In this episode of MISMO Mic’d Up, JazzX AI’s Head of Product for Mortgage, Jagjit Singh joins MISMO President Brian Vieaux to discuss what it will take for mortgage lenders to move beyond AI experimentation and into enterprise-wide adoption. Their conversation explores why the industry’s biggest opportunity isn’tsimply automating individual tasks, but fundamentally rethinking how lending organizations operate around AI.
Drawing on more than 15 years of mortgage lending experience, Jagjit shares why the next generation of lenders will differentiate themselves by institutionalizing expertise, redesigning workflows, and building AI into the fabric of their operations. The discussion also covers governance, change management, and the practical steps lenders can take today to begin operationalizing AI while delivering measurable business value.
Read the transcript of How AI Will Separate Mortgage Winners from Everyone Else with Jagjit Singh:
BRIAN VIEAUX (MISMO)
So, from your vantage point, where do you think the industry actually is today in the AI maturity curve?
JAGJIT SINGH
So, I think most lenders right now are certainly experimenting with AI. They are active with AI. Any lender that we speak to, they’re actively doing something with AI. And right now, what they’re moving from is POC, POV, or pilot, to practical use. And so, when you look at it, many lenders, they’re using AI to improve pieces of the process, but they’re not looking at the full operating model. And I think the next stage in that maturity curve is going to be how do you redesign those workflows around AI, and not just bolt on more features.
So I think right now, lenders are kind of at the low end of that maturity curve, and we’ll start to see over the next 6 to 12 months, lenders start picking up and getting to that higher part where they’re redesigning their workflows around AI. And I think we’ll talk about that a little bit later today.
BRIAN VIEAUX (MISMO)
Joining us today on MISMO Mic’d Up is JAGJIT SINGH. And Jagjit, you’re the Head of Product for the mortgage vertical at JazzX AI. Welcome to MISMO Mic’d Up.
JAGJIT SINGH
Thanks, Brian. I appreciate you having me. Really excited about our conversation today.
BRIAN VIEAUX (MISMO)
Yeah, I’m looking forward to this one. We’ve been planning this for a while. It’s great that we’re finally getting this pulled together. And, you know, it’s one of those things—and I’m sure you, you live and breathe AI every day—in just a 3-week span from when we first started talking about, you know, recording this podcast until today. Things are changing, like, in rapid, rapid, rapid pace. But before we get into, you know, the rapid change and all the things that go along with that, help our audience understand a little bit about JazzX, who you guys are, what you’re—what problems you’re solving for the mortgage industry. And if you don’t mind, give us a little bit of your background too.
JAGJIT SINGH
Yeah, sure. So let me start with my background. I’ve been with JazzX for about a little bit over a year now. Prior to JazzX, I was with a lender. So, I was doing mortgage lending for, you know, a decade and a half or so of my career. And so I’ve been a part of all of those different technology transformations—technology transformations that we’ve seen, air quotes on purpose—over the last, you know, couple of decades. I’ve done roles within operations, I’ve done roles within product, within technology, across origination, servicing, collections, recovery.
I’ve had opportunities across, you know, the full gambit as a lender. So I moved into JazzX because of the vision that JazzX has: how do we disrupt this industry? How do we build this AI-native mortgage application that can sit on top of your existing LOS, on top of your existing POS, on top of your existing product and pricing engine, on top of your existing technology stack? And have those—have purpose-built AI assistants across the entire lifecycle. From that borrower’s first interaction, through underwriting and getting the loan approved, through closing, signing the docs, getting the loan funded, and going through that post-close or investor audit. You know, every assistant that we’re building is purpose-built to automate the day-in-the-life of the persona that they’re emulating across those different functions that I just went through. And we use reasoning, we use learning, we use this, you know, AI to produce these audit-ready outputs because, you know, it is a very regulated industry for these assistants.
So every finding that we put together, everything that we do is fully traceable, it’s supported by evidence, and it’s paired with our platform that we’re building. Sothe lenders can rely on the governance and the AI safety that we can deploy and measure that return on day one. So that’s effectively what we’re doing.
BRIAN VIEAUX (MISMO)
I love it. And what I think is really important, refreshing, frankly, is your background in mortgage, right?
JAGJIT SINGH
Yep.
