The Exchanges

Q And how did you, uh, get the flywheel started on that side initially? Who were the first few, uh, that said the, that they were happy to partner?

A Um, so I think American Express was the first, uh, bank, if you can call them a bank, um, that we launched, uh, I think we kind of went top down from largest to smallest. Um, the, the deeper partnerships though, you know, Chase is the one that we've been working on the longest, and if you look at our history in the press, you can see that we had good times with Chase, bad times with Chase, now very good times with Chase. Um, but we really did go top down. We spend most of our time and attention focused on the largest banks, um, and the technical integrations we'll do with the smaller banks as well. Um, but our business development team, which is a team unique to, to Plaid that only talks to banks, um, that, uh, that, uh, that, that is entirely focused on the largest banks.

AI assessment note: “American Express was the first, uh, bank, if you can call them a bank”

Q To now being a routing layer. Do you want to talk about this?

A Absolutely. So, Uh, historically people used to think about banks as, as, as the vault. Uh, you walk into a bank branch, you bring some money, you give it to your banker, your banker puts up the vault. You need some money later, great, you ask your banker for it, and they give it back to you. Um, maybe you ask them for a loan, anything else, but, but the, the bank is the vault, and your primary bank, um, we, we kind of call them big box banks, uh, where they're truly everything to everyone. Um, they will give you a, a checking account when you're, you're 12 years old and have 25 dollars, Uh, they'll give you a mortgage when you need to buy a house, and they'll help you plan for retirement. Um, the, the, the nuance for us, though, became when we started to look at the actual inflows and outflows into those accounts, um, if you looked at a lot of, uh, especially millennials, um, if you looked at their bank statements, it's just a list of, of I.O. So, uh, it's a list of, uh, money coming in from a checking account, getting transferred out to a friend via Venmo, paying their bills, paying their rent, so on and so forth. Um, and so then, then we had this mindset shift where we said, The bank account is actually the hub for your financial life. It's not the vault. And, uh, like any good hub, you would want it to be highly interconnected. Um, and so, kind of going into that thesis, uh…

AI assessment note: “The bank account is actually the hub for your financial life. It's not the vault.”

Q Has it helped with the other constituency? Have you seen some of the, some of the banks you were trying to convince? There was like the loan developer that happened to, uh, come across The tool and say, hey, we should, our bank should be enabled, um, by, by that or is it happening now?

A We have a lot of that now. It didn't happen a lot at first. A lot of us was us going to the bank and saying, hey, we're going to, we're going to build this thing. We want to build this thing. Here's how it's going to work. Um, there's actually, it's interesting in the early days, there's a provision of Dodd-Frank that says that consumers may have, must have access to a digital copy of their financial data. And we operated under this principle where consumers, Signed us that, and, and we then went and collected the data from the bank. So the banks couldn't say no per se, but they certainly didn't have to make it easy. Um, and so now, uh, we have very good relationship with the banks. They're, they're our customers in many senses. Um, actually most of the biggest banks are, are our customers. Um, and that, that adds a lot of value back to the ecosystem, but it's always been tenuous to, to work with the financial institutions. Um, in one sense they need to enable their customer to do something. Uh, so they need to enable their customer to pay a friend or apply for a mortgage or, or whatever it is. On the flip side though, they also want to be the one giving the customer, uh, that mortgage or, or, or that retirement account. Um, and so there, there's a lot of push and pull. Uh, those relationships are delicate.

AI assessment note: “We have a lot of that now. It didn't happen a lot at first.”

Q And how did you, uh, get the flywheel started on that side initially? Who were the first few, uh, that said the, that they were happy to partner?

A Um, so I think American Express was the first, uh, bank, if you can call them a bank, um, that we launched, uh, I think we kind of went top down from largest to smallest. Um, the, the deeper partnerships though, you know, Chase is the one that we've been working on the longest, and if you look at our history in the press, you can see that we had good times with Chase, bad times with Chase, now very good times with Chase. Um, but we really did go top down. We spend most of our time and attention focused on the largest banks, um, and the technical integrations we'll do with the smaller banks as well. Um, but our business development team, which is a team unique to, to Plaid that only talks to banks, um, that, uh, that, uh, that, that is entirely focused on the largest banks.

