AI Hype vs. Reality: What Mortgage Leaders Need to Know About Agentic AI

Artificial intelligence is generating no shortage of dramatic headlines, but how much of the fear is justified—and where do the real risks actually lie? In this week’s Weekly AI Update, me and Pavan Agarwal cut through the noise surrounding AI and examine what mortgage leaders should truly be paying attention to. From the rapid rise of agentic AI and its ability to deploy thousands of autonomous agents to cybersecurity vulnerabilities, accountability, traceability, and the limitations of today’s large language models, the conversation separates technological reality from sensationalism. They also explore emerging approaches such as JEPA, the growing role of AI in mortgage operations, and why understanding both the capabilities and limitations of these technologies will be critical as lenders decide how—and where—to put AI to work.

[David] Listeners, we’re back with Pavan and it’s time for another AI update. And as you’re listening to the news, it’s pretty much all over the news that AI has gotten out of hand. There’s fear-mongering going on, that this is going to take over, and we need to slow it down. Well, I’ve got some opinions about it, but I don’t have the expertise that Pavan does and then I’m also listening to Bessant get questioned right now as he’s in at a Senate finance committee getting grilled on this very thing. So, Pavan, you’re the expert. You’re the guy that has the knowledge on this. I really liked your answers when you were on national television earlier this morning. Excellent responses. Talk to us. Is this is it is a concern that we need to slow it down legitimate? or is there something else going on here?

[Pavan] Well, let’s talk about for a minute the people who are saying to slow down. On one hand, they speak from both sides of their mouth. On one hand they say slow down, on the other hand says, don’t worry about it, we got under control. Right? So it’s just it’s really great hype and news cycle that they’re creating. Right? They so and then and then you talk look at the this guy who’s the so-called whistleblower who worked for Anthropic for four months and then worked for OpenAI before that. And it’s like he’s out there making all of these statements and supposedly talking to the media and releasing all this inside information about these companies. But he’s already in breach of his employment agreements and so forth. And those companies, instead of exercising their rights, they encouraged him to speak more. So it’s it is so much of it sounds and feels, so much of it feels like a setup, right? It’s like this is all you know, just all part of the whole mechanism to drive up the hype. Because you know, this these companies are trying to go public and the revenue they they’ve spent so much money building these data centers and the revenues aren’t catching up to it. So they they gotta keep the hype going somehow.

[David] Right. Yeah. They keep the well then there’s the reality. I mean, you’re telling they’re trying to convince us that entrepreneurs who are competing against each other are voting mutually to de-escalate this thing and kind of slow it down. I don’t know of a single entrepreneur. I know of entrepreneurs that are political in mind. That will politically mining will play into that just to try to hope the other two will slow down so that they can keep racing ahead. I’m just not buying it. I’m not, I think there’s more to this thing. There’s a lot more. And I I don’t want to get in a conspiracy theory per se, but this is a race. And even if this is the situation, Pavan, with these three US based companies, there’s China out there. And we’re hearing rumors that these guys are not slowing down and that they’re not going to participate in this quote unquote do this in a more throttled manner.

