AI in Banking, Core Systems & Fintech Trust: Insights from AFT Spring Summit

Speaker A: Foreign. This is Sam Kilmer, Managing director, Cornerstone Advisors, coming to you with another unscripted episode of FinTech Hustle podcast. And we are in the hall at the aft spring summit here in Nash, Vegas. We got a little bit of the skyline behind us. It's been a windy cold. I mean, it's a spring summit, but it kind of had a little bit of a winter vibe before being honest. Tornado, winter. Tornado, winter vibe. A, uh, little dangerous, a little challenging. But, um, I'm joined on this episode by four rock stars from the fintech community. We're just going to be here talking in the hall, as we would. The whole point of fintech Hustle is sort of like you're at an event, you push the record button and you record some of those conversations that you loved, um, from, you know, when you're, when you're in the thick of it, you know, we're in the middle of this business. We love what we do. So let me start, uh, over here on my far left is Laura Kornhauser, who is the CEO and founder, um, of Stratify, I believe you're based in New York City, if I'm not mistaken. Um, and also former banker. You were J.P. morgan, right? J.P. morgan, 12 years of JPMorgan. So banker, Fintecher, troublemaker. In the middle of the whole thing. Laffer, she's a great, she has a fantastic lap. We may get some of that later. Um, also, uh, David Eads, who was the CEO and founder of Vine Financial, and, uh, boy, David, I think we've known each other for a while and, uh, I think you were the founder of Growth Solutions as well, as I recall. So you've been around the business quite a while. And so welcome to you guys. Uh, uh, Kirsten Longnecker, who is executive vice president, um, at York Public Relations, and so working a lot with lots of folks in the fintech world and seeing a lot of things. And also I think one of the things that I've loved about some of Kirsten's content is kind of in a sea of. There's a fair amount of boring content out there in the banking fintech space. And she's always been really good at sharing her, you know, kind of free knowledge, um, on ways to make content better. So we'll maybe we'll talk about that

Speaker B: a little bit later.

Speaker A: Um, and then also Tim Hamilton, who's the CEO and founder of Praxent in the, um, I think I'm going to get this right in the sort of in the development space, in developing Applications for fintechs and banks. Uh, a lot more builders these days. So, uh, I know Tim and his team have been real busy, busy. So, um, I guess without any further ado, let's just sort of jump in and, um, start, uh, the conversation. And I guess I would just ask. And Laura, maybe we could just start with you. Uh, what's a day in the life of Laura Kornhauser in New York City look like?

Speaker C: Well, most days start up, uh, getting woken up, uh, way earlier than I'd like to by one of my two boys. I have two boys, five and three. Uh, they always start the day, and first things first. Got to get them out of the house before. Before anything real can happen. So that starts every single day. Uh, and then I am privileged to be able to work from home. We're a fully remote team. And then I'm just diving into whatever is the focus for that day. On the company side, it's such an interesting time we live in right now. So I look to spend as much time possible speaking to bankers, both existing customers and prospects, uh, and then really diving into the new things that our team is working to build. We're in a time right now where the environment is changing so rapidly that the set it and forget it mindset definitely does not work right now. Uh, so we, even as a team, have weekly sprints, weekly focuses that we've pulled down and pulled even tighter, um, over the past year, in particular, to be sure we're responding to the latest in market challenges. So I try to spend half my day talking to customers or prospects, uh, or just friends in the industry about what they're seeing, and then the other half of the day just helping my team remove blockers. I, um, see it as one of the biggest parts of my job, is how can I make everybody that works with me more, uh, effective, more efficient by getting things out of their way.

Speaker A: So, as a father of three, I tip my hat to you. Starting off your day with. And all three of mine are boys. So, uh, between. You said. You said two boys, which is. You said is a different energy. Two boys. And then weekly sprints. This all sounds very exhausting, Laura, if, um, I'm being honest, but it sounds like energizing.

Speaker C: It's energizing too, right? I think that, again, all of us are builders, right? So this is the kind of thing that we wake up every day super excited about. How can we learn more about the problems that our customers are facing, how those problems are changing again? These days? It is on a daily Weekly monthly basis, it's not on a quarterly or annual basis. And how can we be adaptive and responsive and help them understand what is now possible that maybe wasn't even possible three months ago? Um, so it's an exciting time.