AI assessment note: “American Express was the first, uh, bank, if you can call them a bank”

Q uh, topic from me and then we'll open up, uh, to, People, maybe some thoughts on the future of FinTech. Like you had some interesting predictions. So where are we in that, in that arc, right? From the vault we were talking about to the sort of the writing layer. What, what happens next? I mean, do we end up with this financial system that's a hundred percent digital eventually?

A Yeah, and for consumers, absolutely. As a consumer, you actually can live a fully digital financial life right now. You might not get access to certain types of products, but you'll be able to do most things. If your financial life is relatively simple, you can do it now. Um, I think, so some of the interesting trends that we're seeing, fintech obviously growing massively. Um, the banks themselves, uh, they're now actually launching digital persons of themselves. Uh, so Chase has done this, Capital One has done this, a number of other banks have started to launch these fully digital banks where you never have to walk into a bank branch. Really exciting. Um, we're seeing more and more of that. Uh, we're now seeing the banks and the fintechs starting to both Uh, partner, uh, meaning a FinTech will have a product and a bank will distribute that product to its, its customer base, and also compete, uh, meaning the, the banks themselves are cloning FinTech products. Um, ultimately, this is amazing, uh, because the consumer wins, right? With more optionality, with more choices, the prices come down and the consumer gets more choice. Um, and that really goes back to the thesis behind Plaid of making money easier for everyone. Um, so, so we've seen a lot of that. We're actually seeing this interesting trend, um, historically, uh, bundling in banks was geographic. Uh, so you went into a …

AI assessment note: “Yeah, and for consumers, absolutely. As a consumer, you actually can live a fully digital”

Q So what have, uh, been some of the key data challenges? So is that, is that purely like a scaling, sort of engineering, uh, making sure the whole thing works, or was, uh, Like the, the dirty data aspect of this, was that, uh, was that an issue? Was that something you solved reasonably early?

A Yeah, so our, the earliest hooks that we had were just around categorization and, and cleansing. So could we identify a merchant well? Could we identify what was a check being deposited? Uh, could we understand what, what someone's income level was based on the transaction data that was going on in their account? Um, all of that stuff, it's, it's pretty basic data science challenges. Um, but it's important to do correctly, and it's important to do correctly at scale. So, um, that was one of the big hooks for us early on. Um, then, yeah, it shifted more into the mode of scaling. So, um, kind of integrating with all of the banks that we have to integrate with, uh, 15,000 banks having all sorts of different integration methodologies. Um, so that means that we built a lot of really kind of smart infrastructure to, to look at these different banks and figure out how to do it. Uh, data processing on our side was pretty intense, uh, with all the data coming through. Um, crazy challenges around, like, batch updates. So how do you schedule batch updates correctly to get the most fresh transaction data from the bank without taking their servers down because you have too much traffic? Um, that is an actual challenge that we run into. Um, so, uh, a lot of it does become around, kind of, like, these, these infrastructural challenges. Um, and, and now we're shifting back into a mode where, k…

AI assessment note: “the earliest hooks that we had were just around categorization and, and cleansing.”

Q To now being a routing layer. Do you want to talk about this?

A Absolutely. So, Uh, historically people used to think about banks as, as, as the vault. Uh, you walk into a bank branch, you bring some money, you give it to your banker, your banker puts up the vault. You need some money later, great, you ask your banker for it, and they give it back to you. Um, maybe you ask them for a loan, anything else, but, but the, the bank is the vault, and your primary bank, um, we, we kind of call them big box banks, uh, where they're truly everything to everyone. Um, they will give you a, a checking account when you're, you're 12 years old and have 25 dollars, Uh, they'll give you a mortgage when you need to buy a house, and they'll help you plan for retirement. Um, the, the, the nuance for us, though, became when we started to look at the actual inflows and outflows into those accounts, um, if you looked at a lot of, uh, especially millennials, um, if you looked at their bank statements, it's just a list of, of I.O. So, uh, it's a list of, uh, money coming in from a checking account, getting transferred out to a friend via Venmo, paying their bills, paying their rent, so on and so forth. Um, and so then, then we had this mindset shift where we said, The bank account is actually the hub for your financial life. It's not the vault. And, uh, like any good hub, you would want it to be highly interconnected. Um, and so, kind of going into that thesis, uh…

AI assessment note: “The bank account is actually the hub for your financial life. It's not the vault.”

