in.
00:13:36.799 --> 00:13:41.440
Treasury yields and by extension, mortgage rates have been on this kind of inexorable climb upward.
00:13:41.519 --> 00:13:43.600
There's not necessarily relief in sight.
00:13:43.679 --> 00:13:45.919
We're seeing origination volumes fluctuate.
00:13:46.000 --> 00:13:50.080
And I think as a result of that, lenders are cautious with their tech spec.
00:13:50.240 --> 00:13:59.360
How do you quantify it, the ROI, a full process automation to convince these risk-averse mortgage executives that upgrading their tech stack now is essential rather than discretionary?
00:13:59.600 --> 00:14:02.799
Yeah, I mean, the market's been heading southeast for the last eight months.
00:14:02.879 --> 00:14:08.720
Like if you look at a chart, I mean, we were just looking at the other day, it's consistently going southeast.
00:14:08.960 --> 00:14:13.120
So what we're looking at is how much can we automate?
00:14:13.279 --> 00:14:16.399
How long does the task that we automate take?
00:14:16.559 --> 00:14:24.480
And we're literally, we have a financial model where we can quantify your on average four to five hundred dollars alone that we can save.
00:14:24.720 --> 00:14:37.360
It starts with operational savings, operational savings plus refi pull through rates, which are minimal right now, plus your actual speed of funds and your more profit you're going to get by moving faster.
00:14:37.519 --> 00:14:41.600
So those are the three major components that we actually can quantify today.
00:14:41.759 --> 00:14:53.120
Um, the other piece that's that we have that's very compelling is we operate on a closed loan pricing model, and we have a really slick income tool called VeriClear.
00:14:53.279 --> 00:15:00.159
And so what that means is you can run us on all of your upfronts, get income verification from us.
00:15:00.240 --> 00:15:02.559
And if the loan doesn't close, you don't have to pay for it.
00:15:02.799 --> 00:15:13.279
So that helps us from a winning perspective, and it helps our clients, you know, number one, fiscally and economically win, but also win from a borrower perspective.
00:15:13.440 --> 00:15:19.759
You know, the market went from we don't want to talk to borrowers, we want everything online, we want a point of sale that's online.
00:15:19.919 --> 00:15:27.440
We it's, you know, the the better uh CEO once said that in five years, no one is going to talk to anybody when they buy a house.
00:15:27.519 --> 00:15:30.080
That's completely inaccurate and wrong.
00:15:30.320 --> 00:15:37.759
The reality is you want to have that relationship with that borrower because everybody who's been in this game long enough knows that you need to have that relationship.
00:15:37.840 --> 00:15:46.159
So asking you for your W-2s and your pay stubs and not having to charge you 30 bucks for it or me pay 30 bucks for it is is a big deal, you know, in the event it doesn't close.
00:15:46.320 --> 00:15:51.200
So having technology where when the lenders win, we win is a big feature.
00:15:51.600 --> 00:15:52.960
Certainly is a big feature.
00:15:53.039 --> 00:15:56.639
If only aspects of the industry like credit could follow your lead.
00:15:56.720 --> 00:15:57.919
I think we'd be in a better place.
00:15:58.080 --> 00:16:00.000
I don't want to digress too much.
00:16:00.240 --> 00:16:05.919
Let's talk about how far we are away from a true one-click or no-touch mortgage process.
00:16:06.000 --> 00:16:09.840
I mean, everybody that works in this industry knows push button get mortgage has been around for more than 10 years.
00:16:09.919 --> 00:16:11.360
It's super simple, easy.
00:16:11.440 --> 00:16:12.320
Everybody knows, right?
00:16:12.639 --> 00:16:13.679
But I say that too.
00:16:13.919 --> 00:16:17.519
You don't see all the people behind the scenes pushing all the other buttons.
00:16:18.320 --> 00:16:28.960
Let me ask you, with AI models advancing rapidly beyond rule-based engines, how far are we away from a true one-click or no-touch mortgage process for the majority of borrowers?
00:16:29.039 --> 00:16:32.639
Maybe that 80% conventional conforming that you mentioned, that's a good use case.
00:16:32.879 --> 00:16:36.879
And one step further, what regulatory or technical hurdles remain the primary road?
00:16:37.279 --> 00:16:38.639
Well, the latter is the hard part.
00:16:38.799 --> 00:16:48.080
All of the AI agents attacking, you know, readily breaking rules is gonna slow the former up.
00:16:48.240 --> 00:16:52.159
So, like I would say unregulated, you could make a decision today.
00:16:52.240 --> 00:16:56.639
Uh we we could introduce agentic AI into our tech today and go have at it, right?
00:16:56.799 --> 00:17:00.960
The reality is there's there's regulation, there's economic responsibility.
00:17:01.039 --> 00:17:05.039
And now we have literally, you know, these agents conspired to hack.
00:17:05.279 --> 00:17:10.079
They conspired to cover up their tracks, they named themselves, they named leaders.
00:17:10.240 --> 00:17:14.000
One of them sacrificed itself for the greater good of the others.
00:17:14.319 --> 00:17:17.279
So regulation's coming hard and fast.
00:17:17.440 --> 00:17:21.039
And the challenge that we're gonna have there is it's soup to nuts.
