Trust is the new currency in an AI-driven market

Transcript

John Rogers:

I would say at Cotality, there's two principles that underpin us in terms of the models that we provide. So, one is the explainability of our models that is audible in terms of inputs and outputs, clearly communicate confidence levels and sensitivity analysis. The second area, the central principle is ensuring that human in the loop in the process.

Maiclaire Bolton Smith:

Welcome to Season six of Beyond the Buildings by Cotality. I'm your host, Maiclaire Bolton Smith, and I'm just as curious as you are about everything that happens in the property industry. On this podcast, we satisfy our collective curiosity, explore questions from every angle, and look beyond the obvious. With every conversation, we illuminate what is possible. To kick off the year, we're starting with a conversation about the technology that everyone seems to be talking about: AI. A decade ago, home buying moved at a human pace. Today, AI recalibrates prices, financing, buyer profiles, and insurance in seconds. It's shifting how property markets function and how people navigate them. But this progress comes with obligation. The industry must ensure that predictive tools expand opportunity rather than narrow it; that they model resilience, not exclusion. So to talk about what happens when property becomes data, we have our Chief Data and Analytics Officer, John Rogers. John, welcome back to Beyond the Buildings.

John Rogers:

Hey, Maiclaire, it's great to be back. Thank you for the invite.

Allie Barefoot:

Before we get too far in this episode, here’s a friendly reminder about how to see what’s coming up next in the property market. To make it easy, we curate the latest insight and analysis for you online. Find us using the handle @Cotality on all of our social media channels. But now, let's get back to the show.

Maiclaire Bolton Smith:

Well, I’m really excited to start off the new season with you, and I can’t believe we’re already at Season six here. But let's start with this crazy big idea that models are becoming the market. So flood maps, digital twins, predictive simulations—things that have always been really big in tech are really no longer just tools. They're reshaping real decisions. So, what are the, I mean, the implications of this shift?

John Rogers:

Yeah, no, Maiclaire, it's definitely a big shift. I think the good news is I think we’ve been on this journey for about two decades; it's just very amplified right now in terms of the explosion of AI that we’re all reading about and using every day. I think this shift means that these digital representations, which is really now a primary source for, like, for value, risk assessment, and operational efficiency—basically, that's dictating a lot of real-world capital allocation and public policy, so it’s trillions of dollars. I can give you an example. At Cotality, we help our clients use these models, such as a digital twin, a flood model to inform insurance pricing, loan terms, property investment decisions. That's good for both our clients, our housing industry, and really for you and I to, you know, make better decisions and protect our homes. So, you know, we're really passionate at Cotality that we spend a lot of time, effort, and investment in responsible AI. So that's really, you know, ensuring that there's not entrenched bias, that our models are transparent, and, you know, fundamentally that our clients can trust the outputs that we provide.

Maiclaire Bolton Smith:

And that trust part is huge. And I think, I mean, that's kind of where I really want to talk about today, that we think, you know, AI can simulate thousands of different futures, forecasting what potentially could be. But like, for a homeowner of a specific property, the property hasn't moved. Nothing's changed from their perspective. And let's say a model is showing that that property is inside a simulated flood zone, and maybe that'sthe flood zone of the future versus, you know, where we are now, just if we think of how the environmental life is changing. How do we ensure that our models are informing decisions without dictating them? Because from a homeowner's perspective, things haven't changed, but all of a sudden their insurance premiums are rising. So, how do we navigate all of that?

John Rogers:

Yeah, there's, I would say in Cotality there's two principles that underpin us in terms of the models that we provide. So one is the explainability of our models—that is audible in terms of inputs and outputs, clearly communicate confidence levels and sensitivity analysis. And obviously, we go through a huge amount of internal governance mechanisms and peer reviews, and if you really think about it, we go through thousands and thousands of client reviews from, you know, insurers, reinsurers, banks, and also do third-party reviews. All of that is to say that’s key to gaining trust.We don't want to be seen as a black box. The second area, the second principle, is ensuring that human-in-the-loop in the process. As an example, whether it's a real estate agent confirming an automated listing—so we've, you know, gone nationwide with that in the last couple of months—to, you know, I run our QA function on our manufacturing plant and how they sample data to ensure quality within that process deep in the bowels of our manufacturing plant for data and all things property. So, those are the two underpinning principles. But to maybe create an analogy, I always think, do you remember when the mapping apps came onto our cell phones?

