The Art of Serial Entrepreneurship and Fintech with Mike de Vere

Episode Overview

Our Guest: Mike De Vere- Crusaded for Equitable Credit Access Through Innovative Fintech

Mike De Vere is the CEO of Zest AI, a company at the forefront of using advanced machine learning technology to transform the lending industry by promoting equitable credit access. Before his role at Zest AI, Mike had a distinguished career that spanned over three decades, including leadership positions at notable firms such as JD Power, Harris Interactive, and Nielsen. His extensive background in data analytics and customer experience has uniquely positioned him to lead Zest AI in its mission to make financial services fairer and more inclusive. Under Mike’s leadership, Zest AI has focused on developing tools that allow financial institutions to evaluate credit applications with greater accuracy and fairness, thereby reducing biases that traditionally disadvantaged underrepresented groups.

Mike De Vere’s approach to leadership in fintech is deeply influenced by his personal experiences and a robust educational background. He earned an MBA from the University of Southern California, a credential that complements his passion for both the analytical and ethical dimensions of financial technology. His engagement with his alma mater goes beyond academics; he is an enthusiastic supporter of USC athletics, embodying the Trojan spirit in both his professional and personal life. This blend of personal interest and professional expertise underscores Mike’s multifaceted approach to leadership, where he integrates personal values with business acumen.

In addition to his professional pursuits, Mike De Vere is an advocate for using technology to address societal issues, particularly in the financial sector. His vision extends beyond mere business success; he is committed to the idea that financial inclusion is a civil rights issue of our time, a perspective that shapes the strategies and operations at Zest AI. His work is driven by a belief in the power of technology to create opportunities for all, regardless of background, race, or gender. By pushing for changes in how creditworthiness is assessed, Mike aims to dismantle systemic barriers and promote a more inclusive financial landscape. His efforts reflect a broader commitment to social equity, making him a respected voice in both the fintech industry and the wider community advocating for social justice and equality.

Episode Transcript

Mike De Vere: Imagine if you had the same score, but then you had a policy called debt-to-income. So if that female, even though she might have the same amount of debt but is making 83 cents on the dollar with the same job, could trip and be pushed out for manual review and be denied by your underwriter. Just having the right credit score, that’s the right assessment of risk, doesn’t fully solve the problem. You have to deal with the policies associated with it as well.

Kevin Rosenquist: Hey, welcome to PayPod, where we bring you conversations with the trailblazers shaping the future of payments and fintech. My name is Kevin Rosenquist, thanks for listening. Today, I’m chatting with Mike De Vere. He spent his impressive 30-year career at companies like JD Power, Harris Interactive, and Nielsen. Now, he’s the CEO at Zest AI, a company that is using machine learning to help financial institutions safely expand credit access through more accurate risk prediction, faster credit decisions, and more inclusive lending. Both Mike and Zest AI are on a mission to ensure fair and equitable access to credit for all, regardless of race, gender, or sexual orientation. I have no doubt that you’ll enjoy my fascinating conversation with Mike De Vere. So you got your MBA from USC. Are you a big Trojan football fan?

Mike De Vere: Oh my gosh, it’s a family affair. So my grandfather, Grandpa De Vere, is one of four boys, orphaned at age 12 in Indiana, shipped to Los Angeles to work in the farms, and so as a 13-year-old, went to his first USC football game. Now, fast forward to my father being born, I’ve got a picture of my grandfather and my dad at age four in front of Tommy Trojan on campus, the same one for me, the same one for all my kids. So I love the school, my daughter is going there right now. So one out of five got it right.

Kevin Rosenquist: Well, I’m originally from Chicago and I’m a big Bears fan, so I hope your former quarterback is going to help us out.

Mike De Vere: Well, there it looks like they’re going to be sacking the team, Kevin. Like bringing in a receiver. So it’d be nice for the bears to return to their glory.

Kevin Rosenquist: I agree, and hopefully, Caleb Williams can be a help in that category.

Mike De Vere: He’s a generational quarterback. I mean, he’s special.

