The Biggest Technologies Shaping Banking Right Now // Hans Morris | Plugged In 1x36

Coming up, a punk rock-inspired episode of Plugged In. This tech-focused one features Hans Morris, managing partner of NYCA Partners, a fintech VC in New York and San Francisco. Together with my main man Steve Williams out in Scottsdale, Arizona, how's it going?

Well, I'm Al Dominick, holding it down in our nation's snowy capital. The first snow of the year came through the District, and I had to open my window so you could see this is happening. But we're going to have a great conversation today that builds off recent ones with bank CEOs like Dave Brager from Citizens Business Bank, David Findlay of Lake City Bank, Jeff Jackson from WesBanco, and Nitin Mhatre from Berkshire Bank. All of those conversations brought to light various interests and real opportunities to leverage technology tools in some pretty meaningful ways. Did those strategies call to mind other conversations we've had with folks like Chip Mahan, Brent Beardall and Kevin Blair?

Absolutely. But today, I like to say, follow the money. For our bankers, understanding what's out there in the ecosystem, in the VC world and in the tech investing world, including the folks who are going to partner with banking and also disrupt it, makes it great to have Hans here.

Totally agree. As much as we learn from great bank CEOs, it's an opportunity for us to cast a wide net and get perspective from somebody who has been doing some really cool things. Steve, I know we love stats, and Hans, you've probably heard Cornerstone talk data in the past, but let me throw this one out for listeners. NYCA has more than 80 active investments and is currently investing from its fourth fund. In addition to Hans' regular day-to-day job, I believe you're still chairman of the board of LendingClub.

I am, yeah.

And you're on the boards of several private companies, which I had noted as SigFig, Thought Machine, Sentilink, and some others.

Yep.

Right, so you're just bouncing around doing different things. But I mentioned that Hans is joining us for a punk rock-inspired episode, and I have to disclose to you guys that he's also a huge wine connoisseur, as is Steve, as am I. To prove this, I pulled a bottle of Château Cos d'Estournel from my birth year of 1976.

Wow.

It's empty, so we can't share it virtually, but you're keeping company with some folks who love a good vineyard. We also happen to love some good vinyl. Your friend Nigel Morris at QED was the one who connected Steve and me with the idea that you know a lot about punk rock. We're not going to split hairs here. Your listeners can't see it, but that's a picture of Paul Simonon from The Clash in the background, then Tom Verlaine from Television and Patti Smith, and James Brown, who's the Godfather of Soul.

Actually, in the late '70s and '80s, I saw James Brown at a very small club in New York, the Ritz, right after he got out of prison.

See, now this is why you have Hans on Plugged In, because you never know this type of excitement exists in the hallways. What's funny about you showing those posters is I pulled five different bands, thinking they would have lyrics that inspired us. The first one I pulled was The Clash. I also looked at the Sex Pistols, the Ramones, the New York Dolls and the Stooges. So I had a good group. I'm hoping mine is okay.

Yeah.

Steve, when you were a boy and used to take a walk into the music store, are those ones going to resonate with you?

Oh, absolutely. It feels like a lot of good New York vibes, with some connection to London, but a lot of good New York in your list.

Let's kick things off and tell some stories. Steve and I were out in Chicago a few months ago. We heard Hans speaking with Nigel, and Nigel teased up a Visa story and a lollipop story. It's an acronym Hans created. I made some notes, but I can't remember the exact details, so I'm really hoping he can remind me what he was talking about.

In 2007 and 2008, really when smartphones were coming out, you had what I would call an unlimited stream of ideas. It also coincided with the Durbin Amendment. There had been an increase in merchant-funded rewards, and then that suddenly became uneconomic with Durbin for debit cards. So all kinds of entrepreneurs came up with ideas around having your smartphone and location, being able to push things to you in a given location, and being able to combine what you were searching for online with your payment history, maybe social media likes, birthdays and so on.

I would see some new idea every week, sometimes several a week, and I got sick of it. I came up with an acronym that tried to encapsulate all of them, which was LOLLIPOP: linked offers, likes, loyalty, incentives and promotions on payment systems. I've been using that ever since. We're probably on the third or fourth wave of new ideas now.

One of the things I said at that meeting is that it's a lot harder to do than it looks. It's easy to make a video about how it could work. It's very hard to make it all come together in a way that's actually really valuable to you and to the merchant versus just becoming spam.

We invested in a company called Point.me a couple of months ago, which I think is an example of newer strategies. There's a lot more data now, and what's important is focusing on making it extremely valuable to the consumer while also making it extremely valuable to the provider of the reward. That's quite complex. We could probably spend the whole podcast talking about this, but I do think there is reason for optimism. Some of these are interesting.

