EYEON Fintech: Lending Innovation, AI, and the Future of Credit

Simon:
Welcome everyone, and thank you for joining us for another episode of our ION Podcast. I’m Simon Darchis, your host and head of fintech products and strategic initiatives at FinWise Bank. In this edition, we’re keeping an eye on the current trends in lending, from changing our behavior and product mix to the practical realities of product ideation, partner alignment, and scaling. To provide insight on the topic, we have Ed Walters at the mic. Ed is vice president of lending partnerships at Upstart.

Ed, thank you for joining us.

Ed:
Simon, thank you for having me on today.

Simon:
And we’ve known each other for a few years, but I think it would be good for everyone that’s listening to us to hear a little bit more about your background, what you guys do at Upstart, how you got to Upstart. It’s always fun to know about the person that you’re listening.

Ed:
thanks. Let me start with who Upstart is, and then I’ll tell you kind of my journey, So for those of you that haven’t heard of Upstart, Upstart is a leading AI lending marketplace that we really partner with banks and credit unions to expand access to affordable credit. And so what does that mean? So our goal is really to reduce the cost and complexity of borrowing for

all Americans. And we do this through our proprietary AI models that help us, I’ll say, really remake the entire lending process. And if you think of like our long term goal as a business is our vision is to become this always on everything store for credit, where we can automatically approve borrowers at the right prices instantly and effortlessly. And many people have known us from a personal loan kind of product. That’s where we got our start.

to now we’ve expanded into multiple asset categories, ⁓ auto retail, auto refi, ⁓ home equity, as well as small dollar lending. So we’ve got a lot of ⁓ expansive products around the consumer lending space. Now, my journey, I’ve been in financial services for over 30 years. And I really had what someone described as a very

zigzag career all around financial services. And I’d probably say it break it down into four parts. if you kind of think of like chapters of my career, started very early on in technology. And that was kind of that idea of implementing, developing, supporting systems right out of college, know, date myself, but you know, I thought I was going to be a cobalt developer, really, you know, focused on getting into the

the guts of the mainframe systems. And there, I really learned the critical importance of quality and controls. back then when I started, it wasn’t unusual for people to push code straight into production with very little testing. that experience…

in technology and starting my career there was really around respecting speed and control. It’s how do you move fast, but not without having the right rigor. Now the second chapter I’ll say of my career was in risk management and there got involved in business continuity well before COVID, before that was kind of a mainstream. was before hand sanitizers were everywhere, before working with.

Simon:
Right.

Ed:
working from home that we all do now, that really didn’t exist when I was in this risk management space. But it gave me a true enterprise-wide perspective, really understanding the moving parts of an organization, how we connect, how we depend on each other, and really thinking about the value of scenario planning and readiness, especially when things don’t go as expected. And probably the key takeaway from that part of my career is really about

you know, maintaining this just calm demeanor, no matter how stressful situation gets, which is important in a high moving industry that we’re in in the FinTech space. And then the third chapter was around project management. So it was building project management offices, really building these internal consulting groups. And there I got to hone my focus on really making sure we were working on stuff that really matters. And so for any organization,

I’m sure you’re the same. You’re pulled in a lot of different directions. But success really comes down to, we have clarity of purpose and priority? Are we aligning the teams on the right items that are most important? And are we giving those priorities the attention? And then that brings me in kind of my current chapter. the current chapter is where I’m at today of consumer lending. ⁓

I’ve been able to see this world both from inside financial services organization and into my current role at Upstart where I get to work with bank and credit union executives every day to really help them deliver a better lending experience. And this chapter kind of brings everything that I’ve learned, but then also adds this critical dimension that you and I are working on on a regular basis about having an impact on real people. And so that’s, you you take personal loans.

so many of these personal loans, the primary use cases is debt consolidation, right? And there’s all sorts of reasons that people can have debt, right? everyone’s ever had debt, they understand the stress that it creates. And we know that a well-timed personal loan can give someone that first good night’s sleep that they’ve, you know, that they had in a long time.

Simon:
Mm-hmm.

Ed:
And they can see a path forward, right? And so that is what’s rewarding of the work that you and I get to do on a daily basis is we’re doing stuff that can kind of have a direct impact on someone’s life. And so that gets me up and out of bed every morning, which is exciting to be a part of.

Simon:
You know what’s super interesting about your experience and I guess I never really realized it For the many years that you and I have worked together. I didn’t know you started as a tech guy Which is funny because when you think of upstart and I think a lot of people that have been following kind of the upstart story Really cannot tell you whether it’s a tech play or a landing play

And I’m sure you can feel it through the experience that you’ve had in applying those first few years as a tech guy to your current role at Upstart. You have a little bit of tech, have a little bit of financial services, and it’s interesting because I’m assuming you guys think about products differently than a very traditional shop would. You must be also looking at it through tech lenses. So how do you guys think of…

of innovation, right? Do you guys follow the trends and you try to act on it? Do you guys try to be at the forefront of trends and start what’s the process, at Upstart, like how do you guys come to a decision that you’re going to do HELOC like you’ve done, you know, over the past couple of years?