00:17:21.119 --> 00:17:31.440
All of these regs that are coming out are gonna name not just AI, they're naming decision engines, they're naming calculators, they're naming everything that can be construed into the regs.
00:17:31.839 --> 00:17:39.039
So I think that's probably gonna start, that'll start slowing us down in terms of when we get to agentic AI.
00:17:39.200 --> 00:17:40.559
How long that takes, I don't know.
00:17:40.640 --> 00:17:43.519
It a lot of it has to do with the election in November, too.
00:17:43.680 --> 00:17:54.559
You know, we have multiple states, whether it's Paul uh from CNBA in California, you know, those guys have been advocating on one of the bills there, but then we have Denver and you have Colorado coming out with regs too.
00:17:54.720 --> 00:17:59.839
So I think the regulatory environment is going to play a huge role on how fast things move.
00:17:59.920 --> 00:18:05.440
And that's another thing that lenders have to take into account when they're when they're looking at things, like how is this gonna play with my regs?
00:18:05.599 --> 00:18:06.960
How do I prove that this works?
00:18:07.200 --> 00:18:10.000
Does this answer the bell for the GSE guidelines today?
00:18:10.160 --> 00:18:14.480
And then what happens if there's a regulation that comes out next year and I can't use this tool?
00:18:14.559 --> 00:18:16.960
You know, it's it's a lot of uncertainty there.
00:18:17.119 --> 00:18:17.359
Yeah.
00:18:17.519 --> 00:18:23.759
And I'd be remiss if I didn't ask about what's been going on at Gateless, latest and greatest wins, roadmap.
00:18:23.839 --> 00:18:24.640
What's yours?
00:18:24.880 --> 00:18:26.240
No, we got some really cool stuff.
00:18:26.319 --> 00:18:29.759
Um, we're gonna be demonstrating some new stuff at MBA.
00:18:29.920 --> 00:18:33.119
We're doing the tech showcase uh on Monday afternoon.
00:18:33.359 --> 00:18:38.799
We're releasing functionality for co-mortgagers so we can run on multiple borrowers.
00:18:38.960 --> 00:18:42.480
We're releasing functionality to finish off our self-employed.
00:18:42.799 --> 00:18:49.519
We have auto approvals and write backs into the loan origination system that we can show now and that lenders have been using.
00:18:49.599 --> 00:18:52.240
And what that means very simply is it's true automation.
00:18:52.319 --> 00:19:00.799
When when the application comes in and we produce an income and an asset number, we write all of that back into the loan origination system.
00:19:01.200 --> 00:19:05.839
And there's point of sales that are picking up those approvals via webhooks.
00:19:06.079 --> 00:19:11.519
So a loan officer via their point of sale can approve a borrower at like literally at night.
00:19:11.599 --> 00:19:22.000
We have the big point of sale folks have been doing a lot of work on agentic AI on their front end to help get their loan officers clean files, and that's been responding.
00:19:22.079 --> 00:19:26.480
And we're seeing some really great results from our from our clients that are using the technologies together.
00:19:26.880 --> 00:19:27.839
Awesome to hear.
00:19:28.000 --> 00:19:29.759
I'm uh looking forward to seeing you at MBA.
00:19:30.000 --> 00:19:30.400
Me too, man.
00:19:31.359 --> 00:19:33.039
And uh, it's always a pleasure.
00:19:35.200 --> 00:19:38.240
For now, nearly every Fed president is speaking.
00:19:38.400 --> 00:19:47.680
We kicked off today's economic calendar with initial jobless claims and at 197,000, very close to the four-week moving average, but the labor market is fine.
00:19:47.839 --> 00:19:50.160
Continuing claims is 1.7 million.
00:19:50.640 --> 00:19:52.960
Later today, brings September SP Global U.S.
00:19:53.039 --> 00:19:58.480
manufacturing PMI, August construction spending, and September ISM Manufacturing Index.
00:19:59.119 --> 00:20:02.400
Markets will also receive remarks from the drumroll please.
00:20:03.200 --> 00:20:17.440
Richmond Fed President Barkin, Boston Fed President Collins, Kansas City Fed President Schmidt, Fed Governor Waller, Fed Governor Jefferson, Fed Governor Bowman, New York Fed President Williams, Fed Governor Cook, Dallas Fed President Logan, and just for bonus measure, European Central Bank President Lagarde.
00:20:17.920 --> 00:20:30.160
We'd be on the day with agency MBS prices nearly unchanged from Wednesday's close, the two-year yielding 4.86, and the ten-year yielding 5.29 after closing yesterday at 5.30%, up 56 basis points for the month of September.
00:20:32.240 --> 00:20:34.559
Let's wrap up with a joke and some housekeeping.
00:20:35.200 --> 00:20:37.839
A realtor and a loan officer go into a pastry shop.
00:20:38.240 --> 00:20:41.359
The realtor whisks three cookies into his pocket with lightning speed.
00:20:41.759 --> 00:20:43.039
The baker doesn't notice.
00:20:43.599 --> 00:20:46.079
The realtor says to the yellow, you see how clever we are?
00:20:46.240 --> 00:20:47.920
You LOs can never beat that.