Steve, what do you think? As you hear all this, it's kind of the smarter bank concept coming to life.

Yeah, I couldn't agree more. People are trying to say, now I'm going to embed this in a merchant and payments network ecosystem, and then banks are going to have to embed it somehow in all these mobile platforms. It's kind of like Elon Musk saying no one understands how hard the manufacturing process is. I totally agree with Hans. No one gets how hard that concept is to embed in an ecosystem.

And I'd add a couple of lessons. One of the reasons we like Point.me is that there are certain things Google is not good at. Say you want to go to Paris, fly business class and see what your real options are. What rewards programs do you have, and what's the most efficient way to get there? That's a pretty complicated equation. You also have to decide whether you're willing to tell some application a lot of information. Point.me figured that out and figured out a way to incorporate the carriers into this so they want to share a lot of information, too.

If you don't do that, you get something a little like Gresham's law from economics, where bad money drives good money out of circulation. That's basically the problem with a lot of the existing models. They're merchant-funded, and the merchants don't really want to give away things that are valuable, so they give away things that don't really mean much to most consumers. That's the tough nut to crack.

Thank you. I probably could have used some music to tee that one up, but I certainly will for this next one because I mentioned The Clash and they've got a song, Train in Vain. There's a lyric that says, "Tell me something I don't understand." So I'm going to apply that ask to what's happening in the AI space. It's a broad catch-all. We should probably be more specific and talk generative AI, chat, RPA and so on, but it seems anything customer-service oriented with humans deeply involved is something Steve Williams and I have been paying a lot more attention to. Companies like Klarna, where they're doing buy now, pay later deployment at scale, really have our attention. From your investor's lens, where are you seeing companies gain traction?

As you said, there are big differences under the category of AI. If people think about the progression from regression-based models, where you can take several variables in a multivariate regression but models start breaking down as you incorporate more and more variables, that got pretty far, but there are limitations. Then machine learning enabled you to build models with many more variables, and the machine would really figure out the best correlations within a wide range of variables in a data set. Supervised machine learning is another standard that has become pretty common. Lots of banks are using that.

All of this has to be in the context of being a regulated institution. If you're a bank, the model regulations, which I believe came out around 2016, are very specific about what you need to do to incorporate any model. There are requirements you have to meet. Importantly, even if you're using a vendor, you have to supervise that vendor. You have to know what the vendor is doing and how that model works.

When you incorporate generative AI, which obviously everyone is talking about, it creates some very exciting ideas. It's a little bit like the videos we saw in 2008 about how you could use smartphones, incorporate lots of data and payments data, and create an incredible experience. It's harder to do in practice, particularly in customer-facing applications where there's going to be a lot of sensitivity because it involves money.

If I'm telling you how to invest your money and I'm right half the time, you might say that's not very good. If I'm marketing to you and I'm getting a 50% success rate, that could actually be very good. Things that are acceptable in one context, like e-commerce, are not acceptable when they involve money, particularly if the activities themselves have very specific requirements. Saying AI will get me most of the way there to meet those requirements is not acceptable.

An analogy I use a lot is self-driving cars. You can make them very good. You can make them maybe better than humans because humans make a lot of mistakes and have bad incentives when they're making investment recommendations, too. But we still need to meet the requirements of existing regulations. To me, that's very clear.

So what are some examples where you could put AI into production right now? One, as you pointed out, is sophisticated customer service where you have a human who's responsible anyway. A human is going to be in the loop in many cases, but can I allow that machine-assisted human to be a lot smarter, including speaking any language and incorporating many complex decisions?

Think about how frustrating the "How can we help?" chat on a lot of websites often is. Those machines are getting much smarter and will get much smarter. That is clearly a very powerful application. You still have to test it just like you test a script in a call center today. You assess how well your customer service people are responding to customers' questions. You make sure you're monitoring complaints. All of those things are going to remain part of your requirements.

I like putting these in categories. What's in production right now? Meeting summaries. Lots of businesses do it internally to make you smarter and help you do things faster. We use that all the time now here, too. Coding is another application where everyone is saying it can create dramatic improvements. One thing coming out of the data now, though, is that it makes really good coders more efficient. It doesn't necessarily make bad coders more efficient. You still have to test the software. If someone isn't good, it's not necessarily going to have the same dramatic efficiency gains as when you have really good coders using it.

Another summary I would give your listeners is to focus first on making expensive people more efficient. In wealth management, there's no question generative AI can assemble documents effectively, interpret what that customer is seeking to do and make workflow much more effective. Make things more seamless for the expert person who's expensive and make them much more efficient.

