Payments in the AI Era

Every payments and fraud professional right now is asking the same question. Not, "Will AI change my job?" They know it will. The question is, "What do I do about it?"

Today, we're answering that with Julie Ferguson, who co-founded Merchant Risk Council 25 years ago, back when fraud was a backroom problem nobody wanted to own. A legend in the payments industry for those of you that know her, and along with uh Clawbach and Albert Drought, co-founders of Pagos. We're actually going to be talking about what it means to operate payments in the AI era. Julie, welcome.

You left IBM in 1995 to start a payments gateway company. You saw your first fraudulent transactions and built a fraud engine. 30 years later, you're now running the organization that connects the entire fraud and payments community globally. What would the 1995 Julie think about AI agents making purchases on behalf of consumers today?

>> She'd be super excited. I think one of the reasons I started my first company is because I hate to shop. And actually, back in 1999, the local newspaper here in Austin named me the queen of online shopping because [laughter] I wasn't going to do any shopping ever again in a store. And so it would sound kind of futuristic, you know, Jetson-like almost, but I think my younger self would have been super excited.

>> Class, did you think this would uh this was what you envisioned back then?

>> No, I don't think it was anything near what you can envision uh back then. We were expecting flying cars and we got agent commerce, I guess.

>> But I love the Jetsons reference. It feels like more and more the world is actually coming true from that episode. I don't know. The foresight there was amazing. Albert, you probably were probably the most confident in the world changing from a product and tech perspective.

>> Well, that's probably only four or five years till I was like pulling Ethernet cables and uh, you know, trying to connect hardware and stuff back then. But yeah, I mean, I guess I saw I saw kind of the the digital transactions from more of the kind of Wall Street side. So I could see you could kind of see where things were going. I do remember, this is probably dating me, but I do remember running a mailing list on behalf of somebody when I was in college and they were collecting payments in clear text through email for a conference. And I remember it dawned on me that that was probably like, wow, I just got a hundred credit cards like in an email on in plain text. But those days quickly changed. But yeah, you kind of started understanding probably not a not a great thing, but also there wasn't anybody to do anything with it back then much.

>> So, >> it was definitely pre-PCI, but PCI is fairly young in many ways. We we saw that transition for sure.

>> So, yeah, I know I remember back in the in the early 2000s, people didn't even believe uh the data breaches were happening and credit card numbers were being stolen. We had to convince the industry, hey, we really need to encrypt and protect it. And a little fact is a lot of people don't know, the PCI came out of the MRC.

>> So, you know, their first couple meetings at our at our conferences and we said, "We don't want to do security. We're fraud and payments people. Go figure it out."

>> Wow. I did not know that. Actually makes sense in a lot of ways, right? What PCI stands for.

>> It's an actually interesting history of how MRC actually started and how you co-founded this. You want to share that story?

>> Yeah, sure. Um, so back in the the late '90s, I I was running my first company and our customers were all selling online and getting attacked by the same fraudsters. And so one of our data scientists said, you know, we really should get these guys together to talk because all the fraudsters are talking. And so we actually invited HP, Apple, and Dell to lunch to say, hey, you guys should talk about fraudsters. And they were really, really skeptical. And they're like, "Oh, I don't know. They're my competition. I don't know if I'm even allowed to talk to them." But they went to lunch. They shared ideas. And that's really the moment the MRC magic was born. And from there, we set up a mailing list and it just went it went.

>> So, it started in Austin physically. That's cool.

>> Yeah. Yeah. It started in a in North Austin near Dell.

>> The foresight, huh? That's amazing. And the global community that it's become. Well, a lot's happening in that community today. AI is a hot topic for sure. So, Julie, one of the things that kind of stuck with me when we were actually having this conversation about getting together and talking about AI and operating the payments era, uh, you said AI is not replacing jobs. People using AI smartly will replace people who aren't. What does that look like right now and who's getting it right, who's falling behind?

>> Um, so anybody who's really leaning into AI but measuring to make sure that the effort and the investment is returning good results is doing really well. So I see a lot of fraud analysts using AI tools to automate some of the analysis of data to help prioritize and make decisions and queues on the payment side. The same sort of thing of using that as just another tool to take a look at the data and do smart things. Where companies have tried to replace people very quickly, they've learned, "Oh, this isn't going to work," and brought those people back and undid as quickly as possible what they had done. So, so there's definitely a balance. And I think unfortunately or fortunately, depending on how you look at it, all AI is really doing is allowing us to do a whole lot more work more efficiently. It doesn't mean our jobs are going away.