Ed:
Now, that’s a great question. I would start it with, we don’t approach it as we don’t start with a product, right? I think we’ve always approached it by starting with the customer and then the problem that we’re trying to solve, and then just making sure that we’ve got a good, deep understanding of why that matters.

I’ll give you a couple of examples. Obviously, ⁓ where the personal loan product ⁓ materialized is the traditional ways that people were ⁓ given credit and assessed credit worthiness was really pricing many people out of the market, whether they weren’t getting rates at all or

they were getting rates that were extremely high because of traditional factors. And so the problem was there’s got to be a better way that we can enable access to credit. And so that’s where you bring in AI, that’s where you bring in a lot of automation, a lot of things that allowed us to differentiate ourselves and kind of come to the table. So now I’ll pick kind of like our two other newer products. So you think of the auto retail solution.

The consumer is one of the most pain points that you talk about is kind of like going into a dealership and buying a car, getting financing. It’s very confusing. And are you always getting the best rate? And so we felt, okay, here’s another problem, right? The consumer doesn’t feel that they’re getting the best rate. And then you also have this new stakeholder of the dealer who the dealer is also trying to sell a car. And so how do you bring a solution?

and bring your learnings in to kind of solve that problem. And so that’s where we saw an opportunity that we could build upon what we’ve already done in the personal loan space to bring a lot more automation. And so when you have more automation and you have better data, you can make better risk-based decisions, which then leads to lower APRs. Lower APRs make a consumer more happy that they can afford the car. It sets them up for success.

Makes the dealer happy because more people are qualifying for cars. So it’s it’s there of like we were trying to solve a problem and then on the HELOC The HELOC issue is just the time it takes right? We you look at the trillion dollars of equity that’s sitting in all these homes, and a good majority of people took took advantage of when Rates were at their lowest kind of during that code period and have mortgages that are sub three percent and I am

So grateful that all those people got to take advantage of that and they’re rewarded with very low monthly mortgage payments. However, now they have a lot of equity in their home that they’re not going to go and get by refinancing their mortgage because you’d be foolish to do that. So now they’re looking at other ways. And so they thought, okay, here’s that I want to tap in my equity. But the problem is that traditional HELOCs just take too long to get access to that.

that capital and we are a kind of Netflix, Amazon culture, right? I just ordered something today. My smoke detector was out of ⁓ nine volt batteries and instead of going to the store, I grabbed five minutes. I went on Amazon and my batteries are going to be at my front door tomorrow morning. And it was like, all right, I have a problem. I want it now. And so that is spilled over into financial services.

Simon:
Right.

Ed:
And so from a consumer standpoint, you’re looking at, you use the HELOC example or you look at the auto retail example. We started with a problem, we started with the customer, who are we trying to serve? And then come up with solutions that then can make a difference and differentiate ourselves.

Simon:
What’s interesting is that it sounds like your product thesis always goes back to what started this industry, right? The lack of access to credit, but also the speed to delivery, if you will, right?

Do you guys think that…

the whole industry is broken and MPLs have a role to play kind of across the entire financial services space? Or do you think we’re now kind of tapping out in terms of innovation and things that marketplace lenders are going to take on?

Ed:
I am optimistic. And I am always someone that is excited about what’s to come and what we haven’t even thought of yet. And I talked earlier about ⁓ our vision is, do you get to that always on credit? Why does a consumer even have to? ⁓

Simon:
Mm-hmm.

Ed:
have to apply for a particular product even, right? So if it goes down to you have a problem. And so what is your problem, Simon? And it could be, you know what? I need a new car or my problem is I need to pay for a wedding or I have a significant home expense I have to do or a tax bill, something. I have a problem that I have a financial need.

And we will get to a point where Ed will stop going to say, I need a personal loan or I need a mortgage or I need an auto. It is, I have this problem. that then these solutions that, know, the, what I think is that, is the technology organizations and partnering with financial services that we’re going to win are going to be the ones that can figure out just making it so easy for the consumer of.

Hey, I know everything about you, Ed. And based upon everything I have available, here is the best option for you that’s going to give you the best rate. And that to me is like, we’re not there yet. So I mean, I still think there’s plenty of runway for ⁓ innovation, which is exciting, which is neat to be in this field today that we’re working in.

Simon:
Yeah, I think the only thing that we have today is just being able to meet customers at the point of need or the point of sale, ⁓ which, you know, is an improvement to the original kind of lending experience. But you’re right, we’re not very proactive just yet, even in, know, if we take again, just the stance of a marketplace lender ⁓ or a bank, we’re spending a lot of money on marketing.