Q And then on top of that transaction layer, what, what else do you do? You do identity, you do, what else?

A So we do, uh, kind of a bunch of straightforward products, so allowing you to monitor transactions, uh, uh, kind of set up, uh, ACH payments. Um, uh, we also have a, a set of other products around like checking a balance in real time. If you're about to do a payment, you might as well check that there are funds available to do that payment. Um, we have identity validation products, um, we're working on, uh, a set of more specific lending products. So I talked a little bit about pulling in, uh, transaction data into, into a loan. Um, our longer term thesis is that you should be lending to a consumer in the way that you lend to a business, which is based on free cash flow, um, and so we can kind of pull that in and build a model around understanding free cash flow, so there, there's a lot of stuff there, um, and we have an, uh, an assets product for lending as well. Um, and then, uh, we're continuing to do more analytics on top of the data, so it's, it's an immense pile of data that we have. Um, of course, we're doing things only that, that benefit the consumer, so, uh, you could, you could imagine us building things around better risk and fraud, Um, kind of simpler onboarding for a lot of these flows that, that we're seeing. Um, it's a lot of data, but we need to be sure that the products that we're building are in the consumer's best interest.

AI assessment note: “allowing you to monitor transactions, uh, uh, kind of set up, uh, ACH payments.”

Q Has it helped with the other constituency? Have you seen some of the, some of the banks you were trying to convince? There was like the loan developer that happened to, uh, come across The tool and say, hey, we should, our bank should be enabled, um, by, by that or is it happening now?

A We have a lot of that now. It didn't happen a lot at first. A lot of us was us going to the bank and saying, hey, we're going to, we're going to build this thing. We want to build this thing. Here's how it's going to work. Um, there's actually, it's interesting in the early days, there's a provision of Dodd-Frank that says that consumers may have, must have access to a digital copy of their financial data. And we operated under this principle where consumers, Signed us that, and, and we then went and collected the data from the bank. So the banks couldn't say no per se, but they certainly didn't have to make it easy. Um, and so now, uh, we have very good relationship with the banks. They're, they're our customers in many senses. Um, actually most of the biggest banks are, are our customers. Um, and that, that adds a lot of value back to the ecosystem, but it's always been tenuous to, to work with the financial institutions. Um, in one sense they need to enable their customer to do something. Uh, so they need to enable their customer to pay a friend or apply for a mortgage or, or whatever it is. On the flip side though, they also want to be the one giving the customer, uh, that mortgage or, or, or that retirement account. Um, and so there, there's a lot of push and pull. Uh, those relationships are delicate.

AI assessment note: “We have a lot of that now. It didn't happen a lot at first.”

Q And then on top of that transaction layer, what, what else do you do? You do identity, you do, what else?

A So we do, uh, kind of a bunch of straightforward products, so allowing you to monitor transactions, uh, uh, kind of set up, uh, ACH payments. Um, uh, we also have a, a set of other products around like checking a balance in real time. If you're about to do a payment, you might as well check that there are funds available to do that payment. Um, we have identity validation products, um, we're working on, uh, a set of more specific lending products. So I talked a little bit about pulling in, uh, transaction data into, into a loan. Um, our longer term thesis is that you should be lending to a consumer in the way that you lend to a business, which is based on free cash flow, um, and so we can kind of pull that in and build a model around understanding free cash flow, so there, there's a lot of stuff there, um, and we have an, uh, an assets product for lending as well. Um, and then, uh, we're continuing to do more analytics on top of the data, so it's, it's an immense pile of data that we have. Um, of course, we're doing things only that, that benefit the consumer, so, uh, you could, you could imagine us building things around better risk and fraud, Um, kind of simpler onboarding for a lot of these flows that, that we're seeing. Um, it's a lot of data, but we need to be sure that the products that we're building are in the consumer's best interest.

AI assessment note: “we do, uh, kind of a bunch of straightforward products, so allowing you to monitor transactions”

Q uh, topic from me and then we'll open up, uh, to, People, maybe some thoughts on the future of FinTech. Like you had some interesting predictions. So where are we in that, in that arc, right? From the vault we were talking about to the sort of the writing layer. What, what happens next? I mean, do we end up with this financial system that's a hundred percent digital eventually?