Kevin Rosenquist: I agree. Well, I could talk bears football all day, but let’s move on. You talk a lot about equality, equity, and the bias in society, particularly, as it pertains to access to credit, fixing that, at least from a financial perspective, is a mission of Zest. In the company, you’re the CEO of, is it more than that for you? Is it a personal mission as well?

Mike De Vere: Well, for me, it’s a question of what’s right. When my 30 years in business, I’ve got to work for some pretty cool companies. What I love about the company that we’re building at Zest right now is that we have a purpose. It’s not about making credit accessible to one group, taking it away from them, and giving it to another group. It’s really about ensuring that every American gets a fair shot. When I talk about fair shots, it’s really to opportunity. If you have the correct assessment of credit, that opens up possibilities. It’s that new car that you might be able to drive to a job that’s an hour away. Now it’s the ability to get your master’s or even go to college or things of that nature. So for me, it’s an issue of loving my country and wanting to make sure that every American gets a fair shot at credit.

Kevin Rosenquist: That’s good. I’m not a wealthy man. I just get that out of the way. I’m doing all right, but I’m certainly in a better position than many are, and it feels like the wealth gap is getting bigger and quickly. Maybe that’s just me but that’s how I perceive it. Am I right on that? If so, what do you believe is the biggest reason for the widening of the gap?

Mike De Vere: Well, man, we need more than our time. But if you look back 60 years ago and you look at a black American versus a white American evaluating their wealth, it has improved roughly 10% over that 60 years. So in order to close the wealth gap, wealth gap for black Americans, it might take over 500 years to close that gap over time. The issue is, it’s just not the credit score itself. It’s the entire lending ecosystem. That’s not an easy thing to solve because you have to get the government involved. You have to get regulators involved. You have to get the credit reporting agencies involved. You have to come up with a better way to assess credit. Then, oh, by the way, your strategies and your policy also have to align. So it’s not an easy problem to solve. That’s why we all have to be very upfront and purposeful about how we talk about it, that there is this racial wealth gap that persists throughout America. And we have to decide collectively that we can do better because it’s going to take being purposeful and taking purposeful steps about and your ability to do that.

Kevin Rosenquist: So if I’m going for a loan or a line of credit and there’s an African American or a Hispanic person with the same credit score, and same income trying to do the same thing, do I have a better chance of getting approved right now, you think?

Mike De Vere: With the same credit score, I think where your just the watch out would be on policy, is you can have a similar risk score, but then you can have a policy overlay. So I’ll give you a different example. Pew Research released last year a report that said the average female makes 83 cents on the dollar to their male counterpart. So there’s an income disparity still and that income disparity has been consistent over the last 20 years. So that’s also concerning as well between gender income disparity. So imagine, if you had the same score. But then you had a policy called debt-to-income. If that female even though she might have the same amount of debt, but is making 83 cents on the dollar with the same job, they could trip and be pushed out for manual review and be denied by your underwriter. Just having the right credit score, that’s the right assessment of risk, doesn’t fully solve the problem. You have to deal with the policies associated with it as well.

Kevin Rosenquist: Okay, you wrote a blog post a while back about the need for equitable access to credit. I want to read a quick excerpt. The lack of access to consistent, fair, and timely credit is one of our country’s most under-discussed issues. Considering its importance in Equitable Credit Access shapes the core foundation of our nation into one that benefits the privileged to the detriment of the marginalized.

Mike De Vere: That took a while to write that one, though.

Kevin Rosenquist: It’s very well written. Is it fair to say that you see the issue of lack of fair access to credit as sort of a pillar of discrimination in our society? Is it something that is keeping it going? You know.

Mike De Vere: Financial inclusion is a civil rights issue of our time. So Rodney Hood, one of our board members, who’s the former chairman of the NCUA regulatory body that sits on top of the credit unions, that’s his quote. I couldn’t agree more, because if you think about how pervasive credit scores are within so many decisions, listen, Kevin, if you want to get a job, are they going to run your credit score? What if you want to rent an apartment? Are they going to run your credit score? What if you want to get a car so you can drive an extra 30 minutes to get that better job, or they’re going to run your credit score for that auto loan? So it’s all throughout there. So how I think of it is that these industry scores have the average American in a chokehold. I mean like think about the sleeper hold. I’m not talking about a light chokehold, a serious chokehold that they’ve got on the average American. It impacts them in so many ways. So we have to be vocal and have to be talking about it. It shouldn’t just be a small group. It’s all leaders within this industry need to face this issue head-on. That’s the only way that we’re going to be able to solve it.