>> Well, I'm I'm curious because we run into these very large enterprise companies, well-known. Some some of them you think are quite tech-savvy from the outside in, they don't necessarily have their AI policies and plans together yet, no matter how much we talk about it in media or or MRC events. Have you found any surprises around that? I mean, that's an example. We have heard some companies like, "Hey, we don't have the policy down so we can't actually use AI tools in general, certainly not on our own data that we operate." That may also be somewhat of a function of the teams that you're talking to, presumably they don't treat fraud the same as some of the other teams, as an example. But if you flip it on the other end, is there anything that it's surprising in terms of very slow to to the market and how does that play out with engagement around MRC?

>> Yeah, absolutely. We still have MRC members in our community that are big giant companies that are like, "Oh, AI is a fad. It's going to go away." So, they're they're not even necessarily, other than reading the headlines, paying attention or making any investments. So, I think I think there's an entire spectrum. You know, there's a lot of questions around the security around it. So, more conservative companies definitely aren't embracing it and giving their employees free reign to do whatever they want. And where other companies that are maybe smaller and more tech-savvy are doing a lot. You know, we see some of our members. I love Spotify. I'm a Spotify user and it has the AI DJ now to create custom playlists and, you know, it makes my experience so much better. So, we have members that are leaning in from every aspect as well. So, I really think it depends if they're a tech-heavy retailer or a more traditional brick-and-mortar where they are in the spectrum. I also I noticed this year in Europe um there was a lot more, especially around agent commerce, there was a lot more conservativeness because regulations first in Europe and so agentic commerce doesn't really isn't really compatible with PSD2 as it stands today. So, so everybody's kind of like, "We'll watch what the crazy Americans are doing." [laughter]

>> Yeah. Exactly. But in the US, we've seen a lot of our members doing pilots and experiments, doing uh press releases saying, "Hey, we're doing this and that." So, it really I think it depends on where you are in the world. And I think it also depends on the technical expertise and the what your company leads with first. If you're a tech-first company, you're going to be leaning in, but if you're not, you're going to be a little more conservative.

>> It seems like you've been I mean, you know, I've just been following you on LinkedIn, too. It seems like you've been leaning in kind of both agentic and kind of AI. I'm just wondering I'm reflecting a little bit on my conversations. It feels I was just talking about it today even. It feels like even week to week now the pace of kind of understanding how it's going to impact or even drive efficiency or scares you and taking things away. I'm wondering where you've ended up after like the last six months of kind of riding both maybe both the agentic kind of roller coaster but maybe also just the AI one in general. Like have you found it, you know, less bouncing week to week and kind of found a place where it feels like it's really fitting in in that daily work or or things like that?

>> And even tool swapping swapping as part of that since >> quite a few company or individuals have gone through changing the tools that people are using uh when we talk about AI?

>> Yeah, no, I think it's the pace feels not quite as crazy as it did the first six months. So, you know, every morning I used to spend three hours studying all the new tools and companies and now it's probably more like half an hour, right? So, so it's the pace is definitely slowing down and there's a lot more meta instead of new innovative stuff.

>> So, I think I think you have a really good point, right? When you when you say, "Hey, is it changing?" It's absolutely changing. I have caught myself a couple of times using AI to brainstorm and improve on something and the other day I spent two hours on it and at the end of it I was like, "It's way better. Is it two hours of my time better?" Probably not. Right. So, so I think you know, we all need to to be cognizant of not going down rabbit holes.

>> Yeah. Is it a fidget spinner uh kind of thing? Like you're kind of spinning it to see what it'll generate.

>> Yeah. And it's kind of fun too, right? It's like it's,

>> You know, fun to brainstorm and improve, but so I definitely I'm seeing the balance. You know, as far as agent commerce goes, there's a small group of us hardcore nerds who are using it and programming it to order stuff for us. But at our conferences, I always ask the question to the audience, "Who in here has made an agentic purchase?" And there's still less than five people out of, you know, hundreds of people raising their hands.