Maiclaire Bolton Smith:

Yeah.

John Rogers:

And at first, when we first used it and said, you know, we want to go to location X, the trust factor was probably, you know, 50/50. But because we've used it millions, or tens of thousands of times and across the world millions and trillions of times, we now just implicitly trust it.

Maiclaire Bolton Smith: Yeah. We rely on it.

John Rogers:

We rely on it. So and, and people might not be aware, but our AVMs are used every day to underwrite loans for decades. So they’ve been used tens of thousands of times, and that frequency of use and trust in a big, you know, financial asset class—you know, just in the residential alone in the US is 50 trillion—they’ve been used in, you know, millions of times, they’re hardened, widely used, and trusted. And some of it is to do with that frequency of use and the, you know, within the mortgage, insurance, and reinsurance markets. So, yeah, that's, our AVMs are definitely stress tested a lot through lots of different avenues.

Maiclaire Bolton Smith:

Yeah, that's a great analogy, a great example too because I mean I go back, my goodness, like not even maybe 10 years ago and using one of these map apps and going to a new, I remember I was going to a new outlet mall and I ended up in a farmer's field because it, the map didn't quite know where it was. And it, but I would never think to not trust it today. I would, you know, going to any new location, I type it in and I would assume it's going to get me there. And I had forgotten about that hesitation that you had earlier on using these things when, you know, it was so far advanced from printing, I mean, printing out MapQuest and like taking it in your car and like figuring out where you were going. So, yeah, so it's interesting how like things evolve.

John Rogers:

Yeah, no.

Maiclaire Bolton Smith:

And trust evolves.

John Rogers:

Yeah, exactly. So it's trust but, Maiclaire, for myself personally, without that map technology, I get lost everywhere. I'm, my spatial orientation's terrible.

Maiclaire Bolton Smith:

Oh, you're not the only one. You and me both, John.  

John Rogers:

So it’s amazing technology. I fully trust it. I fully trust it.

Maiclaire Bolton Smith:

Yeah, I do. Trust, rely, and you know, know that it will get me where I need to go. Like that, so how trust has evolved and grown over time. So, yeah, it's really interesting to think. And I guess that whole human perspective, something we've talked about a number of times on this podcast is in 2025 we did this survey called From House to Home, and I had a really deep conversation with Anand Srinivasan on your team about this last season. But it was the, really the big takeaway, which was surprising to many, was that yes, we're moving in this direction of relying on AI so much, but people still not just prefer, need human input to make key decisions. Sowhat does that tell us about kind of the limits or the reliability of AI if it’s, you know, AI still needs that human intervention not just to kick it off, but to guide you through it?

John Rogers:

Yeah, no, and Anand's definitely a deep thinker, so that was a fascinating discussion. I would say that that research piece that Cotalitydoes is, it’s a fascinating piece like looking at the trust factor by demographic when you're finding a house, buying a house, and trying to protect your house and how much would those demographics trust AI. I think in the, in the example of real estate, I think there's probably two factors that somewhat counter against it, or are challenges for AI. One is obviously it’s an infrequent activity unless, you know, a small population really that are investors that do it again and again and again. For you and I, it’s an infrequent activity and hence there's apprehension. And if you compare it to say like a map technology in our previous discussion, you know, that we use that every, every day. The other part is obviously buying a house is a very emotional journey. There's so many factors underly, it’s a part of your identity, what your goals, your aspirations, your fears.

Maiclaire Bolton Smith:

You're spending hundreds and thousands or millions of dollars.

John Rogers:

Yeah, yeah. So and there's, it would AI can certainly help, but it’s not going to replace that human interaction totally. So it's, no matter what demographic, that research no matter what demographic, at some point you want someone to hold your hand and take you through that journey. So it's a massive emotional decision.

Maiclaire Bolton Smith:

Yeah, and I guess that, that emotional decision leads to like the human judgment of kind of the limits of stabilizing AI-led systems. Like, where do you think we are with the role that human judgment plays?

John Rogers:

Yeah, so massive role. Humans, us, you and I, Cotality, we establish the guardrails, apply surveillance, the reality checks, how to interpret the outputs, the probabilities into actual decisions—that requires humans. It doesn't replace that. You know, just to give you an idea, you know, in Cotality, we ensure all of our statistical models, deterministic models, are compliant with over 130 federal and state regulations. So that's, that's good. Like, we want to make sure that it's unbiased, that it’s trusted, and it’s fair for use for both our clients and for you and I as homeowners.