Speaker D: Laura, I just wanted to key up on what you said about um, increasing the agility and the velocity. I think that with cloud code and generative AI generally tools that are going to accelerate the delivery of value, um, I think we're going to see a huge wave of customer centricity sweeping across the fintech uh, ecosystem. I think for many, many years or decades software vendors and software providers, it's been so capitally intensive to generate software that there's been a level of complacency that we've been able to get by with um, multi year contracts with incredibly high licensing fees. I think the buy versus build economics are in the process of being completely rewritten. And I loved what you said shrinking from two week sprints down to one week sprints, becoming much more um, iterative. I also think customer centricity is, is, is going to significantly increase you in the next chapter.

Speaker A: Well, since you've got the mic, what's a day in the life of Tim look like?

Speaker D: Yeah, as you might suspect we've been spending a lot of code with generative AI. Um, so Sam, as you said we do a lot of fintech product engineering um, to help uh, uh, owners and operators of legacy systems and legacy software to refactor pay down technical debt or rewrite. Um, and we've been using CLAUDE code to modernize Delphi code bases if you guys have ever heard of that. Cobalt code bases still over 85 or 90% of transactions in the United States travel over cobalt rails. It is uh, incredible. And the number of engineering teams we step into where like Steve, the last man standing, the, the, the 75 year or 80 year old uh, cobalt engineer is dying to retire but hasn't been able to um, we've been able to use CLAUDE code to unlock business requirements and actually pave the way for what's next. So I'm spending a ton of time with clients and prospects and our teams to, to help them to plan those projects and use the latest tools to unlock the future.

Speaker B: I'm glad you brought up AI uh coding because you know there's been a lot of discussion in the halls about AI uh coding and I personally, I feel like the fabric of the universe is ripped with you know, Claude Opus 4.6. And I did some math on Saturday morning while I was Brushing my teeth using Claude. And, uh, I wasn't brushing. I wasn't brushing my teeth with Claude. No, no, no, I wasn't brushing my teeth with Claude. I was brushing my teeth and prompting Claude. And I just. I was thinking, what is the delta between the code that we were able to write, uh, since this new version came out about a month ago and the previous month, and it was up just in lines of code, which we all know is not a great metric. IT understates it 286% more code written month over month because of that. And part of what we were doing with that was APIs. You were talking about the vendors with the expensive APIs and the legacy code base and all that. I think that's a perfect use of AI. To me, that's the main use case, at least the initial use case for using AI in all of our businesses where we have to interact with all of these legacy players. You know, it used to take six months or a year to integrate into a core. Now if, uh, the business part of it stays under control, the technical part of it can be hours.

Speaker A: Okay. And so besides getting AI to brush your teeth, that's the way that I think we might. As my. As my. As my son Zach has been known to say, because undergraduate in software development and AI, he always his, his, his favorite quote is, who told you I could do that? Um, and so I think we found our use case that maybe next year at aft, we'll be brushing our teeth with code. You're right. So that's. That, that's fun. I love stuff like, this is why we don't prep for these things, because you're going to trip over. This is fun. But what's a day in your life look like?

Speaker B: Well, my kids are grown, so, you know, the last company, my kids were young and I had to do what Laura did and, you know, help get the kids out to school and help watch them come in and all that. But now my kids are grown and actually some of them work in fintech. Even, uh, I'm actually, actually at the last. AFT was my first old man moment where, uh, Will Bryant came up to me. He goes, hey, hey, guys, you got to meet. This is Thomas's dad. And I was like, this is the first time I've actually been introduced as, you know, like in a business context, as somebody's dad, you know, But I didn't tell you.

Speaker A: That's fantastic. And it kind of reminds me of when my son said, that's a dad joke. And I'M like, I'm a dad. That's all I got. So. But, but um, yeah, and I have two. It's funny you mentioned that because I, uh, my three sons, the youngest of which is 17. So a little young for fintech other than his own personal, you know. Um, but the, the older two are actually, it just turns out they both have landed in the space. So it's kind of funny. Although they're product, they tend to be product and developer types as opposed to, you know, go to market people and they're like dad, don't go by our booth. And I just want, I want to get on Slack and execute. Just don't do this to me. So I'm like, I just did it. Okay, uh, so I'm going to ask for forgiveness guy. Um, so with that, Kirsten, what's a day in the life of your world look like?

Speaker E: Um, so I actually think of my day in terms of my biorhythms. So I really start, um, I'm most creative and I'm most energized. Um, thinking through concepts, um, turning something, ah, complex into something simple and easy to digest and put into somebody's mitochondria. Um, so mitochondria. Um, um. So, um, I'm often writing first thing and um, that writing in my world is in terms of making our clients look good. It's um, making sure their thought leadership, it sounds like them, represents them, but also furthers conversations, um, between our clients and whomever their audience is. Um, and then by 2 o' clock I'm like, I'm in the clicky pasty zone. You know, I'm, I'm researching and on LinkedIn making sure I'm keeping up with industry insights. Um, and I also work from home. Uh, we're all remote and um, so for me I am on mountain time but my company's on Eastern. So I'm on by 7, 7:30am but by 3:30, you know, it's pretty quiet. So um, I just make sure my days are focused heavily on the kind of pie chart of where I need to spend my time for my clients best interests.