Right, like we want to make sure that we get ⁓ the people that are in need for funds or that actually have a project that they need financing for. So I hear you, think yes, on the marketing and customer acquisition side, there’s definitely plenty of work to still be done. But for you guys at Upstart, so you have a bunch of different customers, right? We think of the users of Upstart when we say customers, but…

You work with plenty of sponsor banks. You have a bunch of investors, You’re a publicly traded company. When you think of that product ideation, right? And that product innovation, how do you kind of fit all of those pieces, right? Because very often it’s easy to see a consumer problem. Like you guys have enough staff and you have the right resources to be able to fix virtually any kind of problem in the financial services.

of an industry, but then you also have your banks that you have to answer to and your Ford Flow investors and your equity investors. How do you reconcile all of it in that product cycle?

Ed:
Yeah. So, one of the things that, and I think why I’ve, I’ve stayed, you I’ve been here now over six years. And one of the things that, has, has kept me here is that, ⁓ this is an organization that is, focused on the mission first and first and foremost. And it is, as we go back to like, how are we investing in the right products? ⁓ it is.

is making sure we’re looking at where we can enable better access to credit and provide the best APRs as possible. And sometimes that doesn’t always have the best returns. I’ll give you an example. One of the areas that we’ve really grown as a business, when I started, we were great.

in serving the needs in, I’ll say individuals that were on the higher risk end of the spectrum. But we were not very competitive at all in the prime space, in the super prime space. And it just wasn’t a focus. We didn’t have competitive rates. it wasn’t a, ⁓ just we weren’t a good fit. We weren’t aligned to serving that.

that consumer. we saw, particularly as we started developing into, ⁓ as we added secured products and other things like, hey, we want to have the best rates for all Americans, not just for some, for all. And so that really challenged us to come up with a prime program. And those prime programs, they don’t have the same returns as

a higher risk program, which you wouldn’t expect. And they don’t have the same margins, but it’s still the right thing in order to build a sustainable business so that you’re serving all Americans. Because then the investment is, as you’re adding on all these other products, you don’t want to forget a certain kind of segment. And that’s how we’ve tried to continue to make those investments is just looking at, hey, are we

are we able to serve all consumers? then we’re always challenging ourselves. Are we trying to always, are we providing the best rates across all the different risk segments?

Simon:
so you guys have very little product segmentation, right? As it pertains to the FICO spectrum or the credit spectrum, like whether you’re a customer that has a 380 FICO or a customer that has an 800 FICO, you’re kind of applying through the same channel, right? Like it’s the same, I’m sorry, page. It’s the same application process. It’s the same pricing process. Do you think that’s what made you guys so successful? Also just the fact that

You’re really not differentiating at least from a UX standpoint, from a user experience standpoint, between the guy that you have a lot of confidence in in their repayment abilities and the guy that, you know, kind of looks doubtful from a credit reporting standpoint.

Ed:
Yeah, so I think what has been something to our advantage is, there’s two sides. So if you think of us in the center as the technology platform, right? So we are the facilitator. We are bringing the solution to the table. And then you need sources of credit, and then you need the consumer, right? And you have consumers that are at different levels of the risk spectrum.

You have some that are very low risk and some that are very high risk. And so you’re trying to price them appropriately. And then you have different sources of capital. have, our organization, our capital comes from traditional banks and credit unions. thinking of those as like, they’re looking for getting new customers, getting new members, putting stuff on their balance sheet. As you would suspect, they’re playing more in the lower risk spectrum.

of those are the consumers that their capital is allocated. And then you bring the private capital. And those are folks that are really focused on just blended returns staying within a risk tolerance. And so they’re able to serve a much broader band of the consumer. And so I think our financing model of by bringing capital that can serve all consumers.

And then having a marketing and product strategy that enables access to all those borrowers, regardless of risk tolerance or risk of where they’re allocated in the risk prediction, then it creates then kind of a platform that truly is a complete consumer lending platform. And then as we continue to add on other products, it then serves different needs based upon what people have at the time.

Simon:
Yeah, it just makes you guys unique, right? I’m kind of thinking of a few players in this space and I can’t think of many platforms that will give you the same products, the same user experience, the same user interface, even though they may be able to line up the same things as you guys, right? The kind of the right investors, ready to take on the right credit boxes and where target returns kind of fall in line with what you guys can deliver on.

there’s very few platforms effectively that have done it. And I think there is even an argument to be made that you can’t be everything to everyone, right? Or you can be the thing to everyone. Sometimes you do need that little differentiation. And I do think it’s very interesting in your success story that you were able to kind of take the deep subprimes and the subprimes and the primes and the super primes and give them the exact…

same, ⁓ the exact same delivery, The exact same product execution. it’s a compliment, right? It’s not, it sounds like an easy fit, but it’s very much not, especially when you’re looking at individuals that are so opposite, right? As deep subprime and super prime, they just have different needs. it’s very interesting what you guys have built in that regard.

