00:12From Chicago with HousingWire, I'm Allison LaBorgia and I am sitting down with Brian
00:18Niles, the head of Cloud for Mortgage with SAP Pioneer. Hi. Hi. How's your day been?
00:26Uh, really interesting today. Started at 5 a.m. meeting with my leadership in Germany. It was
00:33a great start to the morning after a little bit of a late night. Now, Brian, you have worked in
00:39the
00:39mortgage technology space throughout your career. What have those experiences taught you about where
00:46servicing technology needs to go next? Interesting. So I did have a little bit outside of mortgage
00:54early in my career, which is making an odd tie in. But where mortgage servicing needs to go next,
01:00plainly, is efficiency, right? The servicers, the subservicers, that ecosystem is struggling to
01:07maintain profitability and efficiency gets them there. And the modern platforms are going to get
01:14them there, right? It's not going to get them there tomorrow. It's a staged path forward. But
01:20that's the future of the industry, right? Modern servicing platforms. What is a modern servicing
01:26platform? It's a single ecosystem to handle your loan from origination all the way through the end
01:33of the loan, whether that's payoff or default. Truly a single user experience with a data model that sits
01:41underneath it that is consistent. Now, it's interesting that you mention a modern servicing
01:46platform because we know that some of the mortgage servicing infrastructure was built around the
01:52needs of another era. That's one way to put it, yes. From a leadership perspective, I'm curious, how do you
02:00decide what legacy processes should be improved, replaced, or just left behind? So interesting because there is
02:09so many different processes that are driven by module-based systems today, right? And you have to go
02:17out and hire very specific expertise to handle loss mitigation or claims because not just the skill set
02:25required, but the technology is driving you to have resources trained to do a single job, right? So we have
02:32to get to a place where resources are allowed and have the capability to move across modules to support the
02:38business needs, right? So I don't know if I answer your question explicitly, but that's where my mind goes
02:45on the topic. It's, I'm here for a good answer, and that was a good answer. And I want to
02:52talk a little
02:53bit about something that I think people are increasingly addressing, which is data standardization.
02:59Yeah, data standardization is a great topic. There's an entire MISMO standardization by the MBA that's
03:06been ongoing for, I don't even know how long, most of my career, right? And you look at the data
03:15situation a lot of servicers are in, and then where they need to go with their data, it's extremely
03:21challenging. Why? Yes, why, Brian? Why? Because data sits in five different spots. It has to be synced back
03:29and forth via nightly batches, right? Their data was built and stacked on top of data. And I use this
03:36example a lot in meetings. When the original data for a mortgage was defined, it was defined as X.
03:41And then servicers go, well, I need this. So they'd go off and create, well, here's a custom field for
03:46you
03:46and custom data set for you. And then they would standardize it. Well, the clients were using the
03:52specific one for them. Because they're special. Because they're special. And they didn't want to
03:56move to the standard. But yet the other clients were on the standard. So you have this non-standardized
04:02data set that isn't reproducible across your client base, right? So then you go, well, how do I move to
04:08a modern platform, especially one that's AI native, for instance? Well, you can't because the AI is so
04:15heavily dependent on structured data. And the MBA does wonderful things with MISMO. I got to give
04:21them a plug here in working towards that. So our platform, we started out with standardized data
04:27structure, right? We went forward with the idea that all data is defined. And that way, we don't
04:35have that problem. And that way, our AI immediately knows what that it is. And if there is an undefined
04:41data
04:41element, we define it, right? And that way, it's defined in perpetuity across all the enterprises,
04:47right? We don't end up in this situation of data being in a lot of places, 15 may be extreme,
04:54but there's data warehouses and servicers, you have your data that you may have to pay just to get from
04:59your current service provider as well. And then you have your data off with other providers of technology
05:04that has to sync back to that system of record, either programmatically or manually through user
05:10entry. Not ideal when you're talking data. Now, I think it's an important clarification.
05:19What does a truly unified data foundation actually make possible for servicers?
05:24It makes possible what's coming, right? It makes possible what's coming. And what is coming is AI.
05:32And AI has been a hot topic for a couple of years now. So the implementation of AI and the
05:39augmentation of
05:40your staff with AI is the next evolution, right? If you look at how technology has been implemented,
05:46I don't care whether it's farming or automotive, right? The first thing that happens is you augment
05:50the humans with the tools, then they become more efficient, then they can do more,
05:56right? And then the load moves towards technology. So that's, that is the amazing future we have,
06:03right? And people are always looking for numbers, right? And I say, depending on how you do the math,
06:08right, you may be at a thousand loans per FTE today. And in five years, you may be at 5
06:14,000,
06:1510,000 loans per FTE. That's where the efficiency goes. And I correlated a lot to what has happened
06:21with farming over the years. As 30% of the GDP used to be farming, right? Now it's 1%, but
06:29yet we're
06:29producing the same exact amount needed to both sell and survive as a species. So you apply that kind of
06:37model to technology. That's what's coming. You're going to see major efficiencies in a gradual
06:43implementation of AI across the platform. Now, we know that AI is coming. And you said that
06:51having that standard data set really paves the way forward for that. I want to talk for a second
06:57about the distinction between adding AI to an existing system and building an environment where AI can
07:03actually operate effectively. Yeah. And there's many layers to that because there's the technology
07:10aspect and there's the human aspect, right? And I think a lot of people are forgetting about the
07:16human aspect. In order for AI to succeed at technology, data, data, data, infrastructure, security,
07:22those things, right, that we're at, others are working towards, right? But the human aspect,
07:29you have to bring your staff AI enabled, right? They have to change how they're going to operate as
07:36well. And that's a big challenge because people are very accustomed to doing their role in a way
07:43that they've done for 5, 10, 20 years. Well, now you're going to be adding new tools, right? So you
07:49have to train your staff on how to use those tools, right? It's just not, oh, AI is there now,
07:54everybody's going to use it. No, your staff, yeah, exactly. Your staff has to come up to speed on
08:00those tools, right? When we went from legacy years ago, when we were just using an online screens and
08:06we went to queues, right? It was a big change. Like now my work appears in a list, right? Prioritized
08:12at
08:12the top. And I have to go figure out what that means. And you have to use a mouse even
08:16in some cases,
08:18right? So these were big leaps. Well, AI is that next big leap. So you have to get your mortgage
08:22servicing staff ready for AI. What does AI ready mean in the context of mortgage servicing?