[Pavan] Yep. And you know, there’s two things I want to point out on that when we as we’re talking about international politics. first of all, President Trump very clearly came out and said, you know, there’s the risk is way overhyped and I don’t see I don’t see it being a problem, right? and you know, yeah, you can argue like you know, he’s a real estate guy, what does he know about AI? But remember, he gets the national security reports, he gets the national security briefings. And you you know that our intelligence agencies are all over this and they would understand the risks, right? And and you know that they would know if there is really a problem or not. Okay, so so he has he has information, the complete picture, right? And from a neutral third party, from governmental body, the complete picture of what the risk really is. And and you know, the national security agencies, they don’t, they’re not profit-driven. They’re not trying to go IPO next next month, right? Or next year or whatever it is, right? So they’re they’re gonna they’re gonna tell him exactly what what what these risks are and where we where we stand. Okay. So so that’s number one. So if anyone who’s worried about it, you should have comfort in that, right? Whether you support the President or you don’t support the President, it you gotta support the process, the system, right? You you gotta you gotta believe in the system. And the system works. Okay. the president has said that this really isn’t a risk. But and then if you combine that with what the AI companies have said, like I said, on one side they said it’s a risky, on the other side they said, don’t worry, it’s under control. It’s not it’s not a risk, right? So so which is it, right? So we have half of what the AI companies are saying is agreeing with what the president is saying. Okay, so that means you get you voted two out of one, right? The safety vote wins, right? That like like it’s like there’s not really that much to worry about. Is there something to worry about? Let’s just talk about what the risks are, like how agentic AI really is risky. Okay. So there was a research done about two years ago that you know, drew out lines or show that all LMs were converging to a certain point in in intelligence and they can’t get any more intelligence than that that point. And we’ve already crossed that point. So it doesn’t matter how much more training data you throw out, how much more you make it do, right, the intelligence of an LM will not exceed that certain point, right? And we’re we’re past that. So they they’ve hit a wall on on how smart LLMs can get. Okay. and I and I do believe that’s part of the problem that these AI companies are having. the second thing is and you is you think about the hype that AI companies create. Like cloud code, right? cloud’s code generation AI is is amazing. And and when you use it, you’re like, my God, this is as good as an engineer. And the cloud anthropic marketed as this is a whole new model. We just breakthrough, we, you know We threw away the rule book, we’ve done something amazing here, right? And it is really cool stuff. And then recently, earlier this year, the cloud was hacked and the code and the cloud code was leaked to to the internet. Okay. And when it after it got leaked, people looked at it and said, wait, this is really good, interesting code. This is really interesting technology, but there’s no breakthroughs here. There’s n there’s not there’s no like breakthrough AI here that you know is the thinking you know true conscious thinking machine that they were kind of hyping it up to be, right? So so that’s an example where the hype exceeds the reality, right? Now, did they have some really cool algorithms and stuff in there? Yes. And and what happened as soon as the code leaked, within a week, the open source community took over and made a a a cloud a cloud code equivalent and better using those same ideas, right? so now it’s in the hands of of all of us. So is you know, so we th the that’s an evidence that the AI industry is overhyping their capabilities. These companies that are trying to go public really fast, they they they live on hype cycle, they’re trying to get attention to get get get shareholder engagement. and and and then w we just look at like what actually happens when you have something that generates code really well. Okay. and then you can just turn on thousands, tens of thousands of agents and have it have it attack a a problem. And it’s got it’s got all the data, they they’ve they’ve they’ve mined all the data they can from the world and it’s got all the patterns over the decades of say how passwords have been breached, how how firewalls have been breached, right? So unlike a human hacker who tries who has to manually try different hacking attempts. Okay, different strategies and it’s slow and and tedious and it’s only as good as the as the the a human hacker is only as good as as how many attempts he could try himself. Okay. But now if you put 10,000 or 100,000 agents to hack a hack into a bank or into a power grid and right, and and it’s got every possible combination that’s ever been used on on you know it’s common patterns about passwords, like human beings. like if if I if I was to create a 256-word password, right? yes, the the number of combinations are astronomical. But because we’re human beings, we tend to have patterns, like we tend to create passwords in a certain way. It’s predictable. How we create passwords is very predictable. Okay, it’s so so what happens is that the these agents will go and try all the different patterns that are known of people generating passwords and that’s why they’re able to breach these firewalls and and and and so-called escape because they can the AI can can guess human human behavior is very predictable. So it can guess things like what what what how where our security holes would be, where our passwords would be and and things like that. And and so so I call it, you know, y yesterday on TV I called it intelligent brute force, right? So it’s a brute force because we have this massive GPU farms and so we have we have this huge brute force capabilities. And and then, but brute force alone, it’s not enough to break strong passwords. Okay, but if you do intelligent brute force where you are careful on picking what you do, then you can break through. and you’re old enough, David. You remember this movie called The War Games from 1980s? Right?

[Pavan] Right that was like when all all us all us nerdy kids we geeked out in that movie. And and if you and and the the the the climax not well not the climate, but the the crux of the movie was, and and there’s a lot of truth in this, is where how the kid hacked into the government system, right? And and what that kid did was he researched the developer, right, who who built the system and and studied his history, and from there he guessed the his personal history. Right.

[David] Did did he study his pr I mean his his personal history? So he was looking for the patterns.

[Pavan] He’s looking for the patterns. And then from by looking at his personal history, he was able to guess the password, and that’s how he got in. Right? So it’s all the research that he did to guess the password and he got in. So he didn’t. Right.

[David] And now with everything being on land, this can be brute forced and done.