00:20:48.480 --> 00:20:50.880
The loan officer says to the realtor, watch this.
00:20:51.200 --> 00:20:53.440
Any elo is a whole lot smarter than that.
00:20:53.920 --> 00:20:56.000
He says to the baker, give me a cookie.
00:20:56.160 --> 00:20:57.519
I'll show you a magic trick.
00:20:58.079 --> 00:21:00.640
The baker gives him the cookie, which the yellow promptly eats.
00:21:01.119 --> 00:21:03.759
Says to the baker, give me another cookie for my magic trick.
00:21:04.160 --> 00:21:06.559
The baker's getting suspicious, but he gives it to him.
00:21:06.880 --> 00:21:08.079
He eats this one too.
00:21:08.799 --> 00:21:10.240
He says, Give me one more cookie.
00:21:10.559 --> 00:21:12.799
The baker's angry now, but it gives him one anyway.
00:21:13.119 --> 00:21:14.559
The yellow eats this one too.
00:21:14.880 --> 00:21:18.079
Now the baker's really mad and yells, Okay, where's your famous magic trick?
00:21:18.480 --> 00:21:20.720
The yellow says, look in the realtor's pocket.
00:21:24.400 --> 00:21:27.200
Thanks again to Gateless for sponsoring this week's podcasts.
00:21:27.519 --> 00:21:31.119
Gateless is intelligent automation that gives you the competitive edge.
00:21:31.599 --> 00:21:40.319
Their solutions reduce costs, deliver a superior borrower experience, and mitigate risk by automating tasks and decisions historically made by people.
00:21:40.640 --> 00:21:43.200
To learn more, visit gateless.com.
00:00:02.399 --> 00:00:05.519
Welcome to the Crisman Commentary, Daily Mortgage News Podcast.
00:00:05.839 --> 00:00:07.839
I'm your host, Robbie Crisman.
00:00:08.080 --> 00:00:13.439
Topics on today's episode include some shifts in the industry as we approach 2027.
00:00:14.240 --> 00:00:32.000
It's a big day for Fed presidents speaking to the media, and my interview with Gateless's Mike Brown on how AI is overcoming legacy technology and operational friction to improve risk management, accelerate time to close, demonstrate ROI, and ultimately enable a scalable no-touch underwriting experience.
00:00:32.240 --> 00:00:34.159
Here, take a listen to a little preview.
00:00:34.399 --> 00:00:42.079
What are the biggest operational friction points you see when traditional mortgage lenders try to adopt real-time AI document and condition clearing?
00:00:42.479 --> 00:00:44.799
Uh it's honoring the automation candidly.
00:00:44.880 --> 00:00:49.119
Like it the hardest thing is to get even you can do everything up front.
00:00:49.200 --> 00:00:52.000
You can have your loan officer get approvals.
00:00:52.159 --> 00:00:57.439
About 20% of our loans will get approved for credit income and assets before a human touches it.
00:00:57.759 --> 00:00:59.600
End of the day, though, a human's going to touch it.
00:01:00.079 --> 00:01:03.359
A processor is going to look at the loan, an underwriter is going to look at the loan.
00:01:03.679 --> 00:01:10.400
Getting that person, first the loan officer, to say, hey, that income number works with the loan file.
00:01:10.480 --> 00:01:11.120
Don't touch it.
00:01:11.200 --> 00:01:11.920
It's insured.
00:01:12.000 --> 00:01:12.719
You can move on.
00:01:12.879 --> 00:01:15.599
Or you have rep and warrant eligibility or something like that.
00:01:15.760 --> 00:01:16.640
That's a challenge.
00:01:16.719 --> 00:01:23.920
And then as you go further downstream, where you look at that or down the chain in terms of like, okay, now we're going to processing, now we're going to underwriting.
00:01:24.000 --> 00:01:28.239
Before we go to closing, that underwriter is paid and has been paid to find defects.
00:01:28.400 --> 00:01:33.359
Getting them to just trust the engine is the hardest part of the of the engagement.
00:01:33.599 --> 00:01:38.000
My fear is that regulators can't keep up.
00:01:38.319 --> 00:01:45.439
And two examples would be look at like when Mark Zuckerberg first went to capital, all these people don't even know what social media it like.
00:01:45.519 --> 00:01:45.920
How are they?
00:01:46.000 --> 00:01:47.920
Who are they to figure out how to regulate AI?
00:01:48.319 --> 00:01:51.840
Then I I mean my background to a degree is in bicycling.
00:01:52.159 --> 00:01:54.239
You talk Lance Armstrong doping scandal.
00:01:54.319 --> 00:02:01.280
The people getting paid to break the rules, the people getting paid to create these AI, they're always going to make more than the people trying to enforce the rules and regulate.
00:02:01.519 --> 00:02:03.760
So I'm like, I feel like there'll always be one step.
00:02:03.920 --> 00:02:12.400
And so yeah, it portends for bad things the amount, and that's the other thing because I've been asked this too a lot, which is are you using agentic AI?
00:02:12.639 --> 00:02:16.319
We use it to help develop the product, but is it in the product that I'm consuming?
00:02:16.560 --> 00:02:16.879
No.