Kevin Rosenquist: How long have you been in fintech?

Mike De Vere: I’ve been with Zest for five years. If you look at my 30 years of experience, a consistent thing is taking data and translating it into insights. So whether you’re talking about being at J.D. Power or you’re taking data and talking about the customer experience, or you’re talking about the Harris Poll, taking data and maybe predicting an election outcome. So that’s the consistent theme. So today we’re taking data and by applying better math, trying to predict and understand who one should give a loan to.

Kevin Rosenquist: That’s a good Segway. I wanted to talk more about Zest AI. So the platform provides more accurate risk prediction, faster credit decisions, and more inclusive lending. Can you talk a little about how Zest achieves this?

Mike De Vere: It starts with the data. What’s unique about our approach is we believe one size doesn’t fit all. Let’s imagine we’re dealing with a financial institution that’s sitting in Hawaii. Favorite place to visit? We’ve got 15 customers there. I wonder why. So they’ve got a unique issue in the Hawaiian Islands. Guess what? There are Asian Americans and there are a lot of Pacific Islanders. The current system is set up with this industry score to do basically the mainland and the main population centers. So you might have an industry score that under-represents Pacific Islanders. You’ve got to have the right data that’s representative of the population that you’re trying to make this decision on.

Kevin Rosenquist: Sure, certainly.

Mike De Vere: The second area comes down to the model itself. Are you building a model that’s unique to that environment? So instead of building a credit model for the entire US, you’re closing the aperture in on the Hawaiian Islands and the markets that they serve. I’m using that as an illustration because it could be southern Alabama, it could be upstate New York, and what have you. So starting with the data and starting with a model that actually is trained on the communities that you serve but then looking at the model itself it’s beyond 15 variables, which is the current system, 15 to 20 variables. It’s a higher fidelity image of a borrower where you’re able to consume hundreds of points of data. So when I talk about data, I’m not talking about anything creepy like social media data or things of that sort. I’m talking about credit data, which most banks and financial institutions have today. But through the power of machine learning, you can create and engineer new features. So we talked about debt to income. Imagine if you had in your model both income and debt, all throughout your machine learning model. What would be a pretty cool signal though is looking at your debt to income over time.

Mike De Vere: I have five kids on Christmas time rolls, around, guess what? My credit utilization goes up. I’m buying a lot of presents, right for those five kids. So if you are able to look at me over time, you’ll see credit utilization goes up during the holidays, but then it drops down in that first quarter. The current credit system can’t do that because it’s looking at a point in time. It looks like you’re looking at somebody’s heart rate bouncing up and down, and you’re wondering where there’s all that variability in that credit score because it’s just inaccurate. So by consuming more data and applying this better math, you’re able to yield that decision. It doesn’t stop there. This is not about building a machine learning model and just dropping off and stranding you on an island by yourself. With the machine learning model, you have to know how to modernize your policies and the strategies associated with them. That’s where we spend a lot of time beyond the model itself understanding how to optimize those policies themselves so that we can get to that fair and equitable decision.

Kevin Rosenquist: Why do you think it’s taken so long to improve the ability to see someone’s credit, and the ability to make good decisions on a borrower, is it technology, or is it just a lack of desire to change, what do you think?

Mike De Vere: Well, there is the will, that’s for sure. So the majority of credit union executives, when we did a recent Harris poll, we found that over 80% of credit union executives and banking executives wanted to deploy AI within their underwriting ecosystem. So there is a will there. The question is, is there a technology able to do that? That’s where Zest has been mastering the use of AI for the last 15 years. That’s where we have come in, is this ability to create these tailored models specifically for a market, specifically for a business line that other companies just simply cannot match, they cannot match our tech stack. So that’s probably a big barrier, technology can be a big barrier. The second one is fear of change. So nobody gets fired for just keeping the status quo in most organizations, they’re not trying to be pushed to affect change. It’s like we’ve done it this way for 30 years, and I’ve been the chief lending officer for the last 30 years. It’s like, dude, America is different today. We have to modernize this process and embrace that. I’m that gray-haired guy, so if I can figure it out, I think everybody else can come along on that ride with me.