Ed:
Yeah, and it’s important that we, have to find the right partners too, right? You know, working with organizations like yourselves and other financial institutions that are aligned with that mission, right? And you don’t expect everyone to have the same risk tolerance as an organization. get, you know, there’s different factors that go into how each organization, financial services organizations kind of determines their

their risk appetite and what tolerance do they want to be in. But then if you provide that platform that allows them the flexibility to say, OK, here is where I want to play. Here is where I want to partner with you to be able to serve the consumer. Do I have a platform that I can do that? And I feel we offer that, and that’s what’s allowed us to grow.

Simon:
And it’s a good segue in that topic of how you guys work with banks. ⁓ You guys also have a very unique position as it pertains to the banks that you work with. You have multiple sponsor banks, so multiple banks that originate their own lending product on your platform. ⁓ But you also have this…

arm of the business where you almost resell the platform to other banks that want to use the App Store model and the App Store customer experience. So on that first point, the sponsor banks, ⁓ we always, especially lately for the past couple of years, we keep asking sponsor banks, hey, how do you choose a fintech that you’re going to work with? What do you look for? And what’s a red flag, et cetera?

Don’t ask fintechs that question very often because we always take the stance that fintechs will take whatever they’re given, right? They will work with whatever bank is able and allowed to take them on. You guys have the luxury of saying, I want to work with you. I don’t want to work with you. How do you decide who you’re going to work with?

Ed:
No, that’s a really good question. So a lot obviously is, for any company, ⁓ it depends what stage they’re at, right? When they’re new in their establishment, you don’t have as many options. And so you’re trying to find the right partner to start with and to kind of build the foundation. But then as you mature as an organization, what you’re trying to do is to

is to figure out ⁓ where can things be additive, right? And so additive meaning ⁓ is a certain partner bank going to enable us to serve more consumers. And whether it’s because of ⁓ geo structures, however they’re chartered, they’re unable to serve more geographically or

It’s that maybe the constraints of the program, they’re able to ⁓ expand more in kind of the risk tolerance or ⁓ maybe just like the characteristics of a particular asset type that they’re comfortable doing. And so when you’re looking for these partnerships, you’re looking for someone that’s going to be additive to your business. And then the other piece is also on the partnership.

Right is and you know this but and I think folks that are in financial services that are that listening to this now know that consumer lending is not a set it and forget it business. You’re constantly monitoring it. You’re reviewing it. You’re I mean there’s you’re understanding you know how the the vintages are performing how things are originating. What’s the consumer themselves? How’s the consumer performing? Is the consumer

Appetite changing. I mean, there’s there’s so much stuff that’s going in that you’re constantly reviewing and so you need to find a partner that That you know You’re gonna have access to and that you can actually work with which is so which is so important, right? Because I mean, you know the times that that you and I have been able to hop on the phone and say hey Let’s look at this and and hey, want to make this change. What do you think? What do you what do you guys think about this change or what do you need to see in order to get? Comfortable with it and and so

Having a partner that’s going to commit the resources and to be able to respond quickly in this fast moving environment is just as important as, I’ll say, the additive nature of what they would bring to the relationship.

Simon:
Right, and to your point, different banks have different risk tolerances. So do you look for banks that may complement each other, right? In the sense that ⁓ you have a bank that doesn’t want to do that kind of product and you kind of know this from experience because you’ve been talking to them for so many years, or are you going to look for a bank that’s going to be able to do that product and kind of fill that gap in your offering, or is it more than that, right? Is it more of a…

of a strategic alignment, looking for a bank maybe that has never had any sort of regulatory issues. What are those various factors that you think about when you’re selecting a bank?

Ed:
I think, mean, you’re looking at all those that you describe, right? I mean, because I think you, ⁓ as you’re building any of these partnerships and any type of ⁓ relationship, you’re investing a lot of time. So there’s a lot of time in due diligence and there’s a lot of time to actually to start these partnerships up. And so you want to make sure that

that you’re not starting these up for a three month stint. You’re starting these up hopefully for a ⁓ significant multi-year relationship so that you can grow together. And so you have to make sure that this is gonna be a partner that you can do that

Simon:
Yeah, we always make the joke that we’re essentially getting married, right? And the divorce is very painful and very expensive when it needs to happen. totally understood. But so you guys manage kind of that stack of banks, right? You have multiple banks that you work with. Again, just focusing on the sponsor banks here for a second. We all work differently. We all have different expectations. We all want to see different things from compliance. And we want you guys to manage risks in different ways.

How do you bridge that gap? How do you create an organization and processes so that you don’t have to have one compliance in for FinWise and one compliance in for that other bank and one compliance in for that other bank? How do you keep everything standardized and unified within the organization?

