Fraud against small businesses jumped a staggering 70% since the COVID-19 pandemic, and according to Will Tumulty, CEO of Rapid Finance, the shift to online lending has made it easier than ever for fraudsters to strike. On The Credit Union Connection, Tumulty sat down with Sarah Snell Cooke to discuss how this alarming trend is affecting small businesses and why credit unions must step up their game to protect both their members and their operations.
Tumulty breaks down how AI is proving to be a game-changer in the fight against fraud. Imagine AI not only reading documents in the blink of an eye but also sniffing out patterns and detecting shady applications that could easily slip past a human eye. It’s an exciting vision, but Tumulty doesn’t sugarcoat it: AI is only as good as the data you feed it. And here’s where credit unions often hit a roadblock: their data is often scattered and messy. Tumulty points out that unless institutions clean up their act, AI won’t be able to do its magic.
But don’t get too bogged down in the data weeds. Tumulty also highlights the potential AI has for revolutionizing small business lending. While AI can process mountains of data quickly, there’s still a big need for human oversight. He stresses that AI should complement, not replace, human judgment. Credit unions that find the right balance between technology and human expertise will be best positioned to safeguard their members and grow their small business portfolios.
With fraudsters always finding new ways to exploit weaknesses, credit unions are realizing that AI isn’t just a luxury; it’s becoming a necessity. And while the road to clean data and smooth AI integration may be rocky, the payoff for credit unions and their small business clients could be huge.
NOTE: The following transcript was created by our robot overlords. It could have a boo-boo here and there.
Sarah Cooke
Hello and welcome everybody. My name is Sarah Snell Cooke. I’m your host here, of course, at The Credit Union Connection, I’m here today with will Timothy. Welcome.
Will Tumulty
Hi, Sarah. It’s great to be here.
Sarah Cooke
Yeah, you’re, you’re a, maybe a three for at this point. Um, do so will is the CEO at rapid finance, and why don’t you go ahead and give a little more introduction to yourself and your company?
Will Tumulty
Yeah, something’s good. Sarah, so we are a provider of capital of small businesses across the United States, in addition to the capital we provide in the forms of loans and lines of credit, of credit. We provide technology to other small business lenders that either have existing programs or want to start a small business lending program. We’ve been around for about 20 years, so we’ve been doing this for a while, and we’ve figured out a couple of things that can be helpful.
Sarah Cooke
Cool, yeah. So the research that you found and sent over to me said that fraud against small businesses has increased 70% since covid that is, that was just an astounding number I saw. So why are these businesses so attractive to fraudsters?
Will Tumulty
So, so I think it’s really the lenders, right? So the lenders are the ones who will get the who will incur the losses if fraud is not detected. And you know, historically, small business lending has been a brick and mortar operation. The small business owner has a checking account with their bank or their credit union. They walk in they apply for a loan or a line of credit, or maybe they’re doing it via credit card with one of the big credit card providers, a Capital One or a chase or someone like that. There are companies like rapid that have been online lenders to small businesses, but they have been less it has been a sector that has not had as much awareness. In the pandemic. The federal government had two really large small business lending programs, the paycheck Protection Program and the economic injury disaster loan program. Of course, during the pandemic, everything was online, and I think that really opened up both the awareness for legitimate small businesses that you can go online like consumers can to seek capital that you need for your business. But of course, it also raises the awareness of the fraudsters who are out there,
Sarah Cooke
yeah, yeah. And they get it’s always seems like there’s a positive and a negative to whatever it is such as AI that has definitely helped fraudsters scale their operations. I don’t want to call it a business, but I guess it is, but against legitimate businesses and their lenders. So and it can help and but it all, but it also can help credit unions avoid fraud. What sort of impact can using AI have on this type of fraud?
Will Tumulty
Yeah, so there’s a variety of things that you can do with AI. AI is a sort of foundational technology that you can build into your lending processes in a lot of different places. Um, it can help you automatically read documents so you don’t have to have people do it. It can review those documents to see if they look like other legitimate documents, or whether they look like they might be fraudulent. It can also look for patterns across your data. That is something that we spend a lot of time here at rapid doing in our systems, where we will get a new application. That application will come with a number of data elements that are required, like business name, business owners, tax IDs, right? Those sorts of things. We will go to third party data sources to validate all of those elements, but then we’ll take all of that enhanced information, and we’ll compare it to every other application that we’ve seen in the last call it decade or so. And we use our AI driven identity and fraud mitigation platform that we call links to identify those related applications, determine which data elements they are related by, and if there are indicators of fraud that are associated with any of those that might be related, we will propagate those to the applications that you’re looking at today, right? So AI can be used to find those patterns and highlight to your loan officers or underwriters where there is potential fraud that they should investigate.