Speaker A: It's really interesting and I share your um, morning is better for creative uh, sensibility. I do find that sometimes. Isn't it amazing that like you might be dealing with some issue at three or four in the afternoon and you think it's overwhelming and then a good night's sleep and a workout and a decent meal and seeing your family and it's like wait a second. Whole new perspective.

Speaker E: Yeah. So just to build on that, I so appreciate that. But, um, allowing space for, um, problem solving and creativity, the older I get, the more important that has made itself known to me. And so I always trust the process of, um, I start very quickly on a project and then allow a lot of time for it to show how it should complete itself.

Speaker A: Very cool. Uh, okay, so we're here at aft aft Spring. Um, a lot of great sessions, um, and just a lot of great conversations in the hall. Has anything jumped out at any of you that's like. If there was like, one big takeaway for you here? I know there's a lot of them, but. Kirsten.

Speaker E: Yeah. So it's now the Kirsten Show. Sorry about that.

Speaker B: Um,

Speaker E: Michael Birdie Waite said something about practicing rigorous authenticity. And, um, I just thought, gosh, nowhere is that more important for my world. You know, I can only apply it, uh, through my lens. And that is, um, to make sure if, when I am representing a client that I deeply understand conceptually their product, their value prop, what, uh, they're offering, what they're trying to go for. So that practicing, um, rigorous authenticity is about making sure I am not, uh, feigning that I know something that I don't know and trying to purport that it's, uh. Here's this great original idea when it's just a, ah, regurgitation.

Speaker D: Um, so, yeah, amen. I can resonate with that, particularly in this industry where there's just a lot of really, really brilliant people all around us. There's a lot of complexity too, in the value chain. Um, we had a speaker, Donald Miller, who is the author of a book called, uh, Building a Story Brand. And he spoke about the importance of, um, uh, clearly and succinctly articulating what we can do and how we can help our customers on their journey. He talked about how the best product, how this injustice exists where the best product doesn't win, but it's the value proposition that is articulated as clearly as possible, um, so that our customers don't have to expend a bunch of extra calories understanding what is it that we do and how we fit into their journey. And, uh, I came away from that presentation definitely realizing how guilty I am of complicating or over complicating a message.

Speaker E: Um, I'm surprised to hear you such that.

Speaker D: Yeah, it was. It was a powerful one. I, uh, really, uh, really grateful for Donald and his message.

Speaker C: Yeah. Agree that that was an absolutely fantastic session. One of many fantastic sessions this aft, for sure. I also found it very interesting if we blend it with things like the AI session earlier this morning. We've heard from so many customers, banks, and credit unions out there that they're just absolutely. We talked about this. Right. Absolutely inundated by the AI for this AI, for that pitch, and really losing beneath it. What is. What is this AI actually doing? What type of AI technology are you actually using? What are then the risks associated that I need to make sure I'm managing and things like my TPRM process and my ongoing analysis of this vendor. And that soup is very, I think, uh, uh, mushy. I don't know, Stewie. Not clear.

Speaker E: Yeah, yeah.

Speaker C: Um, not clear right now for a lot of the customers that we all are serving. And it is our job to help break through that. Help break through that clutter and recognize that, yes, AI is a very powerful tool. It's a tool to do things. The AI is not the thing that any of our customers are actually trying to do well.

Speaker B: And as much as I'd like to say I have AI figured out and all of that stuff, I think not only is it, you know, mushy for our, uh, customers, but it's mushy for us. It's happening so quick. I mean, Claude came out February 5th. Right. So this stuff is. This stuff is all brand new, and we're all figuring it out. And I feel like I learned something every day as I'm working with this stuff, and I, I just saw the juices flowing in that room that everybody's on the same journey trying to figure it out and try and figure out how to help.

Speaker C: Yeah.

Speaker B: And piggybacking on what you're saying, uh, I think that I keep calling it the Goldilocks path. That, uh, I feel like us as fintechs, helping banks figure out AI and technology in general. We need to take a Goldilocks approach of, uh, we need to take action. We need to do something, but we don't need it to be this black box that, that the banks don't or regulators don't understand what's going on in there. We need to make sure the banker is in control of the AI, understands exactly what's going on, and that us and the banker own the output of that AI in the sense that we're responsible for what it does. And I think that's so important. And, uh, you know, I think technology, often it. There can be technology companies that get it wrong, and, you know that. That can be very dangerous in our industry.