Kevin Rosenquist: I would say so. Just take somebody to lead the charge. Right?

Mike De Vere: Yes, to jump in. Today, we posted a video. Jenny Wipperman, she’s the CEO of ORNL, Oak Ridge National Lab out of Knoxville, Tennessee. She was our first credit union customer from November in 2019. She’s now the CEO of this new credit union. As we thought about the journey we were trying to lead in this industry, it was trying to keep credit unions and community banks and things like that ahead of what’s next within the economy. It’s easier for, let’s say, a large bank of ours to a customer of ours, like let’s say Citibank, for example. It’s easier for them to adopt AI. They’ve got hundreds of data scientists, they’ve got massive risk organizations. What’s cool about what we’ve done and Jenny’s interviewed, she talks about it, is that we’ve made it accessible to the smallest financial institutions that might be doing a few hundred applications a year versus a second.

Kevin Rosenquist: How has the response been thus far to the platform?

Mike De Vere: With our technology, it’s been exciting. I’ll tell you, we increased our customer count last year by 75%. So we’ve shifted from a startup to being that scale-up. How do you go from over 400 credit models in production? How do you scale that to a thousand in the next 12 months? That’s been pretty darn exciting. When we go to these conferences, there these moments where we would walk up to our booth. So imagine a big 20 by 20 structure with two screens and all these people, and there are 20 people deep. Everybody is just so excited that now there’s a technology that is acceptable, accessible by financial institutions of all sizes, but also they can deliver good things for the communities that they serve. Why wouldn’t you want to say yes to all of your members that walk in the door? People don’t wake up in the morning and say, I want to say no to giving them the opportunity, or I want to discriminate against a particular group. I sure hope they don’t. So this is where the Zest technology comes in.

Kevin Rosenquist: There’s so much we could talk about what you guys are doing to combat inequality. Obviously, we talked about it before. How much time do we have? But I’m intrigued to know more about the best race predictor and how it improves fair lending. Could you talk a little bit about how that works?

Mike De Vere: Let’s start with what’s the status quo. So the status quo today, when you build a credit model, you then want to do testing on it to say, hey, how am I doing with certain protected classes, men versus women, Latinex versus white Americans, and what have you? In order to do that, though, when you fill out a loan app, you’re not saying white male, Irish dude, you’re not doing that. You have to try to figure out or guess what somebody’s race and ethnicity is. So that’s where this concept called BISG comes into play. It takes your zip code and your last name. So if you look at my last name there, De Vere, there’s a town over in the northwest of France. So you’re thinking, well, it could be a Frenchman, it might be a Spaniard. Actually, I’m Irish, so this ability for us to predict is predicated on your last name and your zip code. That implies that everybody in my neighborhood is a white Irishman. Well, listen, America is different and it’s completely changed. So the yardstick that we use to evaluate inclusion within a model is completely broken. There is absolutely a better way to do it. So that’s where this Zest Race Predictor has come in. What we’ve been able to do with this Zest Race Predictor is instead of looking at two variables, we’re actually able to look at many variables, 10 to 15 variables, and what you end up with is an accurate prediction of risk.

Kevin Rosenquist: How does that happen? Can you go a little more detail? I’m curious to know how it helps to get a more accurate.

Mike De Vere: You’re pulling in more variables. You’re not dealing with just someone’s last name and zip code. You’re pulling in more variables to predict someone’s race. So what you end up with is something that’s roughly 40% more accurate at predicting race. So that matters. If you’re sitting here and have done all the right things, used the right data, built the right model and you have the right policy, now you’re going to get out your yardstick and say, have I done a good job? If you’re using a faulty yardstick, you don’t know you’re flying blind. So that’s where the Zest Race Predictor comes in.

Kevin Rosenquist: Got you. We talk about AI. Almost every episode that I do of these and it’s pretty hard to avoid, obviously. I don’t think that’s going to change anytime soon. Besides what you guys are doing at Zest, where do you think I could make a difference in fintech?