Sarah Cooke
And so, yeah, I mean that all sounds like great, great things that would be very, very beneficial. How do you address the skepticism, though, that about relying too heavily on AI?
Will Tumulty
Yeah, so, right, we all. I. Have read or seen the things where, sort of, AI goes sideways and gives you nonsensical answers. The technology for AI, sort of the underlying technology, is moving at an incredibly quick pace. So the improvements are are profound. However, if you’re training an AI model, it’s only going to be as good as the data set that is trained on. And so if you have dirty data, unclean data, messy, disorganized data, and in particular, you haven’t tied the outcomes that the model needs to understand what it’s looking for, your model’s not going to be useful for what you’re doing. And I think based on conversations that I’ve had, that is the biggest challenge that most financial institutions have in using AI in this particular kind of way, is that their data is, frankly, it’s kind of a mess. In most cases, it’s siloed, it’s in different parts of the organization. It’s not all together, and so you it is a painstaking effort to get that data to a state where an AI model can be effectively trained by it.
Sarah Cooke
Yeah, yeah. Obviously, there’s a lot of data to sift through which humans can’t reasonably do. And I’ve heard a few people, from a few people, that credit unions data is not always clean, not always tightly wrapped in a bow, so it’s not easy, even sometimes, for AI to handle. So how can credits clean up their data and where it’s coming from and get it all in one place so that AI can do a better job of, you know, or easier job, I guess, of what it’s doing.
Will Tumulty
Yeah. So you know, there’s a good reason that credit unions and other financial institutions why their data is all over the place, and that’s really for kind of two reasons on either end of the value chain. On the one end of the value chain, you have products that you offer to your customers, and those products might be checking accounts or savings accounts or credit cards or auto loans, and each one of those products comes with its own set of regulatory requirements. Therefore, your front end acquisition process is different. For all of them, you collect a lot of the same information, but you’ve got to say different things for mortgages, for instance, than you do for credit cards. On the back end, the providers of the systems that support these products are the core providers, right? So the FIS and the Fiserv, right? The Jack Henrys, those kinds of folks, and very few institutions, keep all of their core systems with one of those providers, which means you’ve site, by nature, got this fragmented data challenge that you have to bring together. There’s really only two ways to tackle that problem. There is, if you want to get all this data together, you can do it on the back end, or you can do it on the front end. If you tackle it on the back end, what you’re doing is you’re getting your nightly downloads or your API updates, however it comes to you, from these various core systems and your front end acquisition flows, and you’re doing a whole lot of ETL data processing and data standardization and all that kind of stuff, particularly for legacy data, that can be a gargantuan task and takes a long time and A lot of people to really sort through it. The other way that you can do this is you can do it on the front end. This is actually easier and quicker, but it does take some infrastructure. It’s helpful if you built it. So for instance, what we do is every one of our front end acquisition flows, where we’re getting an application takes that information and funnels it into a production system, not a data warehouse, but a production system that we call links. And that production system, first it standardizes the data, then all of the data enhancements coming from those third parties come back to that, and those are also standardized, and by doing that, through all of our different flows with all of our different products, we have created an application data layer that is clean, it’s consistent and doesn’t change over time. Importantly, it’s a production system, so we don’t have data analysts going in there and running a script or trying to do reporting off of that. We can spool the information out for all those purposes, but it’s locked down so it doesn’t get corrupted. What we use that for is we get that periodically, every couple of hours, right? We’ll spool out the latest, updated information, and then we can put that into a clean data warehouse that our data scientists can work from.
Sarah Cooke
Okay, yeah, and I’m sure that’s very helpful, having that already done for you. But so now, how can credit unions balance security with the need to go. Or their loan portfolio. How does, how do you ensure that credits and unions aren’t necessarily turning away legitimate business borrowers?
Will Tumulty
Yeah, so, you know, there, there’s a couple of different levels of security, or things to think about in terms of security. There is, you know, how do I secure usernames and passwords and do two factor authentication, all those kinds of things for my members, credit unions are good at that, right? They do that. They work with their vendor suppliers for those kinds of things. I think what you’re really referring to is, how do I determine whether I have an illegitimate or fraudulent applicant that’s coming in right? On the small business side, it’s really a bit of a more complex problem than on the consumer side, if you have a consumer that is opening a new account or applying for a loan, really the tax ID or the social security number is the primary key, right? And it’s like, okay, if I can validate that, and I can attach that to this person that I’m speaking with on the other end of the phone, either through interviews or out of wallet questions, or however else I do that, then you’ve kind of got some pretty good assurance that the person is who they say they are when they’re making an application. Small businesses are more complicated, because from a data standpoint, a small business could have many owners, right? There might be four co owners of a small business, or there may be one owner, like a franchisee, who has four different small businesses, right? I’ve got four McDonald franchises. They are all incorporated separately. They all pay their taxes separately. And so you have this many to many data relationship between small businesses and small business owners, and sometimes they’re very similar. So in that franchisee example, we might have a John Smith who has four McDonald’s all in the same town. In really the only thing that’s different between those are they’re at different addresses and they have different tax have different tax IDs, but they’re all called McDonald’s, right, right? So so it can be difficult, so really managing that data well and knowing what data elements are in common with other things can help you determine whether this is John Smith with, you know, one of his real McDonald’s, or somebody else claiming to be John Smith, who says, hey, now I have a fifth franchise here, and I’m trying to get a loan for that particular fictitious business.
