Should You Let an AI Agent Spend Your Money? (Max Levchin)

Great to be here again.

You know, I think I'm done fighting the BNPL moniker. For a while I was raging against, you know, don't put me in a box, and I'm over that. What's important to understand is BNPL, or at least Affirm within that category, is fundamentally a mind shift, or a vibe shift, as the kids say these days. You can borrow money, it's a good use of personal capital to stretch your money out into the future, and there's a right way to do it and a wrong way to do it, and we represent, certainly to our biased opinion, the right way to do it. We don't charge late fees, we don't do deferred interest, all the sort of things that are kind of scary because they really are. We don't do. We are super transparent, we do everything simple interest, we pre-communicate, pre-price everything, we reward you for paying early, all the things that Affirm stands for. So, if that's called BNPL, so be it. We're still the only company in the industry that does all these things in a very specific, very deliberate way that is so profoundly pro-consumer. And in that sense, we are in a box of our own, and I'm quite happy being there.

Yeah, so as you guys know, every transaction that is powered by Affirm is underwritten individually in the moment of transaction. So, we take all the data that's available about you as a consumer, as a shopper, as a borrower, and the thing that you're buying, drop it all into a giant AI model, and it tells us, "Here's the price of risk. Here is the cost of lending you this money based on everything we know about you." So, if 6 months ago you came in, we never met you before, we can infer a little bit about you based on your credit report data, which we pull, based on kind of what we think we know about what the kind of thing you're buying. There's some very interesting fraud-related data. If you're buying something that can be easily fenced in a street corner, there's a little bit of an increase in risk. If you're buying something that is essentially a capital asset, that's a lower-risk buying a mattress. We're easily convinced that's probably something you're going to sleep on as opposed to try to sell it to the highest bidder, et cetera. So, the models take all the data in and give us a price, and we show it to you and say, "Hey, this is the cost of credit." And a lot of times, the retailer, or sometimes the manufacturer, will pay your cost of credit, will pay your interest for you, so it's a 0% loan. But in either case, there's some associated cost that the model tries to predict, and we're very, very good at that. We've been at it for 15 years. Models are very precise. The next time you come around, assuming you have paid your bills from Affirm on time or early, which we love, we'll say, "Hey, you know what? We know more about you now. It's not just the publicly available credit data. It's not just what you're buying. It's also the fact that we've seen you in action. The cost of risk will go down, and we'll pass it right on to you, and your next transaction will be cheaper." And it goes on and on. Models are built to take that into account every single time, and that's why it's so powerful to underwrite every one of these transactions in real time.

Just ends up being lower interest.

It's a great question, and you know, the future is super interesting and exciting, and it's fun to try to participate in both predicting and building it. This is already happening. This isn't quite figment of someone's imagination. So, if you go into Gemini today and you type in, "Hey, I'm going to Europe. Help me figure out how to get there. Here's my budget. I want to be able to fly a family of four, this city, this city, etc." Gemini will do a pretty good job saying, "All right, I can see some information about airlines. I can help you shop for seats, etc." So, that's already happening. The next thing that's going to start happening more and more, and it's already beginning, and we are an excited part of this, is you can tell Gemini or your favorite bot or your favorite LLM, "Hey, I actually want to close the transaction now. I like the price. I like what I'm seeing. How can I afford this?" And it has been our job for 15 years to answer the question, "How can I afford this?" in the most transparent, sensible, lowest cost possible way. And so, what we've built are these plugins that go into LLMs and pipe up right as you ask this question, "Hey, how do I pay for this thing?" And obviously, we're not the only ones. There'll be plenty of choices, but the thing we're so excited about is think in the past, the thing that I love to rage against, this idea of, you know, zero with a tiny asterisk, don't read the fine print, you're getting screwed, you don't know how, doesn't matter. Like, that's the business model of old. The business model of Affirm is you are not going to get screwed. We take it as our responsibility to you to give you the exact price on super transparent language. One of our core values is no fine print. It's always been hard to compete against those who have a somewhat less scrupulous approach because the fine print is really fine. You can't read it and I can rage against the fine print all I like, but the consumer who's got two kids to travel to Europe with is not actually going to have the time to be like, "Wait a second. I got to read the fine print." So, they still get trapped. They still get into these awful transactions that later on they come to regret, but too late. What's amazing about chatbots, amazing about AI is each one of these bots has a PhD in consumer finance. They are looking out for you every single moment. And so, as you say, "Hey, how can I afford this thing?" Affirm will pipe up with, "Hey, here is a pay in six plan. Here's the exact cost of interest. If you prepaid, the interest is going to come out. Or, here's a trip by an airline that we have a partnership with. They will pay your interest for you, so it's a true no interest loan. Doesn't matter if you're a couple days late. Doesn't matter if you forgot a payment. We're not going to flip on you and charge you a bunch of interest you didn't expect like some of the other characters that I enjoy raging against." And so, that combination of extreme consumer convenience, transparency, and AI being the world's best watchdog of these fine print gotchas, we think is just going to not only do great things for our business, but fundamentally change the bar, raise the bar on financial products. Every single person asking their AI to help them buy something, afford something, plan a financial decision is going to get a better outcome because these things are never going to be fooled. They're capable of doing advanced math right there in front of you and boiling it down for you in a single line. This one has no fees is a pretty good thing. Cool. They're not going to make a decision for you, but they're going to make your decision the most informed decision possible.