>> So, you you've actually you've actually inspired me to try and I haven't. There have been a couple of times where I've been very intentional. I want I want to make this work and I couldn't get it working. So I'm I'm like, cuz I there are pockets that I think it and and ecosystems right where the the thing should work end to end but I feel like I'm being told like, "No, you can't quite do it yet." Like somehow. But I'm curious like what are where are you if you can share like where you're actually shopping a little bit now successfully I guess in the ecosystems but?

>> Yeah, so so I've done something I don't recommend and that's I've shared my user ID and password and my credit card number with my AI platform.

>> And so here's all my credentials, go to this website and buy these three things, right? And so,

>> It can accomplish that. It kind of goes in circles and it's like watching bad TV sometimes like, "No, don't. No, click there. Not there." Like you want to, but it does work. Um,

>> Oh, so you you've done the agent like truly giving it to kind of an agent to go shop, not within one of the closed more kind of closed loop or closed ecosystems. Interesting. Okay.

>> I've done some of the closed ecosystems like on Etsy and that worked. That was pretty slick. But I I honestly I I'm not a big shopper of things that are nice to have like so food and clothes and, you know, the basics. And so my favorite thing to use it for is Sunday night I'll say, "Make a weekly meal plan. Here's the days I'm actually going to be in town. I'm going to shop at H-E-B and Lakeway. So use that as your store." And I I look at what's on sale to come up with the weekly plan. And I need it to be for, you know, two people or four people depending if the kids are home or not from college. And then I say, "Okay, and I want to do minimal cooking, so I'm going to do prep on Sunday night, no more than 20 to 30 minutes. Make me the shopping list and print out the recipes and load it all in my basket for curbside pickup." And so it just spits out the recipes and I say, "Yep, that looks good." Then it loads my basket and then I say, "Yeah, confirmed." And then it goes off and does the purchase. So all I have to do is drive over and pick it up.

>> Sounds like magic.

>> Yeah, that one is awesome. And I could even, if I wasn't so thrifty, I could even have it delivered. But [laughter] I don't want to pay that five bucks.

>> I like that. In terms of tools for putting tools to the side, workflow that you've actually solved for, Julie, you mentioned last time we were talking about the meeting and record, the summarization, the prepping for the next tool that you had loved. I'm not going to mention names, but you I was actually checked it out. It's a really cool tool. Any workflow tools that you guys have used recently that have actually changed how you work that you weren't doing a year ago?

>> Yeah, the the email one that I told you about works phenomenal for me as far as prioritizing emails and it's even starting to expand its feature set to create tasks, tasks and assignments and do follow-ups automatically from the meetings. So, I think as far as running meetings and doing some project management, basic project management, we're going to see a lot of the email tools evolve um in the next six months to to do that and and keep the noise out of our email box. I mean, I'm saving hours a day by having it sorted to different buckets and only seeing the important stuff.

>> Yeah, I'm doing I'm that's one of my use cases that I have my agent like reading a lot of industry news on my behalf and then trying to distill it and then having a conversation saying, "I want more on that." But that's probably given me back time to I don't know if I'm actually smarter. I guess that's still you guys can tell me, Class, if I feel as connected now because I'm not reading it firsthand. But but yeah, it definitely seems to help.

>> You're not sharing 10 articles in our Slack channels though.

>> I know that's probably true. Yeah. Yeah. It's like,

>> Yeah. Well, some of the articles are good to read, too, though. So, it's a complicated balance. I think you lose some nuances that at least if you're experienced, that's hard to pick up if you only read summaries. I think you lose some stuff. But

>> I think,

>> that's tricky.

>> Yeah. One thing I had for Julie when we talk about you you were mentioning payments teams. You you interact with a lot of payments teams. You mentioned that the fraud analysts are using it to crunch through a lot of data. Is there other changes where one interesting aspects of payments obviously touches every team from product to engineering to finance to risk and compliance to legal, whatever. A lot of different teams. Obviously, there's a lot of talk about, you know, cursor to claw to maybe something new again sooner or later. But if we talk a little bit more closer to payments, what kind of stories have companies shared in terms of how actual work have changed or things they haven't changed as a company in the payments teams or the broader set of people that are touching payments that you have heard?