Ed:
No, that’s a really good question. so I think it starts, so you take the compliance piece, right? Is it starts with, are you actually building compliance as part of the product creation? And that’s one of the things that we have invested in as we’ve launched new products is that compliance has been part of the… ⁓

from kind of the day one. Like they’re part of the team. And so, you know, we feel that if we prioritize and have the A team on our risk and compliance organization and we have the best processes and that we are very transparent where we work with all the regulators very frequently of very comfortable getting in front of. ⁓

the various regulatory bodies and talking about our programs and letting them see and getting input on the things that we’re doing, we feel that then when we’re working with any of our partners, whether it’s the partner banks or the balance sheet lenders, we’re able to come to them with, hey, look, we have a strong…

risk management culture here and we have strong risk management processes. So then that helps us avoid all these extra nuances and you tend to I would expect you to have more issues if you had a weaker risk management process that is going to cause your partners to then have a lot of these extra asks because they’re not comfortable with you know with the framework that you’re bringing to the table, but but if we bring the

the kind of best in class kind of risk management program, that is gonna actually benefit us in a number of ways. But one of that is just the operational efficiency working with all our partners, right? Because then if we’re addressing it all, if we have a high bar, then all of our lending partners will say, okay, good, you’re meeting the expectation we have. I don’t have to add anything else to make this.

you know, to add more overhead to this program. So it is, we feel that we’re investing in that framework and in that function in the organization and that it has benefited us as we’ve scaled our organization.

Simon:
Yeah, but it almost sounds like an ideal, right? That everyone is going to be okay with everything that you provide, ⁓ that banks are not going to ask for more. But you and I know that in practice, that’s not what happens, right? ⁓ And know, FinWise is guilty of this. We ask for a lot of things. It’s ⁓ because we have our own experience with our regulators and we know what our region expects and maybe the Northeast expects ⁓ something else. So when…

Ed:
Sure. Yeah.

Simon:
you kind of get that ad hoc request from one of the banks that you’ve never seen before that none of the banks are requesting. How do you go about tackling it? Is there this process that happens within your team or outside of your team where you vet that request and debate whether you now include it as part of your multi-bank packet, like the stuff that you’re going to give to every bank going forward?

Ed:
Yeah, I would say that is kind of normal if you would expect out of a high performance team is that when an LP is asking, let’s say you, Simon, your organization has a certain request that says, is what we’re getting from our regulator. ⁓ One of the first things we’re going to try to do is just let’s understand what is the problem you’re trying to solve.

You and I both know someone could ask for something but it may not be actually what you need. So you may think it’s like, I need to have this report. Well, what are you trying to sell? Why? Because I’m trying to get this data in here and this, well if I provide it this way doesn’t that solve your problem? yeah, it will, right? So I think that the first thing always is when we get these requests is understanding what are you trying to solve?

And then, you know, we’ll assess internally of saying, okay, is this something that, is this unique to your organization or is this, and that could be too, is because of maybe the regulatory body that you’re under versus somebody else. And so, you know, there’s been cases where maybe we apply something that, okay, we know everybody with the same regulatory body, this is kind of something that they’re asking for right now.

let’s make sure at least they all have that because it seems like they’re all gonna get asked it, know, in some of that case. Another is if we find that, okay, this is something good for everybody. And then there’s also cases where you could ask something and it’s not that we tell you no, but it’d be more of, we’ll challenge, particularly if it’s, we have to understand what is it going to take for us to maintain it? And then, you know, often in those cases,

It’s, well, how do we find alternatives? Is there a compromise because we want you to still stay within your risk appetite and be able to meet the needs of your ⁓ regulatory organization that you’re working with and making sure that it’s balanced. there’s not, I think risk management is not necessarily always black and white.

black and white regs, but it is also just making sure you understand like how are we interpreting them and how are we, can we understand the problem and then provide what you need to solve it.

Simon:
Yeah, no, it’s definitely never black and white. But the reason why I’m kind of probing you on this is that, and also for listeners, ⁓ and I have plenty of experience kind of dealing with requests that very specific to a regulatory body, right? Or to a specific bank. And sometimes I understand that, you know, as a fintech that is working with multiple banks, you guys can even be doubtful of what you’re getting requested.

right, of what the banks that you’re working with are requesting or sometimes it could even be this weird kind of not pushback but something that you guys have to manage internally where we kind of go against your secrecy, right, if you will, because we all have to understand that at the core of marketplace standing companies is an underwriting, right, and that underwriting is unique to that

lending company and hopefully that’s what drives their value. ⁓ And you know, thinking about you guys specifically, and I’m sorry I’m rambling a little bit, but you guys use artificial intelligence, right? And you guys have been one of the first few fintechs to use artificial intelligence in underwriting. And I think when you guys started working with banks, ⁓ definitely AI wasn’t as big of a buzzword as it is today. ChatGPT was a really, really… ⁓

far from us, so AI wasn’t talked about as much. Now that it is, I’m assuming that the banks that you work with are asking you to be a little bit more open about your black box, right? And so how do you manage, how do you balance that need for secrecy? Because again, it drives the intrinsic value of your company, but also just meeting the expectations of the banks that you’re working with. And even more so, and I’m sorry, this is a very long worded question, but…

How do you deal with it when it’s only one or two banks out of the entire stack that are asking you to open kimono for them?