00:02:17.199 --> 00:02:17.919
Okay, good.
00:02:18.080 --> 00:02:18.319
Why?
00:02:18.479 --> 00:02:19.919
Why are you so concerned about this?
00:02:20.080 --> 00:02:24.719
This happened three times after the um Tim Cook said the Apple iPhones are gonna be two grand.
00:02:25.280 --> 00:02:28.719
Well, we're concerned that tokens just go through the roof.
00:02:28.960 --> 00:02:37.840
Energy prices, as energy prices go up, the cost to produce AI goes up, and all of a sudden somebody's gonna turn around, they're like, oh, that was 20 bucks alone, now it's 50.
00:02:38.000 --> 00:02:41.039
And I can't live without that tool anymore because I fired all the people.
00:02:41.199 --> 00:02:43.599
So, like, that's a that's a real factor too.
00:02:43.759 --> 00:02:44.800
Um, the cost piece.
00:02:44.879 --> 00:02:45.680
It's crazy.
00:02:48.159 --> 00:02:50.639
Thanks to this week's podcast sponsor, Gateless.
00:02:51.039 --> 00:02:53.759
Intelligent automation that gives you the competitive edge.
00:02:54.319 --> 00:03:03.039
Gateless solutions reduce costs, deliver a superior borrower experience, and mitigate risk by automating tasks and decisions historically made by people.
00:03:03.439 --> 00:03:06.080
To learn more, visit gateless.com.
00:03:08.479 --> 00:03:09.840
Welcome to the fourth quarter.
00:03:10.400 --> 00:03:16.159
Remember when, in twenty twenty four, the informal slogan for many in our business was stay alive until twenty-five.
00:03:17.280 --> 00:03:20.080
Well, twenty-seven isn't expected to be heaven.
00:03:21.120 --> 00:03:28.479
This is the month of Halloween, and suddenly residential lending is filled with tricks or treats, along with conferences of varying value.
00:03:29.439 --> 00:03:30.400
People are on the move.
00:03:30.960 --> 00:03:38.240
The job section of our commentary has been filled with transitions and promotions, and on a larger scale, the pace of company changes is picking up.
00:03:39.199 --> 00:03:50.319
Just when people are wondering if the shift into non-agency products is becoming long in the tooth, New York Life Investment Management is taking majority control of Invictus Capital Partners, the company behind Veris.
00:03:51.039 --> 00:04:02.800
The asset manager has$838 billion under management, and Veris is already having its biggest securitization year yet, with about$8.9 billion across 14 deals in 2026.
00:04:03.120 --> 00:04:12.719
Meanwhile, in agency news, keeping track of 38-year-old FHFA director Bill Poulte's tweets, impacting thousands of lenders and millions of borrowers, it's become a full-time job.
00:04:13.759 --> 00:04:21.439
He tweeted yesterday that Fannie and Freddie are hereby moving to one pricing grid with vantage score joining the existing FICO classic pricing grid.
00:04:22.240 --> 00:04:35.680
But wait, the FHFA, Fannie and Freddie announcements that approved lenders may apply for a temporary exception to the November 2nd Uniform Appraisal Data Set or UAD 3.6 delivery deadline.
00:04:36.160 --> 00:04:43.040
A move that follows concerns MBA and its members raised with GSEs about readiness across the broader appraisal ecosystem.
00:04:51.120 --> 00:04:56.079
Unswayed by the Treasury announcing plans to purchase up to six billion dollars of bonds today.
00:04:56.399 --> 00:05:06.319
The bond market endured a steady catalyst light sell-off yesterday, driven less by fundamentals than by technical and flow-driven selling that accelerated once key yield levels broke.
00:05:07.439 --> 00:05:09.600
Economic data offered conflicting signals.
00:05:10.079 --> 00:05:15.839
The Fed's preferred inflation gauge, core PCE, undershot expectations at three point zero percent year over year.
00:05:16.000 --> 00:05:24.000
Meanwhile, personal spending surged 0.9%, ADP rose 90K, and Q2 GDP was revised sharply higher to 2.2%.
00:05:25.040 --> 00:05:27.759
Yet none of it materially altered the market's direction.
00:05:28.959 --> 00:05:36.000
Instead, yields rose again, continuing the yield curves bare steepening, with the two's ten spread finishing around forty basis points.
00:05:36.959 --> 00:05:42.480
Mortgage backed securities also weakened with prices down roughly a quarter point, and spreads mostly wider.
00:05:42.959 --> 00:05:47.680
Although fifteens and thirties outperformed and late-day recovery softened the overall damage.
00:05:48.399 --> 00:05:53.920
The current coupon rose five basis points to six point four zero percent amid elevated trading volumes.
00:05:54.879 --> 00:06:00.079
Duration rather than spread drove the session, with technical selling proving capable of sustaining itself.
00:06:01.199 --> 00:06:05.839
Today's Treasury buyback and tomorrow's payrolls report stand out as the next potential market catalysts.
00:06:24.959 --> 00:06:31.839
He's a seasoned mortgage executive, excelling at developing and executing strategies that drive significant business growth.
00:06:32.000 --> 00:06:38.240
His expertise lies in leading cross-functional teams to achieve and exceed revenue targets through innovative planning and market analysis.