[Pavan] Right. Now yeah, and and for an AI to search some some search your personal history, search your patterns. and like most most people then make their passwords based on the name of their pet or this or that, you’re right. and and it’ll go it’ll go try all those combinations first. And then so you know f rule number one, everybody that listens to podcasts, harden your passwords. Don’t don’t you use use random password generator. There’s there’s technology that you can use that generates random passwords. Use that, don’t just create passwords based on your personal personal life. and that’s what most people do. So that’s why, you know, AI can easily break through those kinds of things.

[David] Yeah. Yeah. When you’re looking at where this is going, do we have any reason to really I mean, it’s just without getting caught up in the hype. I mean, if it do you ever have any thoughts, man, this is really getting good. Is do you ever have that moment in your s in your head that going like, I’m scared. I mean, this this is starting to get a little scary.

[Pavan] I would I have that moment every day. This is getting really good because I I see the work that we do, right? And and it’s like every single day is like you know the the speed at which we we get data reviewed and analyzed and and the n the new level of of an analysis and understanding that we have is just it’s it’s mind blowing. And there there is this this new model that one of the one of the original fathers of AI is his name is Jan Lacoon. I don’t I don’t know if you’ve heard of him. yeah,

[Pavan] Right, right. So and he just released a new model. J E P A. Now it’s Jeppa too. So so the model the model’s called JEPA. J E P A stands for joint embedding predictive architecture. Okay. And and so remember he he left he he he was one of the the he was like the founding guy inside Meta who who built their AI and one of the key minds behind, you know, modern LLMs. And he quit Meta because he got into a disagreement with Mark Zuckerberg about the direction the research should go and he started started his own company and he immediately instantly raised a few billion dollars. And now he created this this model, it’s out in open source. It’s again it’s called Jeppa. And the way the model works is that it understands, instead of understanding words, which is what large language models do, it understands words and predicts the next word, it understands concepts. Right? It understands that if you drop an apple it’s gonna fall down. And so it understands the concept of gravity. It doesn’t just understand that if I have a sentence that say if I say I’m holding an apple, I’m about to release it, I’m about to open my hand, what will happen next, right? If you prompt that into LLM and it will it will it will write the apple will fall, right? Because it’ll it that’s the most most predictive set of words that follow after it. It doesn’t understand why the apple is falling, it doesn’t understand that the apple is falling. Right. It just it just it’s just predicting those words because that’s the most common pattern in its database that the Apple false. So it’s so you as a human think you think that is thinking, that that it

[Pavan] Yeah, you think that it understands, but it really doesn’t understand. Right. It’s it’s just telling you the most likely pattern of next words. Okay, and and that’s why because of this this concept, this principle. Apple issued a paper exactly a year ago called The Illusion of thinking. So all of you watching look that paper up. And they explained this in detail. It’s a very well researched scientific paper about specifically this issue. Okay, and and so Jan Lacun is he’s like, I want to solve that problem. I want the AI to understand the concept of gravity. So if you if you release the apple, it’s gonna say the apple will fall because it understands gravity and therefore It’s it’s falling. Okay. And so it’s not just predicting those words, it’s it’s understanding the concept and predicting is what it’s doing, is because of concept A that applies to this object, it predicts the concept B, and this concept B is falling. The concept of falling is tied to the concept of gravity, right? Okay, so the this is this is a revolution and this kind of model is like you’re getting closer to actual thinking, human-like thinking.

[David] And and that I mean that doesn’t scare me, Pavan. That but it has I mean the b I see this as an enhancement. I see always have seen AI as an enhancement. Does that ever scare you that this is getting closer to thinking?

[Pavan] It’s just these are just getting better and better tools. Okay. And and getting closer, like thinking means actual conscious awareness, right? And we can have a whole different conversation about conscious awareness and and these machines aren’t aren’t ever gonna get there unless they start working at a quantum level. and so that’s well, I mean qu

[Pavan] Quantum computing is is also like I mean, you know, again another overhyped thing. I mean I overhyped in the sense I believe the research should continue on quantum computing, but it’s nowhere near I mean it it’s i i it’s doing math you know, in and ver y it can it can solve very narrow specialized math problems really well. Okay. But it’s i i is it is it you know, can it i it y you can’t replace a quantum computer. You can’t replace GPUs with a quantum computer. It’s not there yet. Okay, so so that’s the there’s a there’s a there’s a huge divide to cross on on that, right? And the the the and the irony of quantum computers is that the math problem that it solves is prime number generation, like in immediately. Like it does that in a in a in a second. And and and that’s the best math problem to solve when you wanna when you want to crack security keys. So so you know that that’s when they talk about this quantum proof. Security algorithms is is because the the one problem quantum computers solve really well is cracking keys. Right. Okay. So you know but but they’re not general purpose computers and people working towards solving that problem.