>> Yeah, one of the ones that I thought was really neat is um one company I was talking to talked about how writing the requirements documents, they're using AI to write the requirements documents, but then they also say, "Okay, and take this pool of data and validate [clears throat] the feature set that we're building and is it going to work with this and and provide me the analysis and provide it at the technical level, but then also summarize it at the executive level so they can go off to get approval for their budget." But they're using it not just to write, you know, help write that document and edit that document. They're also using it to go off and do that analysis that normally you would have a a data analyst person go off and do. And because the AI engine is smart enough to read it as well as ingest the data and do that analysis, they said they're saving a lot of time. They still have a human review everything for accuracy, but that I thought that was pretty neat. And the other thing they said is they also say, "Okay, what are the security risks? Pretend you're, you know, pretend you're a security guy. What are the security risks we need to consider? And then pretend you're the legal guy. What are the legal, privacy, and compliance things we need to think about?" And so they said very quickly they can build something very solid to present as far as product enhancements or or or feature requests. And uh some the one company I was talking to, they had just done it for potentially adding a new payment type in a different country. And so they didn't know anything about it, but they were able to educate the AI platform they were using and and link it all in and get an answer pretty quickly.

>> Do you feel like the leverage is creating right now is speeding up traditional work or enabling new work for the same role? Like if I was an analyst before, let's say I had to write a requirements document, but I had to do a deep analysis in a market. Is it just speeding up that process or is that being done faster and now they're able to focus on incremental work that they weren't able to get to in the past?

>> I think it's too early to tell. I don't know. Class and Albert, what do you guys think?

>> Yeah, I think exper I think you mentioned a little bit. I think the more experienced you are, the more you'll get done because you have a long list of things that you've been thinking about. So that's the way I perceive a lot of it. But the less the less of that kind of to-do list you have, I think the more scary it is because if you're being told this is what you should do and then you do it in 2 minutes and then you're like, "Uh oh." Maybe that curiosity. I mean, of course, there are people who are super curious and who dive into like the next question or they want to they want to push it. But I think it it exposes your your almost your personality a little bit in terms of that. Do you have the curiosity in the question? But that's that's kind of what I I guess it's too I don't know the the models and the the tools may be prompting the users too like, "You should probably think about putting this on your list next, right?" That's that's how we like to think about it with what the stuff we're building too like some users just don't know what the next step is so you can help them with that. Some people have a long list.

>> But the it is an interesting question though. I mean, is it are we all just doing more and better and more stuff on the list every day that's incremental or I mean, ignoring that you may be using tools or stuff that are new for you, is there anyone doing something radically different as a function of AI?

>> I don't know if we stay in our payment space.

>> Yeah, within the payment space and even within the, you know, the workspace, I think there are people who are doing new things because they have a little bit more time. But when you do those new things, you impact other teams. That's why I I don't really know if we're going to get the big lifts that everybody is projecting, right? Or they're predicting. I I

>> I think you I was going to say you said something that I think we run into just in our use kind of at Pogos, which is like you if you do you do need to hit like being trusting or you need customers to trust you, then you need a human review of some kind. You run out human boundary is the number of hours like Yeah. You know, and then all of a sudden there's a backlog of pull requests or projects that, you know, and then you're still hitting up on boundaries. There's like system boundaries. There's the Visa Mastercard rules. Like it doesn't really matter what you can creatively come up with. You're still going to have to reconcile it. So, it's interesting. You probably hit those boundaries faster, too, in some cases. Um,

>> what about fraud use cases as we talk a lot about more sophisticated fraud, but is anyone doing or have been running into that people doing a very new way of of attacking them for, you know, doing fraudulent transactions and getting away with it? We obviously there's some out there, especially on LinkedIn around identity and the fake and it's harder to identify people. So that's one segment. But if we speak about fraud more broadly on the MRC community, where people just getting hit with things that they hadn't envision before or run into before?

>> Yeah. No, it it is becoming harder and harder for our members to tell the difference between a good transaction and a bad transaction. And it's because of these AI tools. It has become so sophisticated and so perfect that the fraud tools can be programmed to act just like a human. So instead of, you know, systematically clicking, it can add in the randomness and it can add in the flipping back and forth of the pages. And so a bunch of things that our members have always been able to track to say, "Yep, that's that's a human doing that." The fraudsters are recording it and doing replays and they're able to build in that unpredictability of a human. So, it's just getting super hard. Just like the phishing emails have evolved and it's almost impossible to tell a legit email from a phishing email. My bank sent me an email and I honestly I called them and I said, "I don't know if it's legit or not. I can't tell." And it turned out that [laughter]

>> I'm like, there's a lot of class out there. It's interesting you Jul you kind of highlighted this. The verification layer doesn't necessarily go away with AI, but the complexity increases. So, how do you think the verification layer actually gets redefined in this with human and loop?