Ed:
Yeah, no. you know, model explainability is huge, right? That folks, and you’re right, at the different stages, very early on, when you would, you know, walk into a lending partner and you talk about AI and their, you know, their credit risk department would just like, you know, you just, could see them just short circuiting. Like, I don’t know how I’m going to manage this. I don’t know how I’m going to manage my regulator. Now,

over time, obviously, is more and more people are educated about models. then you also kind of, as you understand them, and you start thinking about, well, some of the traditional sources of ⁓ credit underwriting that I’m using today, don’t they have their own proprietary models that create scores that I don’t actually know how they’re creating the score? mean, there’s…

There’s things that actually, they understand machine learning and the use of data, you start seeing other examples that are like, yeah, I don’t see that model, but I’m very comfortable using that score in my underwriting decisioning. Now, what we have done to help, I’ll say manage and help educate folks is kind of what I said earlier is first and foremost is being very transparent with the regulators so that…

we’re understanding from them, all right, what are you asking of our partner banks? And so that then we can explain, hey, this is where our proprietary information is. We’re happy to share with you at a national level, you know, here is kind of how this works. So that when you’re talking to all these individual financial institutions, you don’t have to ask the same question 150 times. So, you you can answer it once for you.

And then the other thing we do is it’s third party model validation. So using external accredited sources that are reviewing the models on a regular basis that then produce reporting that is deemed acceptable by the regulatory bodies that can be used and that our LPs can review. And so even though they’re not looking at the code because it is proprietary, they can have the confidence

that yes, there’s controls on how the model is using, it’s how it’s being trained, just the overall maintenance. And there’s a control process in place so that I can be comfortable when I’m talking to a regulator that I am adhering to the kind of model governance aspects of the requirements.

Simon:
Yeah, I’m interested to see how this all evolves, right? We’re seeing the tone change a little bit with ⁓ regulatory bodies as it pertains to AI. It’s like, on one hand, they really want to use it, but then on the other hand, they want you to explain it in such a detailed way that it almost makes it impractical for you to adopt AI. But yeah, I know kind of throughout your history, you guys have engaged the federal government. I remember, you know,

a few years back when you guys had the first, and I think only ever no action letter from the CFPB, right, as a protein to the model. And that probably helped, you know, I’ll be honest, as a bank that works with you guys, we’ve never heard the FDIC tell us like, you guys are good, they talk to the federal office or they talk to the DC office, but I’m sure it did help in some ways.

Ed:
Yeah, it’s the transparency. mean, let’s face it. mean, this is something, if this was easy, everybody would be doing it. Right? I mean, and it is difficult because you’re changing how consumer lending is applied and you’re bringing new technologies and you’re bringing innovation. so there is a lot of…

Simon:
Right.

Ed:
With anything new there’s there’s people that are optimistic and say, wow, this is great And then there’s the pessimists that are like, this is going to be bad. This is this is going to damage the consumer and so and Each each side is allowed to have their their perspective and so it is around How can you have the conversations educate people talk about here’s how the controls here’s the documentation and and then you know it comes down to is is then the the the results right in in having

Data that you know, it’s it’s it’s different. I mean we’re an organization now that You know, we’ve originated over 53 billion dollars worth of loans and that’s you think of like unit wise it’s it’s it’s over five million units and our model You know our models are trained on it’s like over a hundred million repayment events, right? So so you you have this flywheel of data

And so now you can see the APRs, you can see the consumers that are helped, you can see the performance of the vintages and how are they comparing against ⁓ traditional methods. And when you start seeing, hey, wait a minute, you’re serving more people. Everybody wants that, right? Like who doesn’t want to be able to serve more people? And if you’re doing it in a way that credit performance is not deteriorating, right? Because

You don’t want to serve more people and then wow, the linguencies have gone up, losses have gone up, like you’ve created this financial crisis. No, you’re seeing that ⁓ use of models are actually becoming better at predicting default and risk is priced appropriately and people are giving access to the right amount of credit and they’re not at these elements. And so I think that type of data

is what is also getting people comfortable because they’re seeing the results. They’re seeing the impact that AI is having in the industry.

Simon:
I think on the last point that you’ve made, 2022 is a very trying year for the industry, right? I think what people sometimes fail to recognize, even though it’s something that I’ve talked about a few times on this podcast, is the fact that this is a vertical of fintech that was created in an upcycle of the economy, right? This was all kind of created after 08, 09.

like a few years after 08, 09 and really boomed when institutional investors came back to the table in like 2011, 2012. So all of those models, right? All of those underwriting models were trained during a very prosperous time of the US economy. 2022 happens, we thought that COVID was going to be the Blacks 25th, we’re all wrong. It definitely wasn’t, it was 2022 in the end.