A Yeah, and for consumers, absolutely. As a consumer, you actually can live a fully digital financial life right now. You might not get access to certain types of products, but you'll be able to do most things. If your financial life is relatively simple, you can do it now. Um, I think, so some of the interesting trends that we're seeing, fintech obviously growing massively. Um, the banks themselves, uh, they're now actually launching digital persons of themselves. Uh, so Chase has done this, Capital One has done this, a number of other banks have started to launch these fully digital banks where you never have to walk into a bank branch. Really exciting. Um, we're seeing more and more of that. Uh, we're now seeing the banks and the fintechs starting to both Uh, partner, uh, meaning a FinTech will have a product and a bank will distribute that product to its, its customer base, and also compete, uh, meaning the, the banks themselves are cloning FinTech products. Um, ultimately, this is amazing, uh, because the consumer wins, right? With more optionality, with more choices, the prices come down and the consumer gets more choice. Um, and that really goes back to the thesis behind Plaid of making money easier for everyone. Um, so, so we've seen a lot of that. We're actually seeing this interesting trend, um, historically, uh, bundling in banks was geographic. Uh, so you went into a …

AI assessment note: “Yeah, and for consumers, absolutely. As a consumer, you actually can live a fully digital”

Q the specifics of the product. So I think you, you guys started as a, um, uh, transaction, there was a transaction layer was the first one where you would, Connect with a bunch of bank accounts, extract data, and then enrich it with, uh, categorization, localization, the merchant, the category, you know, the categories, that type of thing. So that's, that's still the core, and then what did you add?

A So, that is at the core of what we do. Actually, the name, the name Plaid came from an early algorithm. Um, one of the hardest challenges we faced early on was we get this, this stream of data for each user. You link an account, we pull in your, effectively, your bank statement or your credit card statement. Um, and then it says all of these really weird merchant names. So, you know, SBX four three. What does that mean? Um, it turns out that means Starbucks in, in Union Square. Um, but we didn't know that. Um, so we pulled in this, this stream of transaction data, uh, started putting it on a map. We, we created this kind of canonical merchant database for all the merchants in the US and, and now more broadly. Um, and then we just tried to do this matching. Um, and so we started building more and more complex, uh, merchant matching, uh, algorithms. Uh, one of them was this cross user comparison where we compare your pattern to another set of, of person's patterns, uh, or to another person's patterns. Um, and we ended up with this cross work pattern. And we said, oh, Plaid. All right. We'll add that to the list of names. Uh, and then later we went through and found out where we could buy the dot com and, and so on and so forth. Um, so that's how we got the name. But, uh, yeah, so at, at the core we're a data plumbing company. Uh, kind of moving data from, from the place that it i…

AI assessment note: “that is at the core of what we do. Actually, the name”

Q the specifics of the product. So I think you, you guys started as a, um, uh, transaction, there was a transaction layer was the first one where you would, Connect with a bunch of bank accounts, extract data, and then enrich it with, uh, categorization, localization, the merchant, the category, you know, the categories, that type of thing. So that's, that's still the core, and then what did you add?

A So, that is at the core of what we do. Actually, the name, the name Plaid came from an early algorithm. Um, one of the hardest challenges we faced early on was we get this, this stream of data for each user. You link an account, we pull in your, effectively, your bank statement or your credit card statement. Um, and then it says all of these really weird merchant names. So, you know, SBX four three. What does that mean? Um, it turns out that means Starbucks in, in Union Square. Um, but we didn't know that. Um, so we pulled in this, this stream of transaction data, uh, started putting it on a map. We, we created this kind of canonical merchant database for all the merchants in the US and, and now more broadly. Um, and then we just tried to do this matching. Um, and so we started building more and more complex, uh, merchant matching, uh, algorithms. Uh, one of them was this cross user comparison where we compare your pattern to another set of, of person's patterns, uh, or to another person's patterns. Um, and we ended up with this cross work pattern. And we said, oh, Plaid. All right. We'll add that to the list of names. Uh, and then later we went through and found out where we could buy the dot com and, and so on and so forth. Um, so that's how we got the name. But, uh, yeah, so at, at the core we're a data plumbing company. Uh, kind of moving data from, from the place that it i…

AI assessment note: “Actually, the name, the name Plaid came from an early algorithm.”