>> Yeah, I think we're going to have to be smarter. I've been commenting now for the last couple of months, my job is so much harder than it was before AI because all day long I'm only doing the hard stuff. All the easy stuff is gone. And I think as people as it becomes progressively more adopted and used, I think that's going to be the biggest problem is is how do we get the brain capacity and then everybody is talking about what do we do about bringing up the younger generation. You know, that one's a tough one.

>> Yeah, it's almost uh we're all we all learn on the job and what's the job they're walking into and how are they going to build that expertise over time. Any concerns because it's an interesting point that you brought up with the younger generation. Will they over-rely on AI and not ask the next question? Like what's the risk there?

>> I think it's huge. Um, you know, we we've we've been talking at MRC about what what can we do for the young professionals um and make sure that they're have a good career path and they're learning the right skills, curiosity, testing, challenging, all those things that we've always done. I was hanging around with a group of young folks this last week and I was surprised how many people when they were having a conversation would just go straight to to one of their AI platforms on their phone to ask the question rather than thinking about it, you know, or searching on Google or, you know, they just took and they took the answer as fact. And I still I'm still very skeptical. So I always try to, you know, have it cite the sources because I I don't trust it all the time. So, it's not there yet.

>> You're still in the verify and trust mode. Uh, Albert, you?

>> Yeah. I mean, I was thinking I think it required like you I think about how I learned and the pain of not of being uncertain and not knowing, but there was no alternative. So, like your manager was kind of like, "You're the you're the guy that has to deal with this now and go figure it out." And you had a you had like weeks and months and quarters to do it. So, I think it's and just thinking about how people learn. We're going to have to be super intentional now in a way that like, "Okay, this is a new person. The good thing is like they should use AI to not to get a lot of their questions answered, but they shouldn't they but they have to actually do some of the reps and we're going to have to figure out the balance." But yeah, it's going to be tricky. It's going to be super tricky. At the same time, what's cool might be that people learn faster and get a bit more of that support, right? Like if you came into payments and like how many times is the first question about BINs? How many times is it like interchange? Like, "What is interchange?" Like, "Okay, well go and spend an hour with Claude and have them walk you through the, you know, three-party, four-party models that are out there and like have them, you know, you can have a dialogue with it." And that's amazing. Like my kids can probably learn about payments now, you know, as opposed to anything else. But but the danger is then do do you want to wait around for somebody to learn or I don't know. It's going to be that's it's going to have to be intentional. We'll have to choose something. That's that's what I think at least. We'll see.

>> Well, Julie, um, in our conversation, you said educating yourself on AI is more than a Google search. And that really stuck with me and I've actually been using that with my kids, by the way [laughter] on this. Uh, they've literally just taken ChatGPT and Claude and turned that into the nicer interface for search, but they haven't gone one step further. So, how do you educate yourself with such simplicity of tools of asking that question and and trusting it? What what's the change?

>> Yeah. So, I think all that stuff we learned about credible sources when we were in high school and college of, you know, how how do you know what you're you're reading is true? I think when we ask questions of an AI platform, we need to be asking those questions as well. So, you know, what are the sources that you're using? Where did you draw this conclusion? And I know when I'm teaching people how to do AI prompting, I even talk about use use AI to create your prompt. Say, "What's missing from this?" Because most of my AI prompts are probably, I don't know, a page worth of text. But that's because I take time to really write it out and think about it. And so I think that's the challenge of when you ask the question, you have to think about it as a researcher and not just as a as a person of, "Hey, what time is this movie?" Right? It's like, "Hey, what time is this movie and what snacks do they have and what be what will the traffic be like? Which is the best theater to go to?" Right? Those are the kinds of rabbit holes I go down when I ask when I ask for some information.