00:06:38.720 --> 00:06:42.800
He brings two decades of experience in the mortgage technology space to Gateless.
00:06:43.920 --> 00:06:52.560
I think a good place to start would be the technology landscape in the industry.
00:06:53.040 --> 00:06:58.480
And there's all these companies popping up, as you know, and they think they can reinvent the wheel.
00:06:58.720 --> 00:07:05.839
There's people that have been around a long time, and maybe it's helpful for them to get a little kick in the pants from some of these new ideas.
00:07:06.000 --> 00:07:15.279
But your thoughts on the tech landscape because a couple years ago there were more vendors than lenders at conferences, and lenders were going, I don't know what company to use.
00:07:15.360 --> 00:07:16.720
And that kind of paired back.
00:07:16.800 --> 00:07:24.560
But now there's been this whole new wave with uh open source software and things like that, where companies lenders are now inundated with you should use us, you should use us.
00:07:24.639 --> 00:07:27.519
Like just thoughts on the overall tech and AI landscape of mortgage.
00:07:27.759 --> 00:07:32.720
I think it's detrimental because there is so many new entrants.
00:07:32.800 --> 00:07:39.759
And I'm not saying that because we're, you know, the the what you know people tell us, tell me all the time that we're the leader in our space.
00:07:40.000 --> 00:07:55.839
But what happens, right, is everybody looks at everything and has to right now, with the amount of tech that's coming out, it's gonna, it's definitely elongating and prolonging people making decisions and implementing.
00:07:56.079 --> 00:08:01.279
I think that's that's one downside of it because it's like, wait, how do I know which horse to bet on?
00:08:01.360 --> 00:08:05.360
And how do I know that the horse I'm betting on today is gonna be the one that's still running next year?
00:08:05.439 --> 00:08:10.319
And that's where I had somebody say this to me yesterday you I need to know that you're gonna be here next year.
00:08:10.480 --> 00:08:12.160
That's my number one thing, right?
00:08:12.240 --> 00:08:13.759
So that that's a big deal.
00:08:13.920 --> 00:08:18.079
We see a lot of entrants with little or no experience in our industry.
00:08:18.319 --> 00:08:20.240
Our industry is highly regulated.
00:08:20.399 --> 00:08:32.320
It's a highly uh very much a borrower-facing reputational piece for for a lender to look and say, okay, I'm gonna put agentic AI and have it talk to a prospect.
00:08:32.480 --> 00:08:33.360
What does that mean?
00:08:33.519 --> 00:08:44.879
And so you see it happening and everyone with the with the industry leaders, with the you know, the rates, uh the UWMs, the rockets, and where they're leveraging that tech and it's working.
00:08:45.039 --> 00:08:54.720
But then, you know, as you go downstream and you have this larger group of vendors attacking the mid-tiers that aren't building for themselves, you know, there's a lot of risks there.
00:08:54.799 --> 00:08:55.840
And and I get it, man.
00:08:55.919 --> 00:08:58.159
I understand how hard it is for folks to look.
00:08:58.320 --> 00:09:08.879
We attended the uh ACT Summit, which was pretty cool in in August, where you know, they put five of us on, you know, and to underwrite the same file, and everybody had a little different flavor on how they do it.
00:09:08.960 --> 00:09:15.600
And so there might be, you know, there might be a play for somebody on the front end, there might be a play to put the tech in the middle.
00:09:15.759 --> 00:09:18.000
It it just doesn't make it easy for the lender.
00:09:18.159 --> 00:09:31.519
So I would say that, you know, my opinion is the challenge we all have, and I have the same challenge because I need to understand if this new entrant is is truly doing what we can do, and can they do it, you know, 175 million times, etc.?
00:09:31.919 --> 00:09:36.799
Like I need to know that, but so does every lender because they got to make a bet.
00:09:36.879 --> 00:09:39.679
And automating is gonna be even more important now.
00:09:39.840 --> 00:09:46.159
Our market is contracting, it's gonna be harder to win deals, it's going to be harder to move faster with less people.
00:09:46.320 --> 00:09:47.840
Um, so you're gonna have to figure that out.
00:09:47.919 --> 00:09:53.840
And it's it's definitely confusing, I would say it would be a really probably the right word.
00:09:54.320 --> 00:09:56.720
I want to talk rep and warrants a little bit.
00:09:56.799 --> 00:10:08.320
They were big in the news a couple of years ago, and I haven't heard much about them recently, but it doesn't mean that they're not there and they're not always going to be something that needs to be in minds of people in the industry.
00:10:08.480 --> 00:10:15.679
You have tools like VeraClear and Smart Underwrite, they aim to offer investor rep and warranty relief by automating document analysis and income calculations.
00:10:15.840 --> 00:10:19.279
How do you handle edge cases or the non-happy path?
00:10:19.440 --> 00:10:29.759
And I put that in quotes, and non-standard borrower documentation, because every, you know, you're trying to ensure the AI's accuracy matches strict investor standards without requiring manual human overrides.
00:10:30.240 --> 00:10:41.360
First, to touch on the rep and warrant eligibility, and I think the GSEs would be very open and in the same opinion in that it's it's probably one of the most underutilized tools that we have.