Speaker E: Yeah. Um, so something a Company that I think is doing this. Well, besides you badasses, ah, is ah, Agent iq. Agent iq, uh, developed an agentic uh, AI process by which um, bank employees who have a high proclivity for going to external AI sources, um, are now using this agentic AI so that it is within a compliant situation, um, and it gives them the responses they need. But also it won't hallucinate when it gets to, to a point at which it is out of information. It will say, I cannot answer that for you. So, and I'm thinking of like a business or commercial banker who has to understand the state of business itself. What the banker client offers, the business offers, um, what their offerings are and kind of put that through their mental algorithm can use this to deliver um, very uh, pointed and uh, right on resources, uh, for that client sitting right across from them.

Speaker A: Yeah, yeah. In some ways this reminds me of my last company before I was at Cornerstone. We were a big Microsoft partner. And obviously Microsoft's been at the forefront of a lot of this stuff too. And um, it kind of reminds me of when they were first releasing um, Azure in cloud. And I remember talking to a Microsoft exec who told me, and I asked him, well, where is this? Like where will the bank be? And he goes, it doesn't matter. It could be in a container, Walmart parking lot in Tacoma. And I'm like, oh Lord, don't tell a regulator that you don't understand. They'll shut your bank down. Like, I know that you think that sounds cute, but they say take me to the place where the wires are coming out of the wall. Um, and they want you to sort of. So it's almost like we just have to be prepared for the use case. That is okay. I'm your regulator, I'm standing in your office. Take me through it. How are you making that decision and making that work? And it seems very similar to that, even though I know AI. Is it different? This is a different cycle. But there are some similarities to kind of pressure testing this in a regulatory thing.

Speaker D: Yeah. What's so interesting is there's this concept that LLMs, large language models, aren't built. They're not programmed, they're not coded, they're grown just like a plant. And so auditability and understanding how it got to be, um, how it made the inference that it made. It's not knowable, um, because they're not built, um, and that, that's going to be a real paradigm shift. It's a probabilistic inference model, not a deterministic reasoning engine. And when you're in a highly regulated industry, you need auditability. And I think adapting ourselves to a probabilistic paradigm is going to be a real shift for the industry that's used to determinism.

Speaker C: And without that determinism, you don't have the reproducibility that this industry absolutely requires. Right. If you ask the same question twice and you get different answers, there's the problem. And that's what happens right now, when you go into an LLM almost by design. That's the way the technology works. And I think one of the things that is, well, I think scary at this point, but could be exciting if people do it right, is how people are then starting to build that into agents and then using those agents. Really not having the right level of control, understanding knowledge about how that agent could behave or act, and not putting the right layers around. Going to what you were saying on Agent iq, layers around what is permissible as far as output, as far as action. Exactly. As far as hallucinations. Right. There was a great comment today about how AI systems are designed to please us. Um, and you can get into some very funny situations, as I'm sure many of us have, with LLMs being like, why did you give me that answer? How do you not know how to add 10 to this number? And that, I think, creates some really interesting and opportunities for financial institutions right now that are ready to go out there and test and do innovative things. Uh, but the risks are massive if they don't have the right tech to do it well.

Speaker B: And I remember aft Miami this, uh, time last year, and there was a session with the, uh, cto, Jack Henry and some other folks. Yeah. And. Yeah, yeah. Ben Metz. Yeah. And wait. And Wade Arnold. Yeah. Yeah. And, um, you know, they were pointing out that the right way to architect AI, to not hallucinate, is to use traditional coding and pattern matching to essentially check the answers. And, um, ultimately that's what we've seen when we, when we were first doing the document reading stuff that we have in our solution, we. The first thing we did in 2022 when ChatGPT came out was we tried some sample documents and throwing it at LLMs, and they were giving us different answers every time. I was like, you can't do math if the answers are different every time. It doesn't add up. Right. And so we kind of stumbled into. Yeah. So, yeah, we had a, we had a strategic plan to stumble into m. Do it. Doing what. What those guys were recommending and when they said that I was like, aha. Yeah. Our approach, this is why it's working, is because I didn't understand why we just went to the solution. But having pattern matching on top of LLMs to really to be able to get the power of AI but then to have the logic of traditional computer programming to check to make sure that everything's okay.