BRIAN VIEAUX (MISMO)
There’s, you know, we see a lot of times in our industry especially, people from outside the industry kind of coming in thinking they can solve big problems, which in some cases they can. But as you know, as we both know, this is a really complex industry with, as you noted, a lot of regulatory oversight. Not just, like,macro-oversight on the entity, but, like, in the weeds, in the loan itself. You’ve got these milestones that happen that are regulatory in nature. And if you don’tunderstand that, you know, you could deploy technology that could really cause a lot of problems at a lot of scale quickly. So kudos to the leadership of JazzX for recognizing that. Bringing in an expert like yourself, you know, is actually a smart, smart approach.
JAGJIT SINGH
Yeah, we talk about this, right, Brian? I think the—like, there’s a lot of four-letter acronyms that we can, you know, spew off that if one thing goes wrong, you will have the auditors, you will have the regulating bodies come in at you. And I’ve been in rooms with the auditor, I’ve been in rooms with the CFPB, with the FDIC, you know, with these different levels—these auditors going loan by loan. So, you know, going back to my experience, having that and understanding what the impact of having your disclosures go out after 3 days, you know, understanding what that does, I think it’s super important when we’re building this technology and making that impact to the industry.
BRIAN VIEAUX (MISMO)
I love it. So, you know, you kind of opened up with this. And, you know, there’s a lot of lenders that are still in that very early maturity curve. There’s a lot of experimentation happening. Some of it is kind of, we’ll call it non-customer-facing, some of it’s more like task-oriented. But I’m really curious, from your perspective, especially with your lending background, as you think about, you know, what separates lenders from other lenders, but specifically, you know, what do you think is separating lenders who are experimenting and maybe just dipping their toes in the water with AI today versus those that are actually bought into operationalizing AI?
JAGJIT SINGH
Yeah, I think the, you know, if I define first what is experimentation versus what is operationalization of AI. You know, experimentation is really you’re at that proof of concept, that POC, proof of value, POV, or pilot stage. And really what you’re doing there is you are doing bits and pieces of the process using AI. So your entire model is not shifted. You’re just trying to see, at that point in time, will this work, how does this work, and you’re learning from it. Then when you look at that word that I don’t know how to say, operationalizing AI, this is now it’s how do I tie this to my end-to-end and how do I tie this to my actual measurable business outcomes that I need to have so I’m not just wasting money. And so, you know, when you look at lenders who are experimenting, they’re using things like, you know, AI chat or some very basic underwriting.
But operationalizing it, those lenders that are truly operationalizing it, and those right now are far and few in between, the ones that are operationalizing it, thatmeans that they’ve embedded AI into their workflows. They’ve embedded AI into their organization. They’ve understood that AI is not just going to be, hey, I put this one rule in and life moves on, I just have to do testing on that and regression testing and all that stuff. This is not what that is. This is, I need to have roles that are managing, governing, ensuring the safety of my AI throughout my organization. And so I think that’s really the key difference between experimentation and operationalizing AI and what the—from a different lender’s perspective.
BRIAN VIEAUX (MISMO)
Yeah, and one thing that you said that stands out is the, you know, the organizational, I’ll call it either culture or mindset or philosophy. How important at the lender level is it to really understand how to impact change at an org—you know, from an organization perspective? Like, how much change management, you know, goes into somebody that moves from a traditional or manual linear process to one that is powered by a platform like JazzX that literally is running in parallel, you know, across the entire lifecycle?
JAGJIT SINGH
Yeah, so maybe let me take a step back for a second. You know, let’s take AI out of the picture for a hot second. So when any technology change is going in, any mass-scale technology change, look at replacing your LOS. Look at, you know, something as mass-scale as replacing your LOS to something as simple as I need to automate this particular function that’s on this screen for the processor. That still requires a level of change management because you don’t want to move something into a production environment, whether that’s mass-scale at replacing your LOS or a small automation piece, without ensuring that your users are trained, your policies are updated, your procedures are there, you know, everything is buttoned up and ready to go. Now, you add AI into that mix. So there’salready this level of change management that you have to do.
But now you add AI into this mix, and you’re effectively redesigning, even if you’re still your same technology stack, you’re redesigning the workflow. You’reredesigning things that are not the same as they are today. And you’re doing it in a way with this technology that is very different from what we’ve seen before.