[David] Yeah, when you say general purpose, they’re not w explain that difference right there for the average person that’s out there.

[Pavan] Right. you know, general purpose means like you can solve one c you know, this this solves specific types of math problems, right? And you could tune fine-tune a quantum computer, do maybe one does prime numbers, maybe one does, you know, integrals or or whatever, right? but y y you can’t take a c a quantum computer that solves prime numbers and have it do, you know, a n a n a neural net calculations, for example, right? It it’s you you would need a different so like one of the ways they could solve this problem is have different quantum computers networked together and and each one doing a different special specialty and and then you combine the results. But anyways, that’s w way off yeah.

[David] Yeah, well yeah. Well, so again to d as we wrap this up on this topic, I have another thing I want to talk to about. you’re not worried about this. You think this is I’m not saying it’s

[Pavan] I’m not I’m not worried about it because if look if AI It the hype is that AI is is has escaped, AI is is can do you know can crack any key and and we’re at what they call it artificial general intelligence, it can it can outthink us, right? If that was the case, okay, then then why haven’t you used the that same AI to create a security system, generate the code to generate secure code that will prevent you from being hacked? Right? if it was that good, then he could also write then he could also write the the AI to block the AI. You could also write un unhackable security locks, right? So s so you know, and so I I think that there there’s there’s again it goes back to w why why aren’t you know the Sam Altman and Anthropic talking about that, right? they they’re only talking about the AI escaped because it’s sensational and it’s exciting and it gets media clicks. But if they if they put out a if they were to put out a press release saying, hey, we wrote an AI that created a an impossible to break lock for your firewall, you think anybody’d be talking about it? That’s that’s not exciting. No now I I don’t know if they did or if they didn’t. I haven’t heard anything about it. I read all the research. I haven’t heard anyone come out saying that that you know, we created a new network security layer that is impossible or is you know, we haven’t been able to break it with another AI. you know, cause I haven’t I haven’t heard about anybody talking about that. That doesn’t mean it doesn’t exist, right? But that’s the point, right? I mean if the if the if the tool gets s smarter, you can also use it to find solutions against itself, right? Right. But whether the tool will, you know, the the the whole terminator kind of fear that we all have in this culture and that the tool will go off on its own and and consciously make a decision to to destroy us all, it would

[Pavan] It would it would have to be instructed to do it. Just like just like you know, you have to inst instruct a gun to shoot, right? You when you pull the trigger, you’re instructing the gun to shoot. You’re the gun isn’t shooting on its own. You’d have to instruct the tool to do those things, to to make those to go down that path.

[David] Because we’ve watched the five year, excuse me, the ten year Treasury go up over five. Wanna get your perspective on interest rates. We wrapped up this interview here.

[Pavan] Yeah, i it it’s interest rates, home values. I think there was a report that home values actually have s still continue to rise recently, which is but whereas in some markets, I think that’s that was the national average, but we are seeing softening as much as ten, fifteen percent in certain markets as well. So it’s so I I think I think as rates continue to rise, we will be in You know, lenders that are doing high high L T V non QM loans and and second mortgages, th they should be very cautious right now. my cause because this is the kind of a a cracking point where home home prices can correct by, you know, ten, fifteen, maybe even twenty percent. and if you’re sitting on a on a second mortgage at a ninety percent loan to value, you just got wiped out. So be careful. Be careful out there.

[David] Right. What about MSRs? What are you thought your thoughts on the value of MSRs?

[Pavan] Well, it’s like we have a huge MSR portfolio. I I just made a bunch of money. So I’m really well hedged. and our production, our production is our loan production has stayed constant to growing through this through this rise cycle because just the demand for the AI keeps going up. So so, you know, w we’re we’re like so well hedged on this thing so it doesn’t bother us. the you know w what where do I think the interest rates I do believe that this is these this is this is all geopolitical problems. I memories are short, right? Don’t forget back in March before the war started, it was in March or February, you know, the tenure was below four percent. Right? And right?