I think that's promising in business banking, too. It can help people prepare credit files and determine what the clients' real needs are. Am I on top of all the activities of the relationships I'm responsible for? Do I actually know what they're up to? Am I meeting all their needs? That's something generative AI can help with, and there are lots of promising ideas there.

I agree with that. Commercial officers are expensive, Al and Hans, and they kind of like the art form of being in their office working on the credit memo. We saw nCino sell to Moody's this week, and I think one of the things they were really leaning into was generative AI for the credit memo using some of their more quantitative data as well. That is a great place to look in banking, not to get rid of officers, but to double their portfolio with a robot attached to them.

Yeah. Think about the way you would typically get information. You'd say, "Let me take you out to lunch," and then you'd get updates from the customer. I could go into that meeting a thousand times better prepared with AI tools. Sales efficiency is another example where I think there are lots of companies looking at ways to make that more efficient.

As you two are talking, I'm jotting a few notes down. You're talking about data and the customer and having a better sense of where they are and what you might do for them. That bleeds into another topic around embedded finance. Over the next five to ten years, it really does seem that anything analog will be digitized, which is a theme I've heard you and Nigel talk about. As it does, institutions can capture data that no one else can really get their arms around. You're talking about transaction data that others can't see like a bank does.

I've got to pull in the Sex Pistols and Anarchy in the U.K. for this one because, "I don't know what I want, but I know how to get it." Bankers keep hearing about the promise of embedded finance. Steve, I think you had a question around embedded finance for Hans.

Yeah. You're in the flow of money and investing in hypotheses all day long. Given that we see the potential of embedded finance to add value because it's low-friction, personalized and driven by data in an integrated ecosystem, pound for pound, do you think embedded finance will be a net loss of shareholder value to traditional banks, or will partnering and creating lines of business around this actually be a net gain for the banking industry?

I think that's somewhat up to them. When I talk to bankers, I say every single business vertical, even if you just take healthcare, probably has a hundred different verticals within it. There are providers, pharma, big pharma, specialty pharma and so many different business processes around all of that.

What we look for is often some new SaaS software company because almost everything has become cloud-based. In fact, almost every operating system is converting to a cloud-based offering. What is going to be the financial operating system in the cloud-based offering for that vertical? It may be thousands of different software companies that do that.

If I were giving advice to a bank, I'd say, what are your areas of business specialization today? Let's say you're a big lender to various philanthropies and cultural organizations. That's a niche you have in a certain community, and maybe you've expanded beyond that because you've developed really good expertise and great referrals. That's a good example. There are a whole bunch of things those entities could use, and you should provide them.

You should be the one saying, "I want to embed that into your ticketing software. We want to create the ability to offer you a way of adding a button where you press and automatically make your annual donations through the same system. It's automated for you. You don't have to do anything. We provide all your receipts right here in one place." I'm just making that up, but if I were a bank CEO, I would be talking to my specialty commercial group and asking who the key software companies are, what the leading new cloud-based offerings are, and how we can work with them.

How can we embed factoring, short-term financing, payments and longer-term credit products? Can we offer payroll solutions right in that app to those companies? What insurance products can we create? We have a company, Grid, that offers earned wage access to lots of companies. If you have employers who need earned wage access, how about an embedded wealth product for a higher-end employee base? Focus on where you already have a competitive edge, but you should absolutely be focusing on how you can build smart, programmable applications into it.

It strikes me, Al, when Hans is talking, that for banks to do that, they're going to have to figure out where that work takes place because it's kind of stuck between IT, the credit group and the treasury management group. It goes back to something Brent Beardall talked about, creating a product group that can bring this together and create product solutions to deliver to customers.

Yeah, and some of these are very simple. They already have APIs in some cases. They have SDKs built, so it's very simple for the business partner, your customer, to download something and incorporate it into some other software they have. I don't know if you know the company April, which is automated tax prep, but that's like a button. I don't know why every bank doesn't say, "We offer all of our small businesses tax prep. Press this button." Embed it into your banking app or into another operating system that all the companies in a vertical use.

Steve, as you and Hans are talking, I'm an optimist, so this is unfamiliar territory for me to venture into, but I also see the challenges that might exist in terms of fraud and real-time risk management. With total apologies to the New York Dolls and their song Personality Crisis, because they did not want this to be used for commercial purposes like this, let's hear what you have to say about fraud and real-time risk management, Hans.

I love a fraud company. All of the changes to payments networks and other tools that enable payments to be real-time obviously create a lot of risk. The friction you had with two-day settlement gave you the opportunity to check things. Now you don't. Many applications are batch, so all of that has to be rebuilt. The same thing is true with corporates. A lot of ERP systems are batch. Large corporates use SAP as a batch system. You want to be able to reconcile cash in real time and assess payment risk in real time.