You know, it’s interesting because I think that that’s really when the Androiding models that you have built or that anyone in the industry has built were being trialed by fire almost. Right. And so what did you guys learn in 2022? you take the opportunity to refit your models? Did you kind of go back to the drawing board and maybe identify to know that

You couldn’t be everything to everyone like we talked about a little bit earlier. Like I’m just really wondering kind of the lessons learned from this whole thing.

Ed:
Yeah, no. Yeah, the COVID period was so fascinating. And like I said, I’ve been in financial services for a long time. And you go back to like, 08 and even before that. you have those periods of you just, the cycles are just abnormal.

And, with, with COVID it, it was that, mean, you, nobody, nobody predicted COVID. Now people were, were, were always worried about planning for a bird flu or some type of like flu pandemic, but like no one predicted, yeah, this is COVID is going to hit this time. And that on top of that, no one then predicted that we’d have a stimulus that was, that was out there. And, and so that created just this, this weird

Simon:
Right.

Ed:
⁓ consumer fanata where the personal savings rate was at like all times highs and Delinquencies was at its all-time lows. And then what happened is then you had we all got back outside ⁓ stimulus dollars ⁓ stopped ⁓ People started spending again and then they didn’t stop they the consumer kept spending

Simon:
Mm-hmm.

Ed:
and kept spending this whole YOLO just, and what’d you see? You saw then vintages that were being originated in 21 and like early 22, but like kind of like that second, late second half 21, early 22, then as though they started maturing in 23, you started seeing performance really.

deteriorate and it’s because you had the inverse. had this shock to the system early on where people weren’t spending, they couldn’t spend anything, and they were getting money, so their savings were high. So they had this huge buffer. And then they got out and they were spending it. And it was fine for a while because the buffer still existed, but then that buffer disappeared, but the consumer didn’t stop. And so then we saw periods where

those vintages kind of underperformed expectations. And that for us is we use that time to where we can invent, like we’re always continuing to look this, like how do we invest? How do we improve the predictability? How do we improve the strength of the models? And during that time coming out of that is when we created the Upstart Macro Index. And it was really kind of understanding

not just the of the health of the consumer, but what were some of the other kind of macro risks that were going on? You we were seeing, you know, the change in interest rates, we were looking at change in inflation, all these factors that were actually weighing on a consumer so that even though Ed Walters, maybe my credit profile has not changed, like I still have the same kind of credit profile, I have the same

DTI, I’m making the same amount of money. However, now there’s factors around me that make it more expensive or there’s more risk in the environment. And so we really didn’t have a capability to kind of assess that and that incorporate that into the model. And that’s really where we came up with the Upstart Macro Index as a way and we publicize it on our website upstart.com slash UMI.

we publish what we’re seeing. And so if you think of like an index of one, and an index of one is like, that’s what you expect during a like benign period, that this is normal credit performance, it should be about a one, right? Well, you know, during those periods, when those vintages that I told you that were originated during periods where now ⁓ performance ⁓ worsened because of all those factors, you know, we saw UMI in the

And it’s like kind of adjusted for inflation and some other things, seasonality, like, UMI is in 1.5, 1.6. What does that mean? That means that the defaults that we’re seeing was at like 1.6 times the normal of what you’d see during a benign period. so then, so the first was identifying that. And then by having that visibility, the model then can incorporate that.

Simon:
That’s

Ed:
is part of the risk. And then further having those kind of risk adjusted APRs. So again, like what I told you is that, hey, maybe Ed Walters hasn’t changed, but because of the environment factors that are going around and what we’re seeing that how consumers are kind of stressed in the environment, ⁓ we need to adjust the risk appropriately. And that has really allowed us to then,

as we’ve had subsequent, kind of nothing like that we’ve seen in 21 and kind of those 22, as we’ve seen some stress and we talked about it in a prior earnings call that as we see stress, well, that is going to impact then how the model approves consumers and that’ll have a direct impact on our volume.

And at the end of the day, I think you and I have talked about like what’s the What’s the most important thing as we look at is at the company? Is it is it the volume? Is it revenue? First and foremost for us, it’s credit performance. I mean that You don’t have we’re a trust business, right? You don’t have a business if you can’t give the confidence that you have the right tools and technologies over the long time It’s it doesn’t mean we’re going to be perfect because you because nobody can

predict the future. But if you have the right tools that over the lifetime of relationship that you can accurately predict and build a performance within the risk tolerance is to get the returns that you expect, that’s the trust that that’s what keeps capital. That’s what signs up new partners. That what is enable us to launch new products. That’s what enable us to serve more consumers. so we’ve invested a lot in the capital kind of the

performance ⁓ space to just make sure that as these things are happening, ⁓ the model is able to adjust. it’s a tool that we never waste a ⁓ tough situation, right? So we took advantage of those kind of those challenging vintages and saying like, okay, how do we learn? What tools can we build? So the next time this happens, we’re that much better.