[Pavan] And mortgage companies and banks started hiring like crazy again, like massive like like here we go, and then

[David] Now so what you’re suggesting is we could be back to those good days. Those good days are

[Pavan] Yeah, we we yeah. Rem remember the Houthis the Houthis just blew up the east-west pipeline, which was the way around the Strait of Hormuz. Right. And so the the oil just suddenly stopped coming out of Saudi Arabia. Okay. And y you know that that eventually this will get solved. One way the other it’s gonna get solved, right? And I I

[Pavan] I do believe in in in in the American military, I do believe in what our soldiers are doing, and I and I I think eventually we’re gonna put China in this in this place and they’re gonna stop supporting Iran because the the oil’s gotta keep moving, right? We’re gonna we’re gonna get this figured out. Okay, and and the oil prices will w ill will will fall. Shipping like a lot of the inflation’s come because the the shipping lines have been disrupted. Okay. Right?

[Pavan] And and so just moving the goods from from both directions, right? Trade has been disrupted and moving goods around has has you know costs have gone up four or five times. Okay. And that that all comes down to consumer consumers feeling those pressures, right? So and and these are logistical problems. The this isn’t a fundamental problem in the you know like i it’s not it’s not due to like say overstimulation or something. These are just these are geopolitical logistics problem and they will be solved eventually. I we don’t know when, but they will be solved. And we hope, we pray that they’re solved soon. And when the when they are solved, we’re gonna see inflation will collapse, right? And interest rates will follow. Now here’s the good news. Here’s the good news that we do have inflation but jobs numbers are still strong, right? Right? We’re still at like historic low unemployment. Right? Now I really feel for for for working, you know, hardworking Americans because their buying power has eroded and the and they’re they’re feeling the squeeze, but at least we’re not in stagflation. They were not in nineteen seventies style stag stagflation where n when nobody’s working and inflation and and you’re you’re and and then you get the double whammy you you you can’t get a job and the money that you have in your pocket is suddenly fallen in half, right? least those days aren’t here, right? and and those were the the the w some of the worst time in American history. Probably, you know, next to the Great Depression’s probably, you know, those those seventies was probably one of the worst times to be in America. So yeah, as so as as you know, inflation is terrible, but i i i but when you have unemployment in the five in the low five, you know, basically at five percent, I I think we’re gonna be all right. Okay, I’m not I’m not terribly worried. And then and especially since we know the cause of the inflation, the root cause is this is these these you know rogue regimes out there shaking up the world. We get those things straightened out, we’ll we’ll be in good shape again. Yeah.

[David] Right. Well, I love the optimism I share that with you, Pavan. Thanks so much for being on here with me today. I know you’re in Las Vegas at the Guerra Casas event. Not real. Tell us a little bit about that real quickly.

[Pavan] Yeah. Yeah. Yeah. Yeah. Yeah. it’s an amazing event and yesterday it was the first day of the event and we stole the show. We opened up and the Angel AI robot opened up the show. So we we’re working on the world’s first robot account executive. And and it was on stage it was it was it was it was it was dancing with I forgot his name, but he’s he’s he’s a big big Latin star. and and we’ve loaded the the Angel AI. Yeah.

[David] Yeah. and it it I I would love to see what that you have to send us some videos of that.

[Pavan] Yeah, just just follow Gary Acosta on Instagram. He’s got he’s got all all over his Instagram. So yeah,

[Pavan] so we we’ve installed the we loaded the angel AI transactional language model into the robot. So the robot understands mortgages. And so we pe people are interacting with yesterday, asking all kinds of questions. It’s like it’s got all the answers and everything. So the world’s first basically robot account executive we’ve created. and and it’s so much fun and interactive. and think about it like like everyone was having a good time and and interfacing with it, right? And it was inviting. It was it’s great user interface. It’s inviting, it gets you involved, gets you engaged. So think about it as a loan officer, having having your robot assistant with you, right? As as you as you work with clients and and open houses and so forth, right? That’s that’s the world that we’re it’s such an exciting future that we’re we’re we’re heading heading into.

[Pavan] If you have a few minutes, I want to talk about one more thing. I don’t know if

[Pavan] You’re talking about the risks of AI, right? And and all of this all of all this talk about AI going rogue and and taking over and derailing humanity. The the MIT issued out MIT Slow and Business issued a paper explained. Agentic AI explained. Okay. And they walk through all of the capabilities of agentic AI and all of the problems and the risks that agentic AI exposes us to. Right? The the promise is massive because you could you could fire up thousands of agents and all like working together and making decisions and and and solving problems. Huh? Right?