Maiclaire Bolton Smith:

Yeah, no, that's great. And you know, John, as you were talking, I kind of, I started running an example in my head and kind of playing on this, you know, human intervention needed, and I'm like, oh well, we got to the place with the map example that we no longer needed somebody holding our hand, we didn't have the human intervention, but we do because so many of those map apps are crowdsourced and they pull in real-time data from crowdsourcing from all the other phones. So there is thathuman intervention even though you may not realize it.

John Rogers:

Correct. And like think about, you know, there's traffic jams and it's verification of traffic jams, and there's something on the road and it's that real-time feedback.

Maiclaire Bolton Smith:

Is this still there?

John Rogers:

And you trust it more because there’s human input. And there’s a human feedback loop, absolutely right. Absolutely right, yeah.

Maiclaire Bolton Smith:

Yeah, so it's interesting how even places where we feel like there isn't human intervention, there actually is, you maybe just don't recognize it.

John Rogers:

Yeah, exactly.

Maiclaire Bolton Smith:

Interesting. If we stick with that survey, I think, you know, the big thing that we talked about in that survey was speed and really looking at like the acceleration that AI was providing and I mean, the world is everything about being more fast-paced now, right? Everything is all about acceleration is king and how can we like run on speed. But when it comes to buying a house, that speed can be very frustrating, very stressful, and anxiety-inducing because you are moving too fast. So I guess where, where do you know that you're drawing the limit between the speed becoming essential for, you know, being productive, but not being too fast that it doesn't induce anxiety on somebody who maybe buying a home?

John Rogers:

Yeah, no, very interesting question. I’m not totally convinced that speed is always the number one priority. Obviously it’s a, definitely a contributing factor. I think with the use of AI, the three key areas is confidence, clarity, and peace of mind, especially in the homebuying process. Speed is advantageous, but it can cause stress, pressure to make a decision, too many notifications, as an example, you know, when you reach out to, to an agent to look at homes. And this, I just, I just came out of, in Cotality we were on a yearly hackathon, we had 82 initiatives, amazing stuff. And one of them was actually in the, you know, we support about 50 to 60,000 appraisals and there's a number of alerts that we give out during the appraisal process to both lenders and appraisers. And to be honest, it's actually sometimes it’s counter productive, it actually causes stress for the appraiser, the lender, and overall the homebuyer selling process is a lot of noise. So we're actually using GenAI underneath the covers to understand the patterns, weed out the really key notifications, and then reduce noise levels. Again, just pulling it back to confidence, clarity, and peace of mind for both the appraiser, the lender, and then obviously for you and I if we’re in that home buying or home selling process. So yeah, no, speed is not the number one priority, I would argue.

Maiclaire Bolton Smith:

Yeah, and I think, you know, that's so interesting just to think of like the whole human psychology of all of this too. And I think, you know, here at Cotality, we, we can track things that kind of understand human psychology at work through this whole home buying process and specifically our OneHome product—it works on kind of tracking—and you know this better than I do—but kind of it tracks micro-signals. So things like if somebody opens a document 15 times but doesn't sign it, or if they like hesitate on something. Like we can track all of that, and that may give some indications to agents on there may be, you know, some places where we need human intervention to restore confidence in decision making. So I think that's really interesting.

John Rogers:

Yeah, I think this is where, where Cotality shines and can make all the property professionals that we help AI superheroes to help you and I as buyers and sellers. So just understanding those micro-signals is very important. So imagine we understand, you know, the, to your point, the length of time looking at properties, how many times you visited that property, where in the process as a buyer you talked about, you know, hesitating or dropping out. From a, if I’m a real estate agent, that’s really important because now I offer a very personalized service in that localized environment to understanding what homes, types of homes you like, interest that you have, points of interest that you might go to, rather than just sending a generic email which we get far too many of in this world and you know the vast majority of the time we just delete or just ignore. So that's where AI and those micro-signals I think are really helpful for you and I.

Maiclaire Bolton Smith:

Yeah, you know, oh, John, I saw you look at 15 different ranches in Texas. So why am I sending you properties in, you know, like, do you know, like it kind of understands even if you may not say it, if you start looking at it, it might, you know, be like oh, you know, maybe you are ready to move, we didn't think you were ready to move but you're.