Right? Now I can have a conversation with my loan. I can have a—I can talk to my loan, and I can understand what’s the policy in there. I can understand the data that’s in there. It’s a very different technology. It’s a very different technology that’s being used that now requires this additional layer of change, just from the end user’s point of view. But now when you talk about it at the organizational level, right, your roles, your people that you need to have, the types of skill set that you need to have within the organization changes as well.
And I’m not saying that you’re having to add, you know, 20 different people to do one additional process. That’s not at all what I’m saying. I’m saying that you have to reskill, you have to upskill, you have to reimagine the way that the organization will be operating within the same number of people, of course, right? And so that they’re—I think right now people are underestimating the change management that’s going to be there in order to deliver the value that AI is going to bring. Even if you look at all the technology that we’ve implemented in the past, you know, OCR as an example, I remember back in 2018, 2019, you know, evaluating these things. It was cheaper to do it offshore because I still had to do all this additional stuff on top of OCR. And so you didn’t get the value because you didn’t put the right training into it, you didn’t put the right effort into it, you didn’t put the right, you know, organization behind it.
But now with AI, you know, in order to get that value, you want to make sure that you’re driving that adoption, you’re driving that trust, you’re driving all of those things so you can drive that value.
BRIAN VIEAUX (MISMO)
You said something at the beginning there, which I want to double-click on because I think it actually lit up in my head. It’s like I have a picture of the platform, and you described talking to the loan. Tell us what you mean when you say talking to the loan.
JAGJIT SINGH
So in the past, when have you been able to type in a question with your assistant and get a near real-time response with logic behind it? Right? You may have had chatbots and, you know, I can do a search on policies, procedures. But now, you know, at least with Jazz, right, with JazzX, we have an assistant on the bottom right corner that is Jazz. Jazz can answer any question about the loan, about the policy, in near real-time for the user. So I just now have to go in as the loan officer, as the loan processor, as the underwriter, hey, what was the initial closing date for this loan and when did it change and what document was used to make that change? And Jazz can spit that answer right back to me.
Hey, you know, Brian has this special type of income, right? Maybe he’s getting some RSU income or he’s getting some sort of income that I don’t see very often. I see it maybe once or twice a year because that happens. But I can ask Jazz, hey, Jazz, or I can ask my AI, hey, what needs to be true in order for me to accept this income? Instead of me having to go to the underwriter and ask that question, I can just ask the loan. I can ask Jazz, what needs to be true? And so that’s part of the power of AI, what we’re doing.
BRIAN VIEAUX (MISMO)
Yeah. And I think what you just described, what you opened with when you talked about JazzX sits on top of your existing technology stack, right? And so in the example where I’m talking to the loan and asking a very specific question about Brian’s income, you’re actually asking about the data in the loan file from the LOS against a set of guidelines, loan guidelines from the investor that that loan ultimately is going to, perhaps.
JAGJIT SINGH
That’s right. And the overlays that the lender has.
BRIAN VIEAUX (MISMO)
That’s fascinating.
JAGJIT SINGH
What they accept, what the lender accepts, what the lender doesn’t accept.
BRIAN VIEAUX (MISMO)
So powerful. I mean.
JAGJIT SINGH
It is very powerful, yeah.
BRIAN VIEAUX (MISMO)
I see the power of it from the originator’s perspective, the processor’s perspective. But my risk compliance hat, which is always on, I could, you know, downstream, I could see the power in that when that loan’s going through some kind of a review, right? It’s really, really strong. All right, I want to shift a little bit, not a whole lot, but, you know, it dawns on me that based on, you know, based on what we’re describing here, this is really how lending is going to look in the future, right? And so what do you think are the, like, primary capabilities of the next generation lender, right? There’s, let’s face it, there’s some old, stodgy, you know, lenders still around. And then there’s a lot that are not just new companies, but companies that have been around for decades that are embracing new technology and kind of a new way of thinking.
And I’m just curious from your seat, my guess is the folks you’re talking to are pretty progressive. They’re forward-leaning typically. What are some of those common attributes that make them that next generation lender, in your opinion?
JAGJIT SINGH
I think the innovation mindset, right, that innovator’s mindset is what I think differentiates between that lender that is going to get to that next generation versus a lender that is going to be stuck in, well, I’ll use your words, not my words, dodgy, right? Just for the record, using Brian’s words on that, not Jagjit’s. And so when you think about that innovator’s mindset, they start to think about what can I do with AI?