Simon:
It’s rebuilding your underrunning to be little bit more…

a little bit more, it’s not even robust, but a little bit less sensitive to the economic downturns, right? And just make sure that at least you factor it in when ⁓ you kind of issue your underwriting decisions, as opposed to, you know, the model thinking that we’re still operating under the same macroeconomic conditions that we’ve been in when all of a sudden we’re not. it’s, yeah, it’s nice to see that the industry reacted.

so quickly to it, but also try to learn as much as possible from this situation. think, do you, and feel free to tell me that you don’t want to answer this, but how do you rebuild the trust with ⁓ forward flow investors after that kind of event? Because the narrative you have is the one that I expect to knowing the industry the way I know it. But a forward flow investor that may just be looking at numbers on a spreadsheet,

may not think of it the same way. So how do you rebuild that trust with them?

Ed:
Yeah, it’s a great question. And so I think you do it a couple of ways. mean, one is, first and foremost, during that period, everybody experienced performance challenges, right? And so that is one. So it’s less about the comfort in the company versus it becomes then comfort in the asset class itself, right?

Simon:
Okay.

Ed:
So, mean, because that’s where want to differentiate is like, is even traditional models, like every, everybody saw, saw stress during that period. And you can look at, know, DVO on data and stuff that you can see. It wasn’t, it wasn’t isolated to any particular consumer lending platform. Everybody saw stress. And, and so you’re right. You, during that period, you, have individuals ⁓ that both on the private credit as well as is on the

the banks and credit unions that are like, hey, I don’t feel comfortable in that asset class because of the variability in it. And so how you rebuild trust is, ⁓ for those that just aren’t comfortable with the asset class anymore, is you continue to show ⁓ data and be able to show the performance. You show the investments that you’ve made. over time,

Simon:
Right.

Ed:
we’ve been able to bring those people back to the table. The other piece that you look at is ⁓ you look at different asset classes or different risk profiles. Like, for example, I talked earlier about the T prime program, just the lowest risk consumers. Well, someone that maybe was okay with personal loans but didn’t like the expanded risk, well, they could get more comfortable saying, but you know what?

I like the prime borrower. okay, now you’ve made a change that I could actually get back in, but in that space. Right? And so it’s kind of understanding what people need. And then as I talked about the other products, now as we’ve added secured products and different things, it’s, hey, I still have the capital. Maybe I’m not as comfortable with the unsecured space, but I really like the secured space. I really like auto and I really like Heloc. And ultimately,

where our partnerships will go is this multi-product solution. My utopia is the best partnerships I have is you have the CLO engaged, you have the CFO, you have the chief risk officer, so you have the leadership team that’s engaged. And they are using all the products and then really becomes, how do you want, what are your business needs?

Simon:
Mm-hmm.

Ed:
how do you want to then, if you have everything kind of enabled, now you can just enable ⁓ diversification based upon what you’re doing organically, how you want to manage your balance sheet, what your runoff is on your other assets, so that you have the flexibility to say, okay, this month I want more PL. Next month, for the next two quarters, I really want to grow in auto. Now I’m seeing as, you know what, I’m not hitting my mark in HELOC. And so that’s where that…

future state of where we’re investing a lot of time with our partner base is exploring, getting more and more of them on a full multi-product solution because I think that’s where we can have the biggest impact and it provides them the most flexibility.

Simon:
then it’s all about diversification, right? Just having investors interested in different things and making sure that you can kind of place all trenches or all segments of your product with the right capital. So, that makes a lot of sense. And we’re kind of almost at time, right? Anything else you want to share with our listeners, any words of wisdom, any trends that you’ve noticed recently in the industry or at Upstart specifically.

Ed:
Yeah, so I would say is I’ll make a plug because we’re an AI company is everyone is inundated now with AI, all these different AI tools. But I encourage everyone that ⁓ take the time to get educated and figuring out what all these different tools or what all these different companies are doing with AI.

because the, you know, I’ve often heard people say, well, AI is going to take my job. And it’s like, no, AI ⁓ is not necessarily going to take your job. ⁓ Who’s going to take your job is your coworker that’s using AI to be more productive that then makes you look unproductive. That’s what’s going to impact. And so, so I encourage everyone that, you know, how are you investing your own personal time to just understand

Simon:
Right.

Ed:
the different tools, how can AI be a part of your business? Because it’s not something that is down the road. It’s here now. And some people are doing some really cool things with it on top of what we’re doing in consumer lending. it’s across multiple verticals right now, which is pretty cool to see it be leveraged and be leveraged in the right way.

Simon:
I think that’s a great piece of advice. Thank you very much Ed. It was really nice chatting with you and thanks for all the wisdom that you’ve shared today.

Ed:
It’s always a pleasure, Simon. Thank you so much.