08:30Well, AI ready means enablement, right? So you're either going to leverage an existing LLM that your
08:37technology service provider provides, like we have an AI that comes with our platform, or you bring your
08:45own large language model. So you have to be ready for those options. And then you have to be ready
08:50for
08:50the cost, right? You have to budget for your AI expenses and mortgage servicing operations,
08:55because you don't want to run out of budget halfway through the month and go, okay, what do we do?
08:59We
08:59go back to manual, we turn AI off for a month, half a month, right? That's, that's not realistic. So
09:05you
09:05have to plan and understand the cost and what's coming. And what we've seen on the cost of AI is
09:12it hockey sticks up, right? Because the adoption becomes so prominent, but it also becomes a
09:18quality standard for a mortgage servicer too. We see who's leveraging AI in our organization.
09:25And the higher the leverage, what we're noticing is the higher the performing employee was before
09:31they had access to AI. So it really gives you another quantitative measurement of your staff
09:38on their ability to advance. And you have to be able to adopt AI, not just to survive in the
09:45job,
09:46but to survive in what's coming in the world, right? It's, it's, it's inevitable. It isn't, it's,
09:51it's transformative. When I, when I sit down every morning, the first thing I do is I hit co-pilot
09:56and
09:56go, what are the three most important things today? And it gives me a readout, right? It gives me a
10:02readout of
10:02today. The three things I must do before I end my day. So just, just little things like that are
10:10transferring. Now, Brian, we have to have a human moment and we have to talk about how you see the
10:17role of the service employee, like changing with this technology where you have individual point
10:24solutions. And now we're moving away from that into autonomous agent, agent workflows and more
10:32strategic implementation of AI. What about the humans? Like so many times, uh, the guy with the
10:40shovel was replaced by the excavator and he must learn to drive the excavator, right? Uh, with AI to,
10:46to maintain what you have, you either have business expertise or technology expertise. Well,
10:52the technology in this case needs to be trained on the business expertise and then it needs to be
10:57monitored, right? It needs to be insured. And then those quality agents, right? The humans in the middle
11:02and in mortgage with the regulatory and compliance structure we have, humans are always going to be
11:07there. We're, that's going to be the check and balance, uh, for the foreseeable future. So again,
11:13it comes down to, you have to be willing to accept AI, the future of your role, uh, your value
11:21to the
11:21organization is going to increase with the more knowledge you have on the topic and your ability
11:26to help train and maintain those models, right? And the skills embedded in those models as the
11:32industry shifts and changes, right? So increasing your value by taking on more interesting responsibilities,
11:39but that's kind of been the role of the mortgage servicing employee throughout is if you're very
11:45specific in your role and you only do SCRA checks. Well, when that volume decreases, unless there's
11:50another role you have a skill in, you may not have a position in that organization going forward. So
11:57continue to expand your skill, use the tools available to you. Um, you know, download Gemini,
12:03right? Go out there, use ChatGPT, use Copilot. These are great tools, very specific, but they're only as
12:11good as the data being fed into it. And whether that's the underlying data of the servicing system
12:15or the data about the situation that that potential, uh, servicing represent is in. So if they get a call
12:23from a customer feeding that information into the AI, so it has an understanding as well, because it's
12:28only as good as the information it has access to. That context layer is key. Exactly. Thank you for
12:34the, the word was escaping me, but context layers. I got you, Brian. Now I'm going to ask you to
12:39look into
12:40your crystal ball here. Yes. What do you think the most forward thinking mortgage
12:45servicers will be able to accomplish in the next few years that would have been impossible with
12:51legacy infrastructure? Uh, efficiency, right? I, the efficiency gains moving to a modern platform
13:00are transferable. They may put your organization in a place to be in business. They may put your
13:07organization in a place to consume other organizations who are not as efficient. If
13:12you're locked in on a legacy platform and you can't move to one of the more modern platforms,
13:18you're going to see your margins disappear. You're going to see your business leave to more efficient
13:24organizations who can respond to the needs of the business, right? We may see 50 year mortgages,
13:31right? That product will need to be deployed quickly. We may see changing products across the
13:37ecosystem. We may see credit cards off your home equity line to tap into that, you know,
13:43$20 trillion in available equity out there. These products require rapid deployment, right? While
13:50maintaining regulatory and compliance, which a modern servicing system can give you very quickly.
13:57Brian, thank you so much for taking me through what a modern mortgage servicing platform can unlock for
14:03the industry and for telling me what SAP Pioneer is up to. Thank you for your time. I appreciate it.