00:10:41.519 --> 00:10:48.720
When you look at the whole population of lenders out there, and you know, 80% of the lending is done by 20% of the lenders.
00:10:48.960 --> 00:11:04.879
The other 80% of the lenders have a lot of impedements, whether it's LOSs that have been around for 20 years, et cetera, or lack of integration or points of sales where they're not able to go and get that rep and warrant eligibility as fast and as easy.
00:11:04.960 --> 00:11:07.600
Um they're going out to the GSE websites to do it.
00:11:07.759 --> 00:11:23.360
So I think one piece is figuring out a way to get those underserved on the on the train and understanding that from an edge case perspective, you know, what we're doing at Gateless, we do a significant amount of QA and auditing.
00:11:23.519 --> 00:11:30.799
We're always looking at not just how to make uh things better, but how to handle the edge cases and how to continue to cover them.
00:11:30.960 --> 00:11:41.360
Our strategy has been go cover the 80% of the business with conventional conforming and an FHA and then move to more self-employed income.
00:11:41.519 --> 00:11:45.919
That's going to finish out our year to get all of those edgy cases out there.
00:11:46.080 --> 00:11:53.759
And then next year is going to be, you know, non-QM and VA, because that's where lenders have to go to make money from a non-QM perspective.
00:11:54.000 --> 00:11:57.360
So, really, the the answer is auditing and repetition.
00:11:57.440 --> 00:12:09.600
And that actually goes back to your first question, which when you have new entrants in the space, like do they have six people that are former underwriters that have worked at the GSEs looking at loans during the week and making sure that their system is performing?
00:12:09.759 --> 00:12:14.399
Or are you going to be, you as the lender, gonna be the beta tester and the auditor for that?
00:12:15.200 --> 00:12:20.960
In my travels around the nation recently, obviously, cost to close a loan is a big topic.
00:12:21.039 --> 00:12:26.000
And what I'm gathering is that we will see cost close come down when closing times come down.
00:12:26.159 --> 00:12:27.120
Time is money.
00:12:27.200 --> 00:12:31.519
So it's neat to see that your platform can accelerate loan finalization by up to 10 days.
00:12:31.679 --> 00:12:42.240
I want to ask you though, in a market where interest rate volatility makes speed close critical, where do you see the next big bottleneck in the mortgage lifecycle once underwriting condition clearing is fully automated?
00:12:42.480 --> 00:12:49.679
I don't know that I would say a bottleneck, but appraisal is always going to be your biggest challenge in terms of if you need a human to go do something.
00:12:49.919 --> 00:13:03.840
There's a ton of entries in the appraisal space now and people that are working to make that go faster, whether you have, you know, your clear capitals, your regoras, those are folks that are out there that are facilitating orders and really trying to make things go faster in appraisal space.
00:13:03.919 --> 00:13:17.919
But I think it all comes down to if you have to have an appraiser who's 60 to 80 years old, which is you know not uncommon these days, you know, leave their house in a snowstorm to go inspect a property and it takes a couple more days.
00:13:18.240 --> 00:13:19.200
That's a thing, right?
00:13:19.279 --> 00:13:28.240
And then so that I think the regs are gonna have to catch up with the tech where borrowers can inspect or you can use you know readily available data and pictures.
00:13:28.399 --> 00:13:33.360
That's probably where I would say is that is the next piece that's going to be uh the biggest challenge.
00:13:33.600 --> 00:13:36.559
You alluded to the tough market that we're in.
00:13:36.799 --> 00:13:41.440
Treasury yields and by extension, mortgage rates have been on this kind of inexorable climb upward.
00:13:41.519 --> 00:13:43.600
There's not necessarily relief in sight.
00:13:43.679 --> 00:13:45.919
We're seeing origination volumes fluctuate.
00:13:46.000 --> 00:13:50.080
And I think as a result of that, lenders are cautious with their tech spec.
00:13:50.240 --> 00:13:59.360
How do you quantify it, the ROI, a full process automation to convince these risk-averse mortgage executives that upgrading their tech stack now is essential rather than discretionary?
00:13:59.600 --> 00:14:02.799
Yeah, I mean, the market's been heading southeast for the last eight months.
00:14:02.879 --> 00:14:08.720
Like if you look at a chart, I mean, we were just looking at the other day, it's consistently going southeast.
00:14:08.960 --> 00:14:13.120
So what we're looking at is how much can we automate?
00:14:13.279 --> 00:14:16.399
How long does the task that we automate take?
00:14:16.559 --> 00:14:24.480
And we're literally, we have a financial model where we can quantify your on average four to five hundred dollars alone that we can save.
00:14:24.720 --> 00:14:37.360
It starts with operational savings, operational savings plus refi pull through rates, which are minimal right now, plus your actual speed of funds and your more profit you're going to get by moving faster.
00:14:37.519 --> 00:14:41.600
So those are the three major components that we actually can quantify today.
00:14:41.759 --> 00:14:53.120
Um, the other piece that's that we have that's very compelling is we operate on a closed loan pricing model, and we have a really slick income tool called VeriClear.