How can AI help me? And one, you know, I think there’s a lot of things that you can have with this next generation of AI. There’s reasoning. There’s learning. There’s institutionalizing intelligence. So, you know, if I hit on this institutionalizing intelligence, this is where if you can start to get your best-trained underwriters and you can institutionalize their knowledge within the, you know, system of intelligence or the JazzX system of intelligence, you can start to make sure that every loan is underwritten as if your best underwriter is there underwriting it. And that knowledge and that intelligence just compounds with every single loan that you do. I’ll give you an example, a non-mortgage example of this, what, you know, institutionalizing intelligence and having that memory there.
My wife was looking for a gift for my son’s sixth birthday. So I have two kids, six and almost three. And we were looking for a present for him and we wanted to get a ride-on toy. And that ride-on toy, I wanted to get something powerful. I wanted to want, you know, something cheap that, you know, you’re just going to break after a couple weeks. So I wanted to get something powerful. So I said, you know, look for, you know, these types of characteristics that are on that and ask ChatGPT, what are those options?
So she goes into ChatGPT, she asks those options. And ChatGPT comes back with, here’s a few options. However, keep in mind, this is too fast, too big for your three-year-old. The question didn’t, she didn’t ask the question of, you know, for my three-year-old, she asked the question for the six-year-old, what can I do? And the AI was smart enough to understand what the patterns in the past were to then give a better risk-based decision, enable her to make a better risk-based decision on that toy that she’s going to buy. So I translate that back to mortgage, what are we doing? We are making risk-based decisions on loans so we can get those, you know, sold to investors or portfolio them or do what we want to do.
And so imagine if, as I’m making that risk-based decision, I can learn from all of the previous loans that I’ve done, even years back, and the ones that are more recent, like the last three months, six months, yesterday. Imagine the power that that can bring. And I think, you know, tying it back to your questions about what differentiates those next generation lenders is understanding that mindset, understanding that one example of what AI can do, what technology can do, and then extrapolating that across your entire business, entire operations, entire secondary, your headquarter staff, etc.
BRIAN VIEAUX (MISMO)
Yeah, I love that. I think that’s one of the most powerful pieces of AI is harnessing that legacy knowledge and, to your point, institutionalizing it, right?
And that has its own challenges, and it’s part of that whole change management philosophy that we talked about earlier. But, you know, if you’re a leader of an organization, you’re walking a fine line with your staff, asking them to, quote-unquote, “train the system” with their knowledge. There has to be a lot of trust that exists between the staff and leadership that effectively I’m not, you know, training my replacement, which is this, you know, this AI tool.
JAGJIT SINGH
So let me hit on two points there. So point number one, I think your lenders will have, going back to that change management, right, lenders will have to rethink about their incentive structures, rethink about how they perform, mismanage their teams. All of that has to get, you know, rethought through. So that’s point number one. Point number two is, you know, you talk about trust between the leadership and the staff. And again, you know, a point that I was hitting on earlier on the change management aspect of, you know, ensuring that you have the right organizational support around it. You do need, we’ve been talking about having this, you know, policy owner that is a new type of role that can be fused from a different role, but a new type of role that is a policy owner role that can review the feedback that the staff is providing and be that single point of contact that’s reviewing the feedback and then re-institutionalizing that knowledge into the AI, taking that and saying, I approve this, I don’t approve this, or I approve this with these particular guardrails. And so you still need to make sure that, yes, that trust between leadership and staff is extremely critical, but you have to enable that trust in between. And so those are, I think, the two points that I just wanted to hit on that comment.
BRIAN VIEAUX (MISMO)
Yeah, I love that. And that’s, you know, I think as leaders, we tend to assume that, you know, our vision is clear. And my, you know, my answer is always, if you think you’re communicating enough, communicate more, because some people aren’t hearing the message. So you just got to keep kind of hitting that nail on the head.
JAGJIT SINGH
That applies for everybody except for me though, Brian, because remember, everything that I say is crystal clear.
BRIAN VIEAUX (MISMO)
That’s right.
JAGJIT SINGH
And everybody understands the vision. I’m joking.