>> That's interesting. Credible sources. My son always has a list of credible sources. I'm like, "How much did you actually go group and click through and read the whole source versus just take the summary of it?" One area I'd love to get your take on uh because you've spoken with a lot of payment and fraud professionals and continuously do this and travel around the world and everybody's talking about a lot of evolution of what the role is going to be over time and the skills needed to be successful in that role in the AI era. What what do you think is the role three years from now? Payments and fraud analyst role three years from now look like?

>> I think they're going to have to be curious and strategic. I think that's the that's one of the big differences. I think we're seeing everybody payments become more important in companies and organizations. It's it's some companies it's even has a seat at the C-suite table. But it's people are going to have to be very strategic about what they're doing when they're making those decisions and and not just rinse, lather, repeat, right? They're going to have to be thinking, "How can I do this better?" and constantly thinking about the evolution of their role.

>> Claus, any thoughts? You've seen all different types of role evolution.

>> Yeah, I was just thinking about you were asking about fraud analysts and payments analysts, but what about the head of payments or the head of fraud? How are their roles changed and are they all going to keep it up? As an example, some people there's always a percentage of the market, you know, that from when you were building your startups. They don't necessarily like a better for tools where there's less people involved because their self-worth is tied to how many people work in their team. So, how do you manage a world where you have fewer people that do a lot more, but it's not tied to how many employees are in the team? And and in fairness to the individuals, many companies have bad compensation and promotional schemes that are actually tying the chances of getting more money and getting a bigger role tied to how many people you have managed before. So, I'd probably think about a little bit more like like that. Capel, I don't know if if Julie, if you have any thoughts on on that sort of one level up and then I'll ask another question about one level above the payments team, but we'll get back to that. So,

>> it's interesting because I do spend a lot of time when I'm doing mentoring for folks talking about being an individual contributor versus being a leader. And I wonder as you're talking about it if we're going to see more individual contributors but getting compensated at a leadership level, right? Are we going to start to see that that playing field leveled? Because the the really strong amazing individual contributors often make as much as the leader of the team because they're so important. They produce so much great great worth to the company, great value. But um, you're right. I the for the people who are climbing the ladder and have their checklist of, "Yep, I'm now a manager, yep, now I'm a director," and you know, wanting to go down that path. I bet the path is going to change because of AI.

>> Yeah, I feel like the ladder has been cut short a lot. There's no concept of it either in that regards. I love the comment that you made about individual contributors even on as more senior you go and and focus on the output. One thing um we kind of talked about was around merging of payments teams more broadly where it traditionally was sitting maybe into a finance function or an ops function and now potentially in the cyber organization. Can you talk a little bit about where you've seen it in the space and uh where you think that's evolving to?

>> Yeah, I think the lines between fraud and security are getting really really blurred because the fraudsters are leveraging security holes and the security hackers are leveraging fraud holes and social engineering and customer service teams. And so we've already seen a couple of companies move their fraud teams under the cybersecurity department in the companies and and I think that's where we're headed.

>> Got it. So payments and fraud potentially going in separate directions in an organization.

>> Yeah, I think so.

>> Okay. So then that's how like this Chief Payment Officer almost stands alone with payments and then the CISO officer actually owns uh fraud and risk along with overall cyber.

>> Yeah. Yeah. And it and it's partly it's because payment is becoming almost a consumer preference, right? So how how do I want to pay and where do I want to pay is really a strategic thing companies need to figure out and understand, especially as they go global and their online offline kiosk, you know, we can buy everywhere now. So I payments is a lot harder job than just having a little point of sale terminal and swiping a card.

>> Yeah, no, I agree. Any thoughts Albert on just the infrastructure changes uh as a result of this kind of evolution?

>> I mean, yeah, I think I mean I think when you have a world of of uh powerful individual contributors running their own teams essentially, I think all of a sudden the demand I mean I think what we think about is like the demands on all your infrastructure just went up. Could be data, or it will be data, it will be, you know, a lot of the communication is going to probably be happening faster. You're have to articulate more clearly. You're going to be working with people who have perfect memory. All of a sudden, "It's how do you know that? I didn't say that two meetings ago." So, I think yeah, I mean I think I think of it as a demand on all of the systems and then the h the hunger that everyone's going to have to get facts and context in. So you better have all that stuff saved and saving and the loops. I think we think about that at Pogos too, right? Like what are the loops that customers are trying to run now where what you did yesterday should be coming into today. It's now kind of that was always true, but now it's even it's even more of a requirement and it's also an expectation and it's probably a require you know, it's how people are going to shine and be even better, right? So, yeah, a lot of I think every every nook and cranny of a of a company is going to have data that's probably worth using and people are going to be looking for it as a kind of as an input. So,