The nature of fraud continues to compound at a rapid rate. Fraudsters don't have any of the requirements banks have to test software for a period of time and make sure it's ready. They don't have a regulator ensuring that when they put something into production, they've checked lots of things. A fraud tool may add a lot of friction, may upset customers or may bias against certain customers. Those things don't really matter to fraudsters. They will just put things into production.

I imagine these fraudsters in an agile development class talking about speed to market, but you're saying the banks don't have that luxury.

Yes. There's also a technique that was originally developed, I believe, for modeling infectious diseases called agent-based modeling. It requires unbelievable computational power. It's also used for things like modeling whether an asteroid will hit and what its path will be, so it has all kinds of important applications in science. It requires tremendous computing power because you model how every agent, every participant in the system, reacts to the activity of every other agent in the system.

If you have an infectious disease and I walk into the house with you, do you get sick? What actions do you take? Do you stop meeting with people as it spreads? You model all of that. It's a very good application for fraud because whatever protections I put into place in my institution, which may be a very good signal that blocks a certain type of fraud, the fraud just mutates and responds. They will go attack weaker institutions that don't have that protection, or they'll come up with some other new thing that exploits another problem I have.

We've given this quite a bit of thought. We have some very good companies in our portfolio focused on fraud. That includes SentiLink, which is used primarily for account origination, account takeover and first-party fraud. It's a great company, very smart, and they keep adding new products. They have a lot of confidence from their clients.

Sardine is another company focused on a lot of payment fraud applications. Another company, Allaria, is interesting because it came from the crypto world, as did Sardine initially. There are a lot of bad actors in the crypto world, so seeing the activity of those bad actors and being able to model great signals from that is valuable. Allaria is focusing on scams and has a very promising scam product.

What you really want to have is something beyond point solutions. In fraud, card-not-present fraud is completely different from HR fraud, account origination fraud, AML screening or transaction screening. They're all different systems. But in fact, what you would like to do is see the signals and have a graph from all this activity. We see the convergence of fraud, AML, cyber and crypto because a lot of the bad actors transact on crypto now, and that has become a giant-volume business.

A lot of cyber signals involve state actors and associated groups. Those are people who are also doing AML and other fraud activity and trying to wash money through the system. How can you create a view of all of that, and what would that really look like? That's something we're intrigued by. I think this is going to be the story for the next long while, maybe forever.

I also like pointing out that we have another company, Figs, which is in tenant screening. That is changing rapidly. Tenant screening used to be, I need your Social Security number, a credit report, employment verification and a criminal background check. That's it. I didn't care about anything else. If I did those things, I had given the landlord a tenant screening and they were happy.

There are criminal gangs that will, if you pay them, create a synthetic identity and allow you to rent free in Los Angeles because it can take two years to evict you. For the next two years, you don't make a single payment on a $4,000-a-month apartment.

All right, Hans has given us a wealth of ideas and information, and I love that you're bringing real-life examples of companies doing cool things that can help strengthen an institution's position. I'm going to wrap this up, and I'm not sure if I want to go with the Stooges' Gimme Danger or have the Ramones put me in a wheelchair and get me to the show, but I've got to ask you: we've got the holidays coming up. What wine is appropriate to give as a gift to friends or family members who may not be as much of a connoisseur as you are? Any off-the-cuff suggestions that aren't just a Pinot or a Chardonnay?

I'm a good-value type person, so I like picking regions that are out of favor and that people haven't heard of. One I think is just a really good deal right now is white Bordeaux, which is mainly Sauvignon Blanc but is usually blended with other grapes. There are excellent producers, really top-tier producers, where you can buy a bottle for $50. To me, it's much better value than high-end California whites, which are now very expensive.

If you go into a good wine store and ask for a really nice white Bordeaux for $50, they'll come up with something. I could give you some producers, but that's an example of a really good deal right now, and it's a treat when you get it.

Similarly, I like sweet wines. There are lots of them, and they're also a treat when you bring one to someone. You can spend a lot of money on Château d'Yquem, but you can get some fantastic Sauternes and Barsac from other regions. And Vin Santo, if you've never had that from Tuscany, is absolutely wonderful. It's impossible to drink it and not feel better about yourself.

Well, that's what we need, to feel better. When I see you again, I'm going to bring a $50-or-less bottle of white Bordeaux so that we can toast to this episode and all the others we've had on Cornerstone. I just want to take a quick second to thank Steve Williams, thank Hans Morris and, again, thank everyone who's listening in to Cornerstone and getting plugged in with us today.