BRIAN VIEAUX (MISMO)
That’s right. So, you know, from your vantage point, given we opened this conversation talking about our industry still being at the early stages of the maturity curve, where do you see AI adoption heading in the next, you know, 90 plus days?
JAGJIT SINGH
The next 90 days is, you know, I think people will continue to do those PLCs, PLVs, and pilots. And I think you’ll start to see more of that happen in the next 90 days. And you’ll start to see that these lenders are coming in and saying, okay, you know, what I thought was AI before is not really AI. So let me understand what that concept is, understand what that value is, and pilot it. I think you’ll start to see that kind of happen in the next 90 days with all the conversations, all the discussions that are being had around AI. But I think, you know, longer term than that, I think you’ll start to see that lenders will, even those, again, your word, stodgy lenders, not those innovators, they will also start to pick up the pace and start to see that, okay, hey, these other guys, these innovative companies are growing. They’re reducing their costs.
They’re giving better pricing to the consumer. They’re making their consumer’s experience much better. And I’m losing business. I’m losing volume to them. And they’ll start to come up that curve as well. And their curve will be a little bit faster because those innovators have already kind of set up that flow for them. So I think in the next, you know, six months, 12 months, 18 months in this industry, I think we’ll start to see that shift.
And the most important piece of that shift, I think you’ll see the technology shift, but I think the most important piece of that shift is how can the lenders ensure that they’re driving the value and getting the ROI from that shift? Because this is not, you know, inexpensive to run.
BRIAN VIEAUX (MISMO)
Right. Yeah, I love it. That’s a good segue into my last question before we wrap up here. So let’s call this the segment for the stodgy mortgage executive. Even the stodgiest of mortgage executives and leaders out there, I think intuitively understand that AI matters today.
But they’re still not exactly sure where to start. Maybe they’re worried about the governance and, you know, all the stuff. What advice would you give them to kind of help them feel a little bit less overwhelmed? Like, where would you tell them to start today if they haven’t already?
JAGJIT SINGH
I think, you know, first, just educate yourself on what is AI, what are the pitfalls of AI. I think it’s very important. You can understand all of the great things that AI can do, but also understand what are the pitfalls of AI, because you need to go in eyes wide open. But then when you start, start somewhere practical, right? You know, when I talked about PLC, PLV, pilot, start there. You don’t have to change your entire organization overnight. Start somewhere where you can leverage AI, maybe not even in the mortgage operations, right?
Use it to go get Claude, go get, you know, OpenAI, ChatGPT, and use it to help build your decks. Understand what AI can do, what AI cannot do, understand what is AI slop, what is not AI slop. Understand what is this human in the loop versus human on the loop versus, you know, a fully autonomous. And so understand what those things are and start small and then start to expand, then go into, you know, maybe I can put in a, you know, I can underwrite one part of my loan or I can do it, start with document intelligence. I can have AI run my document intelligence instead of my old OCR technology. And I can start, basically you start small and expand, but it’s important for two things. One, understand the pitfalls, but second is make sure you don’t lose sight of the bigger picture of what you’retrying to do.
You can get these bolt-on tools, you can do this, you can start there, but your bigger picture is how do I make sure that my entire flow, my entire business, my entire organization is as efficient or more efficient than it was yesterday?
BRIAN VIEAUX (MISMO)
I love it. Yeah, totally makes sense. Start small, pick a spot, pick a spot that’s maybe not customer-facing, it doesn’t have as much inherent risk, and just start testing and learning. I think it’s great advice. Jagjit, I appreciate you spending some time with me today on MISMO Mic’d Up. For the audience, make sure you subscribe to us. You can go to Spotify and Apple Podcasts, but really the best place to find us is on LinkedIn.
We stream here every Friday morning on LinkedIn. And make sure you follow MISMO as well. Follow the LinkedIn, the MISMO LinkedIn page, the MISMO YouTube channel to get updates on the various initiatives that are happening within the MISMO community, including a lot of great work being done in and around AI. And of course, make sure you follow Jagjit, make sure you follow JazzX AI. Once again, thanks for joining me today.
JAGJIT SINGH
Thanks for having me, Brian.
BRIAN VIEAUX (MISMO)
All right, everybody, come back next Friday for our next episode of MISMO Mic’d Up.