We are embracing competition wholeheartedly. So, I think there are two fundamental things that make us different, maybe three. I'll start with two and then I'll add one more. Number one, much to my chagrin, by the way. Look, this advantage I'll give away. Anyone who wants it is welcome to it. So, stop charging late fees, stop doing compounding interest, stop having transaction fees, all the fees that the industry loves to throw in. We don't do any of that. And for years, I feel like we just never quite got our fair share of love and recognition for it. Those who use us know exactly what we stand for. In fact, our best customers are the ones that have been late by a day or two. I used to love listening in on phone calls like, "Hey, I'm late by a day or two. Can I just be current and you'll be okay." Like, "What's the late fee?" "There is no late fee." Like, "Oh my god, that is so cool." I've heard "that is so cool" so many times. It's the happiest app. We are still largely the only player that does this at any kind of scale. And so, right now, as these AIs compare all these providers to each other, we tend to stand out in a good way of like, "Hey, this is the one with no fees." So, I will gladly share that podium with anyone who wants to be on it, but for now, we are unique. Maybe the more permanent or at least harder to dismantle advantage, we've been at this idea for 15 years, which also means we've gone to literally hundreds of thousands of retailers and struck special pricing contracts with them where they will pay your interest either fully or partially, sometimes during a promotional period, sometimes during every day of the year, but the same level of transparency, the same level of no fees, no gotchas, zero means zero, all the stuff that we just bring from ourselves, we pass on or invite retailers and brands and manufacturers to participate in. So, when you go into a chatbot and you're buying a Samsung TV on a 0% transaction, you know it's zero because it's Affirm making a promise, but it's also unique. You can try to buy it with any other payment provider BNPL or otherwise, you're going to find yourself paying interest. You're going to find yourself also wondering if there's going to be any fees, but even without it you're going to find it a little bit more expensive. So, quite happy to compete on that facet of price because we do have these contracts and they take a long time to negotiate. They take a long time to execute very carefully because retailers don't have infinite margins. So, each time they subsidize the service 0% loan, they have to believe it's a consumer who's only going to buy when they are seeing this particular promotion powered by us. And we have incredible amount of experience doing it, but also huge systems of software that we've built over the years to provide visibility and transparency to the retailers to make sure that they can run these ad campaigns responsibly, thoughtfully, and get the most for their money here. So, that's a huge, very, very sophisticated advantage that we bring to bear that we've used very successfully over the years outside of AI. We think it's going to be multiplicative within AI. And then just one last bit since we're the two and a half, the other thing we've done for the last 15 years is we've gotten better and better at underwriting. And so, at any given moment when we are side by side with any one of our competitors, especially for things like 6-month transactions or 12-month transactions or something that's actually quite long-term and has real risk to it, you have, you know, as people in credit say, 12 opportunities to default, 18 opportunities to default kind of. It's a bad joke, but it's not completely wrong. We're very good at sizing that risk and pricing it correctly. And so, typically we approve more people while having lower losses. And so, that again creates a moat or an opportunity for us to outperform our competitors. So, we love competing. I'm generally a sort of competitive person, but feel very good competing in a space where the judge has a PhD in consumer finance.

The answer is actually kind of both. Maybe yes is the answer. We are better at saying no in a sense that we have the right or the opportunity to be the consultant or the, you know, sometimes people ask me like, what is Affirm? Like, what are you to me as a person? And I'm not really a brand marketer, so I always struggle with these like, what is this? But someone I know who is a brilliant brand marketer said, you're like the smart older sister. Like, you're there like, hey, that's a bad idea. Put down that wallet. Like, don't buy that thing. You should not indulge. That's the Or, hey, that's cool. You deserve it. Go get it. And so, the ability to say no or yes in the right moment, the right way as a sort of a friend or a coach or somebody who's actually out to look for your best financial outcome is something that we're very, very good at. And so, because that is a fairly unique value proposition, we appeal to a particular type of a consumer. Like, the person who wants that participatory engagement from their credit provider is someone who's actually quite motivated not to be behind. Our consumer knows we're going to say no when we feel you really shouldn't be buying this. We're also very frequently going to say things like, hey, a $600 bicycle is a little bit more than you can afford right now. However, we're quite happy to lend you two-thirds of this, but you do have to put one-third down as a down payment because that's what we think is the right sort of contours of a loan for you for this purpose right now. And so, that consumer is doing fine, but it's also because they want to do fine. They're motivated to improve their financial reality. Like, maybe the most shocking thing about Affirm is that we are living proof that in the last 12 months, 27-ish million such people exist. You would think that the headlines of various newspapers is people don't care about their financial reality. They're sort of shopping with their eyes closed. They're not reading their financial statements. And I'm here to tell you that's not true. We have tens of millions of people who are coming back to us now almost seven times a year saying, hey, make sure the thing I'm buying isn't going to bite me. I'm not going to regret it. I'm not going to be figuring out how the hell did I get myself into this kind of debt. And so, that customer exists. They shop with us. They're doing fine. I think some of the less scrupulous providers are happy to lend money because they'll make it back in things like late fees and delinquency charges and we don't do that and turns out there's a lot of people in America who actually quite prefer that sort of way of living.