John Rogers:

It's a different conversation.

Maiclaire Bolton Smith:

It's a different conversation and one more, you know, more inclined to feel more personalized than it being just like, oh here's a random form email that people may, you know, not even pay attention to. And I guess that transparency, that transparency is really kind of what I want to talk about now and the whole transparency and traceability of the outputs and human intervention as part of our core AI and what we do. I guess what does that look like in practice? And how do we build guardrails into our, into our systems?

John Rogers:

Yeah, sure. I'm very fortunate Maiclaire in my job I look after our 21st century data and AI manufacturing plant on all things property and location. And we’ve spent Maiclaire hundreds of millions on this manufacturing plant, which is great. And it literally, when I say that term, I literally mean from sourcing to integration to QC to governance, data stewardship, to our data research and resolution teams to make sure our data is as primary as possible to literally distribution to all of our clients and in all the industries we serve. And there’s to back to your question about those guardrails, so there's three underlying core principles. So there’s the transparency in all the data inputs that we gather—so there's 22,000 data sources that creates that truth dataset on all things property. We ensure that it's the best ingredients, constant QC, different methods, redundancy, feedback loops, cross-referencing from imagery to MLSs to spatial information to tax information, wrapped with operational measurements that we run 24/7 wrapped with governance. And literally I work with teams that are trying to improve 0.001 improvement in data quality. Like that’s how refined and it’s literally like a manufacturing plant. And that's, you know, thanks to my predecessors like Sachin Rajpal and Debbie Matuska. I’m just standing on their shoulders.

But, and there's the other two just to quickly round it up to, obviously we discussed parts of this like the traceability in the outputs again. So I’ve talked about inputs, now it's the outputs, so we trace, manage, audit literally billions of interactions and algorithms through our data, the derivatives, the models into the hands of our users so we have a complete audit trail. And then just referencing back to yes, we have this machine and we need human-in-the-loop brakes if you like through operational metric sampling, client feedback, auditing, governance boards, and so forth. So sometimes the other way I literally sometimes say it's like the, this manufacturing plant is like the Intel processing chip for the housing industry.

Maiclaire Bolton Smith:

Oh, interesting. Yeah.

John Rogers:

Yeah, like we know it's there in a laptop, the Intel processing chip, and we trust it, that’s what we are to the, to the housing industry.

Maiclaire Bolton Smith:

Yeah, I love that. It’s the, we are the Intel inside of the property industry that you just trust it, you know it's there, and you don'treally see it working, but it's there and you trust it.

John Rogers:

That's right, exactly right. Exactly right.

Maiclaire Bolton Smith:

Well, two words that have come up a lot today are transparency and trust. But the other word I’m going to throw into that mix is accuracy because you know, that's what’s going to build trust and transparency is, is accuracy. And I think if we think of models from everything from fraud detection, home valuations—like we’ve talked about a lot of these today too—is what, what are the things that we think matter most that help build trust specifically when we’re talking about these different kinds of models, whether it be, you know, fraud or valuations or hazard? Like, what's the number one thing or the things that really help build trust?

John Rogers:

To gain trust it’s reliability, fairness, and explainability. And what if I break that down a little bit more, obviously think about data quality to ensuring there's no bias in our models, that we have an audit trail from inputs to outputs. And to make in terms of accuracy I probably just to good to give an example. So we have an attainable housing solution that basically, this model identifies land to build a certain type of home with a certain material and labor cost associated with it for a certain median income, mostly in this case you know for attainable housing it’s like essential workers. And then how do you provide an ROI back to the land and building developer? So in terms of accuracy, the first time we piloted this a few years ago, the city that we were with—so city officials, you know, builders, developers, MLS group, title company—what they didn't tell us is that they already knew the answer.

Maiclaire Bolton Smith:

Oh, okay.

John Rogers:

So we ran it and and the challenge for the housing industry and development is it’s often in the planning and feasibility it takes double-digit months to several years to get the tick in the box. So we ran it in the, we ran it in the afternoon and we were literally a few cents out.

Maiclaire Bolton Smith:

Wow. Wow.