Mike De Vere: Let me start broadly with AI. So AI is not the boogeyman. It’s not like the advent of the internet. Al Gore created it for us. It’s not about the advent of the internet. It’s not the boogeyman. Ai, when built purposely, can be a force for good. What I love about what we’re doing is we’re actually being purpose-built around how we can help. Where I see AI evolving for fintech, in particular, is in the area of identifying fraud. So you have entire countries that are set up to defraud Americans and financial institutions. So can you leverage AI within a fraud environment? The answer is “Yes.” At Zest Labs, we’ve come up with some cool applications of AI to find out if somebody’s trying to cheat the system. That’s kind of cool. The other thing that we’re super excited about, and this came out of the Zest Labs is Lulu. I don’t think, Kevin, you’ve ever met Lulu, but she is our lending companion. She’s been trained in the last decade and a half of the learnings that we’ve had at Zest on how to apply machine learning to underwriting. Then what we’ve done is we’ve had Lulu look at industry data like, let’s say, a financial call, reports for all the credit unions, or let’s look at the inclusion data that we’re reporting to the government within the mortgage, and HELOC and inclusion for certain protected classes. So Lulu is able to look at all the history and all the questions we’ve ever gotten at Zest. It’s able to look at publicly available information. Through a large language model, she’s able to help us answer questions. So if I’m the CEO of a credit union, I can sit back and ask, how are my charge-offs versus my peer set at a billion dollars? So what would normally take a week or so for their analytics group to come back with an answer she’s answering in 200 milliseconds, responses like that. So for me, I think where AI, I get excited specifically on fintech, is we can create an environment where the speed of decision-making for executives becomes that much more rapid. We’re able to be agile and adapt to changes in the market conditions.

Kevin Rosenquist: What are you most excited about outside of fintech as far as AI is concerned?/

Mike De Vere: Gosh, you know what I really love is getting into the medical industry because of the ability to consume articles and research reports globally, let’s say you have a particular illness, the ability for AI to consume trillions of points of data that a human could not. I think it has great promise in helping humans in our longevity and solving issues that have been, frankly, vexing us. They’re beyond medication. It’s like homeopathic remedies. What about the 4000-year-old Chinese herbal teas that they’re using to help Eastern medicine solve issues? What are their principles within that? So AI, I think can help create a healthier, more vibrant society for humanity. So for me, that’s a big one, super scary. But I’m excited about it.

Kevin Rosenquist: That’s a good one. You see stuff all the time about the progress they’re making and maybe curing diseases or at least detecting certain things early because of AI. I saw something recently about, I believe it was prostate cancer that they’re using AI to catch that earlier. That’s pretty cool, there’s nothing bad about that.

Mike De Vere: To evaluate the images that you’re getting off of, let’s say an MRI or a CAT scan or what have you. I mean, dude, I do math. I have no idea what this stuff is. My daughter just finished her MCAT, which was like the seven-hour medical school test. She and I were talking about this very point of all the basic data you just have to memorize versus a doctor harnessing AI to help make better decisions to access data that they might not recall, which I think holds great promise.

Kevin Rosenquist: I agree, that’s a really good one. What’s your biggest fear as far as AI is concerned?

Mike De Vere: A tech company without purpose is dangerous. If we just wanted to optimize solely for, like, if you think of the credit system, if we were just solely optimizing for no risk, no charge offs, the most accurate model but also at the same time weren’t looking to make sure that you’re doing good for society? I think it does worry me. So for me, when I think of AI, it can unlock true human potential where we’re truly being able to be creative. For me, it has the ability to unlock true human potential, which is we’re all creative. Let’s stop trying to do the basic tasks in the day and spend more time imagining where we can go as humankind, how we can solve global warming, how we can solve societal issues, and how we can solve hunger. I think spending time like that, but using AI at the same time is pretty cool.

Kevin Rosenquist: I’m always curious because you’ve been at Zest for, I think, five and a half years. I think I saw it on your LinkedIn. Most people didn’t know AI was coming, at least not in terms of the practical use that we have now with ChatGPT and all these other AI programs and stuff. When you first started at Zest, did you kind of see this wave coming at any point? When did you realize that this was going to be big?