[David] That’s kind of what they’re hyping up here. Yeah. That’s what they’re hyping up.

[Pavan] Yeah, that’s and and that’s and that’s what’s, you know, when when you when you if you want to breach somebody’s security, right, you you can

[Pavan] You can you can instead of having one hacker, one human hacker, what if you had a hundred thousand AI hackers trying out different ways to get in, right? And that that’s how you can break through security so that’s the whole point of agentic AI. So it it’s it’s caused a quantum leap. Yes, I talked earlier about how how LLMs converge to a s to a single point of peak intelligence, right? But now with agents, right, so like if you let’s say your peak intelligence is a sixth grader, but when you have 10,000 sixth graders all trying out different things and working towards a a single solution, so collectively, they have a collective IQ of a genius, right? So that’s that’s the the thing with agentic AI, right? That’s why the that’s why there’s this promise. The promise is massive. Okay. And but this this paper, right, and it’s a really easy read. It’s written in it’s not written in techie terms, it’s written in way, you know and any business people can understand, huh?

[Pavan] Like February, March of this year. So it’s very it’s a recent paper.

[Pavan] It’s a very recent paper. And it’s very easy to understand. It’s written in common, you know, everyday, everyday English, everyday language. Right. It’s not techy. so the paper said, yeah, there’s a lot of interesting things it can do, but there’s a lot of problems with it because you don’t know. it’s hard to you can’t explain agentic AI when an agent makes the decision, there’s no traceability. So you so you fail to meet the OCC standard of an AI being to explain how it came to a decision, right? So it lacks it lacks traceability and and then they in this particular paper they cited a specific example of AI agents, okay, agentic AI making the wrong decision on a mortgage application, okay?

[Pavan] And then they’re saying who will be responsible, who’s liable for that, right? ‘Cause ’cause i so if you if you run your mortgage company or your bank using a Agentic AI and it makes the wrong decision, who is liable? So they they they pose that same question. So they’ve they’ve done the research and they’ve looked at this and says, yes, there’s a reliability problem and and and you don’t know whether it’s gonna make the right decision or not. And then if it makes the wrong decision, who’s gonna be liable? And then in and whatever decisions it makes, i it’s it’s hard to ex you can’t explain how it got the decision, right? So these are the These are the limits of the technology, right? So if you don’t understand these limits, right, then you can’t really you know, if if you believe the hype, if you believe the overhype that that this could take over the world, it’s gonna do everything, without understanding the limits, then you’re gonna make some fundamental mistakes in implementing it. And I’m gonna I’m gonna give you an example of of that. A real world this just happened yesterday. We helped, you know, Angel AI helped a borrow yesterday. And and this borrower is is is so funny, like. She she she was getting a HELOC with with a with this other lender, big very very overhyped, or I’ll be I’ll be I’ll be kind. a very a very talked-about lender who’s public and who’s public and and you know claims to have AI completely AI originated HELOCs, okay. So what happened was she accidentally put her mom’s bank statement into the AI instead of hers, right? And the AI couldn’t tell the difference and and and accepted that bank statement. Right? Right? So that’s that’s the ch like like like such a fundamental mistake. The Agentic AI didn’t understand, couldn’t figure out that this bank statement wasn’t hers and it was her mom’s bank statement. Right? And

[Pavan] And it and it moved forward, right? Now now she’s in a mess. Okay. So so so that’s what happens when you don’t understand the limitations of technology and you implement in your business. I mean that’s you know, if they if they had funded that loan, that that’s an unsaleable loan, right? It’s it’s it’s junk. I mean that that’s that’s two thousand eight all over again, right? It’s just you know, it the g the the bar’s got a pulse, bundle loan.

[David] But if they’re use but but if you’re angel using angel AI, that’s a non issue.

[Pavan] That’s not an issue because we have we have the proprietary, what we call the transactional language model, and it’s got all that I mean, like we were creating agents back in 2012. We call them cells. So we’ve been working on this technology a long time and we know how to make it accurate and and traceable. And and you know, obviously our track our track record, we’ve done, we’ve done 200,000 plus loans, and then we’re the only company that issues a warranty on on our decisions.

[David] That’s so powerful. Pavan, good to have you with us, friend. Appreciate you being here today.

Important Links

Pavan Agarwal is a renowned leader in the mortgage lending industry and a pioneer in bringing artificial intelligence to the financial markets. Agarwal serves as the President and CEO of Sun West Mortgage Company and Celligence International, LLC.