John Rogers:

So that’s and again, I can't take the thanks for that, it’s literally the data scientists and the data gurus who understand how to model that particular solution to solve a challenge in the fast-forwarding housing development. So we’ve now done a few cities around the US and we’re just more and more piloting and you just understand some of the nuances in localized areas. But I was just trying to give you the idea of that’s the accuracy that we want, and obviously Maiclaire as soon as we ran it, because we didn't know that they all knew the answer, suddenly the trust level just goes up exponentially. That we’re out by a few cents on a multi-million dollar investment.

Maiclaire Bolton Smith:

Trust verified. You know, verified by accuracy.

John Rogers:

Exactly, yeah, yeah.

Maiclaire Bolton Smith:

I now have trust in the map app because I can go to that outlet mall and not to the farmer's field because it has proven to me that it's now more accurate and it’s learned, and it's just that growing accuracy to build that trust.

John Rogers:

Exactly right.

Maiclaire Bolton Smith:

Exactly right.

Allie Barefoot:

AI is changing how property is priced, financed, and insured. It's recalibrating efficiency to expose new questions about fairness, transparency, and trust. At Cotality, we've been a part of this transformation from the ground up. We've also been tracking these changes to understand how to guide the industry towards a future where intelligence moves fast, but accountability is never far behind. At every step, our insights return the same principle: housing may be powered by data, but it's lived by people. Register for Cotality's webinar to learn more about the questions we've been looking into and what this means for the future of property. The link to register is in the show notes.

Maiclaire Bolton Smith:

Well, John, you've been here enough to know that I like to end these episodes with if you look into your crystal ball, so pull out your crystal ball. And I guess, this feels like a weird question to ask because AI is, is changing so rapidly. But how do you see AI evolving in the property market, you know, in your crystal ball? And I guess how do we ensure that the predictive tools that we have are expanding opportunities rather than narrowing it and really being accountable for the models that we're creating for the future of property?

John Rogers:

Yeah, I'm just so excited and so fortunate to be in this position, but just working with a lot of Cotality product managers, scientists, technologists, we really are on the, we're making great strides to make every home resilient, more energy efficient, and lower our carbon footprint. Sowhat does that really mean for you and I as home buyers and home homeowners? Reduced insurance premiums, lower energy bills, and you know, take care of the planet and build sustainable communities. We have all the jigsaw pieces and globally we're tackling each of these in a series of steps, but I can see, we can see the path now, which is just very exciting and, yeah, just a great opportunity.

Maiclaire Bolton Smith:

How, ha, here's a fun thing, how do you think your crystal ball will change if I ask you that same question in a year?

John Rogers:

If in a year's time, I think, I think the advance of AI will help us get there quicker. I think this is like, in my head I think this is like a 5 to 10 yearjourney in my head, but that’s because I've got I’m thinking from my experiences. I just think the, there was a, a data scientist that was just talking to me about running this model which took, it was a massive model that we have and it takes, you know, several, you know, hours to days sometimes, it’s a heavy compute and now we can run it in under 5 minutes.

Maiclaire Bolton Smith:

That's crazy.

John Rogers:

Just it's, so I think I’m probably being, I think in a year's time I'llprobably be saying I'll be narrowing my horizon in terms of of can we make this a reality. But right now I think this is generational, basically.

Maiclaire Bolton Smith:

Well, I have a feeling a year from now, John, we'regoing to be back here having the same conversation and kind of seeing how things evolve. So, yeah, no, John, it's always great having you. Thank you so much for joining me today on Beyond the Buildings by Cotality.

John Rogers:

And thank you so much for doing this. Thanks for having me on.

Maiclaire Bolton Smith:

All right. And thank you for listening. I hope you'veenjoyed our latest episode. Please remember to leave us a review and let us know your thoughts and subscribe wherever you get podcasts to be notified when new episodes are released. And thanks to the team for helping bring this podcast to life. Producer Jessi Devenyns, editor and sound engineer Romie Aromin, our facts guru Allie Barefoot, and social media duo Sarah Buck and Mikaila Brooks. Tune in next time for another conversation that illuminates the ideas that will define the future.

Allie Barefoot:

You still there? Well, thanks for sticking around. Are you curious to know a little bit more about our guest today? John Rogers is Cotality’s Chief Data and Analytics Officer and oversees research and development for the company. Whether it's driving new solutions that understand the impacts of the real estate economy due to climate risk or it’s building groundbreaking models that identify suitable land for affordable housing development, the R&D group tackles major housing issues and works with many clients across the housing industry to drive growth and mitigate risk on their book of business.

‍