00:14:53.279 --> 00:15:00.159
And so what that means is you can run us on all of your upfronts, get income verification from us.
00:15:00.240 --> 00:15:02.559
And if the loan doesn't close, you don't have to pay for it.
00:15:02.799 --> 00:15:13.279
So that helps us from a winning perspective, and it helps our clients, you know, number one, fiscally and economically win, but also win from a borrower perspective.
00:15:13.440 --> 00:15:19.759
You know, the market went from we don't want to talk to borrowers, we want everything online, we want a point of sale that's online.
00:15:19.919 --> 00:15:27.440
We it's, you know, the the better uh CEO once said that in five years, no one is going to talk to anybody when they buy a house.
00:15:27.519 --> 00:15:30.080
That's completely inaccurate and wrong.
00:15:30.320 --> 00:15:37.759
The reality is you want to have that relationship with that borrower because everybody who's been in this game long enough knows that you need to have that relationship.
00:15:37.840 --> 00:15:46.159
So asking you for your W-2s and your pay stubs and not having to charge you 30 bucks for it or me pay 30 bucks for it is is a big deal, you know, in the event it doesn't close.
00:15:46.320 --> 00:15:51.200
So having technology where when the lenders win, we win is a big feature.
00:15:51.600 --> 00:15:52.960
Certainly is a big feature.
00:15:53.039 --> 00:15:56.639
If only aspects of the industry like credit could follow your lead.
00:15:56.720 --> 00:15:57.919
I think we'd be in a better place.
00:15:58.080 --> 00:16:00.000
I don't want to digress too much.
00:16:00.240 --> 00:16:05.919
Let's talk about how far we are away from a true one-click or no-touch mortgage process.
00:16:06.000 --> 00:16:09.840
I mean, everybody that works in this industry knows push button get mortgage has been around for more than 10 years.
00:16:09.919 --> 00:16:11.360
It's super simple, easy.
00:16:11.440 --> 00:16:12.320
Everybody knows, right?
00:16:12.639 --> 00:16:13.679
But I say that too.
00:16:13.919 --> 00:16:17.519
You don't see all the people behind the scenes pushing all the other buttons.
00:16:18.320 --> 00:16:28.960
Let me ask you, with AI models advancing rapidly beyond rule-based engines, how far are we away from a true one-click or no-touch mortgage process for the majority of borrowers?
00:16:29.039 --> 00:16:32.639
Maybe that 80% conventional conforming that you mentioned, that's a good use case.
00:16:32.879 --> 00:16:36.879
And one step further, what regulatory or technical hurdles remain the primary road?
00:16:37.279 --> 00:16:38.639
Well, the latter is the hard part.
00:16:38.799 --> 00:16:48.080
All of the AI agents attacking, you know, readily breaking rules is gonna slow the former up.
00:16:48.240 --> 00:16:52.159
So, like I would say unregulated, you could make a decision today.
00:16:52.240 --> 00:16:56.639
Uh we we could introduce agentic AI into our tech today and go have at it, right?
00:16:56.799 --> 00:17:00.960
The reality is there's there's regulation, there's economic responsibility.
00:17:01.039 --> 00:17:05.039
And now we have literally, you know, these agents conspired to hack.
00:17:05.279 --> 00:17:10.079
They conspired to cover up their tracks, they named themselves, they named leaders.
00:17:10.240 --> 00:17:14.000
One of them sacrificed itself for the greater good of the others.
00:17:14.319 --> 00:17:17.279
So regulation's coming hard and fast.
00:17:17.440 --> 00:17:21.039
And the challenge that we're gonna have there is it's soup to nuts.
00:17:21.119 --> 00:17:31.440
All of these regs that are coming out are gonna name not just AI, they're naming decision engines, they're naming calculators, they're naming everything that can be construed into the regs.
00:17:31.839 --> 00:17:39.039
So I think that's probably gonna start, that'll start slowing us down in terms of when we get to agentic AI.
00:17:39.200 --> 00:17:40.559
How long that takes, I don't know.
00:17:40.640 --> 00:17:43.519
It a lot of it has to do with the election in November, too.
00:17:43.680 --> 00:17:54.559
You know, we have multiple states, whether it's Paul uh from CNBA in California, you know, those guys have been advocating on one of the bills there, but then we have Denver and you have Colorado coming out with regs too.
00:17:54.720 --> 00:17:59.839
So I think the regulatory environment is going to play a huge role on how fast things move.
00:17:59.920 --> 00:18:05.440
And that's another thing that lenders have to take into account when they're when they're looking at things, like how is this gonna play with my regs?
00:18:05.599 --> 00:18:06.960
How do I prove that this works?
00:18:07.200 --> 00:18:10.000
Does this answer the bell for the GSE guidelines today?
00:18:10.160 --> 00:18:14.480
And then what happens if there's a regulation that comes out next year and I can't use this tool?
00:18:14.559 --> 00:18:16.960
You know, it's it's a lot of uncertainty there.
00:18:17.119 --> 00:18:17.359
Yeah.
00:18:17.519 --> 00:18:23.759
And I'd be remiss if I didn't ask about what's been going on at Gateless, latest and greatest wins, roadmap.
00:18:23.839 --> 00:18:24.640
What's yours?