Mike De Vere: Well, I’ll tell you, we were founded in 2009, so the founders of Zest were so far ahead of the curve as far as the application of AI to this underwriting question, I think it’s pretty cool.

Kevin Rosenquist: That’s remarkable, 2009.

Mike De Vere: For us, though, our momentum in adoption happened right around the pandemic, when organizations were trying to reflect and think about how they could broaden access to credit at a time when America needed access to credit. So there was this need for new tools that were more accurate. Now you build on that momentum because then you have a banking crisis and inflation goes through the roof, and all the banks and financial institutions are lent out. So they have to figure out how to lend to the right people, decrease their charge-offs, but also increase yield. So for us, some of these challenges and headwinds in the economy have helped with the adoption of AI because you’re looking for nontraditional solutions. If everything was going swimmingly, everybody had equal access to credit. There’s no need for innovation. But the very fact that there is that foundational issue in this credit ecosystem but then you layer on top of some of the financial headwinds, it drove some significant adoption for us.

Kevin Rosenquist: We’ve talked a lot about equity today. What do we need to do to ensure equitable access to AI products and services?

Mike De Vere: Equitable access to AI products and services or credit?

Kevin Rosenquist: I would say just to the AI side, I’m just kind of still talking about AI. I’m curious just to know what you think because not everybody has equal access to the programs and to the possibilities that it provides. I was just curious what you thought about how to keep it in everybody’s hands.

Mike De Vere: Well, first is access. So there are technology deserts all across our country where there’s limited wi-fi and limited access to computers. So having large tech get involved there, I think is going to be important. I mean, we’re super blessed in our community that Google offers Chromebooks through the school at no cost to families. So that wasn’t something that existed in the past. So you have big tech coming in and helping and providing access. The second thing is you have to have a connection to the Internet. So whether we’re talking about Starlink, Elon’s, or things of that nature, or Google, that trying to promote broad access to the internet I think is going to be critical. Then I think every company, thinking about what they can open source and make broadly available at no cost to the market. So we talked about this Zest Race Predictor and spent a lot of time developing that IP. We felt it was so important for society that we just open-sourced it and made it readily available at no cost to everyone. So having the right organization, you’ve got access to the hardware, you’ve got access to the internet and then the third area is around really trying to make available what you can as a technology company more broadly to the community.

Kevin Rosenquist: What’s next for Zest that you can talk about anyway?

Mike De Vere: We are excited about moving beyond the origination question, which is very important. Now you’re talking about looking at evaluating a bank or financial institution’s portfolio to understand if it’s healthy. Why I’m excited about that one, Kevin, is certainly from a business perspective, you want to understand, hey, have I made good loans? How does my portfolio look? But what’s cool is we can start predicting and identifying customers who might be at risk of going to collections and who might be having new financial problems. Let’s say they were an A-Tier loan and they migrated down to a C-Tier in the last 90 days. We’re able to flag that for the financial institution so they can get out ahead of it and so that they’re not stuck in collections or some collections agency is calling them that you’ve got a bank credit union that’s doing the right thing, reaching out to a consumer and offering them to restructure their loan so that they don’t go underwater. So, excited about that. Certainly, from a defensive perspective, we’ve got a cool fraud offering that it’s unique to Zest that will be announced here shortly, I think, next month. I think for that one in particular, I’m looking forward to that new force field putting up. I mean, I guess I’m a Star Wars and Star Trek junkie, but imagine that force field that you’re able to put around your lending ecosystem because the fraudsters out there are using very advanced techniques. So why not build a force field that’s strong enough to withhold that attack? That’s what we’ve created. So excited about that. Then there’s Lulu, our darling, She has such great promise. All throughout this year, we’re going to be rolling out new versions. She’s just going to continue to learn. So, stoked about that.

Kevin Rosenquist: That’s awesome. Well, I love what you guys are doing, I love the crusade. I love the desire to make things equal. It’s certainly something that we need very badly. So kudos to you and Zest for doing that. The website is zest.ai. Mike, thanks for being here.

Mike De Vere: All right, brother, I’ll see you soon, cheers.