Sarah Cooke
Okay, yeah, that makes sense. It’s so sad, like the fraudsters, if they’re often brilliant, if they had just put it toward good, right? But also, one of the things with AI is not just the technology itself, but also the people that are working with it and around it. So there’s the people the culture to kind of shift. How do you suggest credit unions handle
Will Tumulty
that? Yeah, so I think from what you read in the headlines, there can be a lot of fear around AI is coming for everyone’s job. And you know, I think I expect what we will learn over time as this whole AI boom plays out is it will make things much more efficient in certain areas, but there we’re still going to need humans, right? We’re going to need people to do a lot of the thinking, a lot of the training, a lot of what should I be asking AI to do? Even on the software development side, when we develop new AI modules for our various systems, we do those, but then we need traditional software engineering to put a wrap around that to fit in the rest of our business processes. And so there is, I’d say it’s like the next extension of tools and capabilities in the world of technology that is really useful, because it can do some things that historically haven’t been able computers have not been able to understand, and it can do it faster, it can do it more reliably, but it’s really a partnership between the people and the AI that makes it work.
Sarah Cooke
Yeah, yeah. I agree. As a writer, I was a little concerned at first, but I totally see that that happening. So looking ahead, what’s the largest blind spot that credit unions may have when it comes to business lending
Will Tumulty
fraud? So I think it depends on as a credit union, where are you with respect to small business lending? There are some credit unions that do a lot of small business lending, some that are thinking about doing small business lending, that are but are not doing it today, and there are others who are that’s just not part of their strategic outlook, right, not something that they’re looking to do. But if you’re doing small business lending, or you’re looking to do small business lending, there is this combination of capable. Abilities that you want to bring together that really leverage things that you would do on the consumer side, which tend to be data intensive, high volume, low touch, at least early in the application process, and things on the commercial lending side, which tend to be much larger, dollar more data and documentation intensive. And for small businesses, because you have to deal with the business owner, right? You need to be able to do those things that you do on the consumer side. And because you’re dealing with a business, you need to be able to do those things on that you would do on the commercial side, and really meld the two together so that, that, I think, is the challenge. It is easier to do a $2 million loan to a specialty contractor, for instance, where someone else and put $2 million to work as a credit union, than it is to say, well, rather than do that, I’m going to do 40 $50,000 loans. To put that same $2 million to work. Right to do 40 $50,000 loans, you’re not going to do that with your loan officers and doing site visits and all those kinds of things. So you need some technology infrastructure to enable you to do that, or you can partner with a financial technology company, like rapid to bring those offers to your customers. So even if you’re like, hey, I don’t want to put it on my balance sheet, I don’t have the capabilities for it, can still be very valuable to a credit union, even for your existing members who happen to be small business owners, to say, Yeah, through a partnership, I can bring capital solutions to my small business customers, make a stickier deposit relationship, have those deposits be larger to help on the Depository side and keep them from going down the street to a competitor.
Sarah Cooke
Yeah, yeah. Definitely a big upside on the business lending for credit unions, potentially so as we wrap up here, you know, I always allow my guests final thoughts. What do you want to leave our credit union audience
Will Tumulty
with? Yeah, you know, credit unions are really in kind of an interesting space. So historically, there’s been a limitation for credit unions, and there’s still limitations on how much commercial lending credit unions can do, but it’s a really valuable thing for the community mission that credit unions have, right? So credit unions serve people who live on Main Street, and they can also serve the business owners who work and have their businesses on Main Street. In a lot of cases, I would venture that there are many credit unions that have personal accounts, checking accounts, savings accounts, credit cards of many small business owners. When I was a small business owner, I had my accounts with credit unions, but I didn’t have my business accounts there. Had my credit union been able to offer the things that I needed as a business. I would have loved to have had them with my credit union, and keep all of that financial relationship in one place. And so I think there is a great opportunity to serve those small businesses as part of the overall strategy.
Sarah Cooke
Awesome. Well, thank you so much. Will for your time today, appreciate it.
Will Tumulty
Thanks, Sarah, great to talk to you as always, bye, bye.