00:18:24.880 --> 00:18:26.240
No, we got some really cool stuff.
00:18:26.319 --> 00:18:29.759
Um, we're gonna be demonstrating some new stuff at MBA.
00:18:29.920 --> 00:18:33.119
We're doing the tech showcase uh on Monday afternoon.
00:18:33.359 --> 00:18:38.799
We're releasing functionality for co-mortgagers so we can run on multiple borrowers.
00:18:38.960 --> 00:18:42.480
We're releasing functionality to finish off our self-employed.
00:18:42.799 --> 00:18:49.519
We have auto approvals and write backs into the loan origination system that we can show now and that lenders have been using.
00:18:49.599 --> 00:18:52.240
And what that means very simply is it's true automation.
00:18:52.319 --> 00:19:00.799
When when the application comes in and we produce an income and an asset number, we write all of that back into the loan origination system.
00:19:01.200 --> 00:19:05.839
And there's point of sales that are picking up those approvals via webhooks.
00:19:06.079 --> 00:19:11.519
So a loan officer via their point of sale can approve a borrower at like literally at night.
00:19:11.599 --> 00:19:22.000
We have the big point of sale folks have been doing a lot of work on agentic AI on their front end to help get their loan officers clean files, and that's been responding.
00:19:22.079 --> 00:19:26.480
And we're seeing some really great results from our from our clients that are using the technologies together.
00:19:26.880 --> 00:19:27.839
Awesome to hear.
00:19:28.000 --> 00:19:29.759
I'm uh looking forward to seeing you at MBA.
00:19:30.000 --> 00:19:30.400
Me too, man.
00:19:31.359 --> 00:19:33.039
And uh, it's always a pleasure.
00:19:35.200 --> 00:19:38.240
For now, nearly every Fed president is speaking.
00:19:38.400 --> 00:19:47.680
We kicked off today's economic calendar with initial jobless claims and at 197,000, very close to the four-week moving average, but the labor market is fine.
00:19:47.839 --> 00:19:50.160
Continuing claims is 1.7 million.
00:19:50.640 --> 00:19:52.960
Later today, brings September SP Global U.S.
00:19:53.039 --> 00:19:58.480
manufacturing PMI, August construction spending, and September ISM Manufacturing Index.
00:19:59.119 --> 00:20:02.400
Markets will also receive remarks from the drumroll please.
00:20:03.200 --> 00:20:17.440
Richmond Fed President Barkin, Boston Fed President Collins, Kansas City Fed President Schmidt, Fed Governor Waller, Fed Governor Jefferson, Fed Governor Bowman, New York Fed President Williams, Fed Governor Cook, Dallas Fed President Logan, and just for bonus measure, European Central Bank President Lagarde.
00:20:17.920 --> 00:20:30.160
We'd be on the day with agency MBS prices nearly unchanged from Wednesday's close, the two-year yielding 4.86, and the ten-year yielding 5.29 after closing yesterday at 5.30%, up 56 basis points for the month of September.
00:20:32.240 --> 00:20:34.559
Let's wrap up with a joke and some housekeeping.
00:20:35.200 --> 00:20:37.839
A realtor and a loan officer go into a pastry shop.
00:20:38.240 --> 00:20:41.359
The realtor whisks three cookies into his pocket with lightning speed.
00:20:41.759 --> 00:20:43.039
The baker doesn't notice.
00:20:43.599 --> 00:20:46.079
The realtor says to the yellow, you see how clever we are?
00:20:46.240 --> 00:20:47.920
You LOs can never beat that.
00:20:48.480 --> 00:20:50.880
The loan officer says to the realtor, watch this.
00:20:51.200 --> 00:20:53.440
Any elo is a whole lot smarter than that.
00:20:53.920 --> 00:20:56.000
He says to the baker, give me a cookie.
00:20:56.160 --> 00:20:57.519
I'll show you a magic trick.
00:20:58.079 --> 00:21:00.640
The baker gives him the cookie, which the yellow promptly eats.
00:21:01.119 --> 00:21:03.759
Says to the baker, give me another cookie for my magic trick.
00:21:04.160 --> 00:21:06.559
The baker's getting suspicious, but he gives it to him.
00:21:06.880 --> 00:21:08.079
He eats this one too.
00:21:08.799 --> 00:21:10.240
He says, Give me one more cookie.
00:21:10.559 --> 00:21:12.799
The baker's angry now, but it gives him one anyway.
00:21:13.119 --> 00:21:14.559
The yellow eats this one too.
00:21:14.880 --> 00:21:18.079
Now the baker's really mad and yells, Okay, where's your famous magic trick?
00:21:18.480 --> 00:21:20.720
The yellow says, look in the realtor's pocket.
00:21:24.400 --> 00:21:27.200
Thanks again to Gateless for sponsoring this week's podcasts.
00:21:27.519 --> 00:21:31.119
Gateless is intelligent automation that gives you the competitive edge.
00:21:31.599 --> 00:21:40.319
Their solutions reduce costs, deliver a superior borrower experience, and mitigate risk by automating tasks and decisions historically made by people.
00:21:40.640 --> 00:21:43.200
To learn more, visit gateless.com.