Speaker A: Good stuff. And I'm hearing marimbas in the background and I've got a couple of hotshot aft board members here that I know cannot be late for a session. I know, I know where my bread's buttered and I don't want to get in trouble here either. Um, so let's do a little speed round here. Um, and we'll let you guys jump in with whatever. I would just say, um, tell me something you think is either um, doing really well and jazzes you or is still messed up in this business and is screaming for an entrepreneur. And I'll let you pick whether to go positive or negative and you have free reign. Who wants to go first?

Speaker E: I'd like to see more representation across the board. So less packaging.

Speaker C: Mhm.

Speaker E: Like this and more packaging that's just more diverse at all levels and more lifting up from early career.

Speaker B: Yes, agree.

Speaker C: To better decision making.

Speaker E: Well and better outcome.

Speaker B: I 100% agree. I 100% agree with that. Although uh, I've got a different answer that I think we, I think we still need to. I'm going back to the old, the old standard of the core banking systems. I think that they're. It's still too hard for banks to be able to do what they need to do with their own data and we have the technology now. That problem has been solved. Uh, there are some core vendors that are easier to work with than others, but there are still some core vendors that are making things really difficult and that's holding us all back because these solutions can't work without the bank's own data. But yes, we need more representation.

Speaker C: Yeah, uh, agreed and agreed. I'll say something that's working well and not working is trust. Um, so in places where we can actually get the trust between the fintechs and the more traditional financial institutions, banks and credit unions, then we can actually have the real conversations around the real pain points, the real problems and the best way to solve those problems. I think a lot that's the positive. The negative is I think a lot of fis have been burned by fintechs that over promise and under deliver, uh, by vaporware demos that don't end up actually being real, um, by call it Tech Bros or Sisses, uh, coming in and saying they have all the answers and all the information and they don't understand the nuances of working in the regulated space, let alone financial services. So I'm seeing it cut both ways. And I really would encourage bankers out there, uh, to use your networks, use the systems you have in place to find the fintechs that you can actually trust and actually partner with. That is how you win. That's how you compete with the fintechs that are trying to eat your lunch. That's how you compete with the top 10 banks. That's how you thrive and survive. Right now.

Speaker E: M. Take us home.

Speaker D: Yeah. Uh, let me see what I can do here. So we've all read the headlines that like, entry level jobs are on the decline and the AI is going to wipe out the entry level opportunities for the software engineer. I take a contrarian view. My intuition tells me that with AI, we're in a season of transformative, disruptive change and that requires us to change within it. The challenge though, is that veterans who've been around for a long time are reluctant to change, resistant to change. Clayton Christensen taught us this with the innovators dilemma. Successful organizations become vulnerable specifically because of their success and their size. Um, my intuition tells me that entry level professionals, folks who are just graduating from school right now, who've got three years or four years of using large language models and generative AI, they're going to be the ones who, with a beginner's mindset, a sense of curiosity and intellectual humility, are going to be the ones who bring us forward and help us to imagine what's next. Um, and so we're enthusiastically offering internships to people and we've got some amazing talent we've never had access to before. Um, so that's, that's my contrarian take on the job market.

Speaker A: Love contrarian takes. And, uh, I'll just throw out a couple quick from my perspective of Pure Day ft, because I didn't weigh in on that earlier. Um, I loved all the points that you all made. I would add to that. I love Peter Gliman's, um, session on stablecoin, not even because of the topic, but because he sort of let the, the crowd marinate, get involved. I think he even said that you tinker with it.

Speaker E: You.

Speaker A: We learn by playing around with things. I love that aspect of it. So great job, Peter Garman. And then I would also say, man, we're all creators here. Um, on a good day. And it's great to be inspired by songwriters, um, sharing their most recent song. And so shout out also to Avery Payne, who is, uh, the daughter of Danny Payne, who was one of those songwriters. It's just great to see young people creating and soul. Nope. That's right. Did I say Danny was a songwriter?

Speaker E: Sentence structure. Sorry. Sorry to go back to your sentence structure on your own podcast.

Speaker A: Love it. Uh, I've just been language modeled. Um, and I love it. I love every minute. Thank you for that, by the way. Okay, so we are going to check out here from the aft spring summit with, um, another in the hall episode unscripted of fintech Hustle in the hall with industry leaders Tim Hamilton, Kirsten Longnecker, David Eads and Laura Kornhauser coming to you. Thank you so much, you guys, for spending some time with us. And back to you. See you on the road.

Speaker B: Hey there.

Speaker A: If you really dig this episode of the always unscripted FinTech Hustle podcast, hit the follow button on Apple, Google, Spotify, YouTube or wherever you jam your podcasts. And hey, tell your fintech friends more shop talk chats are coming in the hall with industry leaders. Look to see you out there on the road.