>> that kind of trigger a thought. Curious what you think about Julie, but I I always felt that payments is quite opaque and often from a vendor perspective, not trying to step on anyone's toes here, but it's almost intentional. It's very complicated to understand your cost. People are nickel and diming you like crazy in terms of creating very complicated fee structures or explaining what they actually are and how they work and how you can change them and what happens if you manage the tool or not manage the tool. Are the buyers, if we can call them that, going to be more informed finally in payments and push back on some of the opaqueness that's been going on and some of the places where people make money that even very sophisticated merchants are unaware of like FX or margins or FX I should say, and things like that. What's your thought tied to that?

>> 100%. And I think we're going to see a lot more analysis on the costs and people are a lot of merchants are already using AI to try and understand that. I always say, "Understand my payment bill," right? That's a very hard thing to do. There's companies popping up left and right that are AI-based that claim they can do that as well. But that is something that people have wanted to unscramble for a long time and and I am already seeing AI start to start to peel that away. It's it's going to be complicated because the bills really are incredibly complex and there's I'm not even sure the people who are creating the bill understand them completely. So, um [laughter] it'll be it'll be interesting to see to see how it ends up, but um absolutely that's that is a hot topic that people talk about all the time.

>> I was surprised just yesterday. I have to say it. Sorry, because we're on I was looking at something yesterday and it's like, "Okay, the cart type is credit, but the interchange on a refund is debit." Like, what? Like, does that actually make sense? So then you're like, it kind of does, but like you see this stuff in reporting and you're like, "Wait, so is it a is it credit or is it debit?" Right. And then of course there's always a bunch of data points around where these things can happen. But anyway, yeah, you're you're right. It gets complicated. Yeah, we did an economics of an e-commerce transaction workshop and at MRC and one of the assignments was, "Okay, figure out the most cost-efficient way to route this transaction." And we were all given this assignment and I was at a table with two of the smartest merchants in our industry. One works for Sony, one works for Google. And I was like, "This is going to be a home run. We're going to win." We we spent a half an hour and still didn't get the right answer. I mean, it it's that hard because is it which is the processor? Is it going to hit the debit rails or the credit rails? And there's just so many factors that that come into play. It is incredibly difficult.

>> And so, thank goodness there's technology to do it, right? [laughter]

>> I think uh we had that at one of our off-sites. You made me think of it. It's one of the things I want to use AI for is create those training games. The role-playing the payment network game where you take turns and you're sitting at a table with your friends and trying to be the best issuer you can be and minimizing and maximizing risks and revenue. Yeah, it's exactly that. But that's the

>> Yeah, that would be we have that we've had this idea that at Pogos that we might uh we might have to create a a kind of game that the industry could use and you could sign up and be massively multiplayer and now that Claude could just probably write that in an afternoon. I think uh I think about making that real. Yeah, we we hired a governance consultant to come in and, you know, help us shape because MRC's tripled in in revenue since I since I got here and

>> and so we need,

>> there's pros and cons to that, right? So, [laughter] so what do we need to change and grow as a company? And I was like, "I wonder what the governance consultant is going to ask me." So, I I turned on my AI platform. I said, "Okay, pretend you're a governance consultant coming into the MRC. Here's our URLs and ask me questions." And it I put it on the audio mode and I had it ask me and then it analyzed every answer I gave and coached me on how I should answer it. Now, luckily, the governance consultant is way nicer than the AI platform. [laughter] It was really fun to do that to set that up and and have that conversation. I'd love to do it as a fraudster and as an issuer and there's all kinds of possibilities.

>> When besides companies growing larger, but what is the tipping point for new companies to join MRC? Where does it come from? Is it all over the place or do they get burned by fraud or what what drives new companies to come to MRC?

>> For the merchants, I think the first time they get hit by fraud or if they've grown to a certain size, usually around $10 million in in annual revenue and they they're trying to figure out their bill and where they can save money, that's that's usually when they start searching on the internet and find us. A related question that goes above the payment seems and the folks that goes to MRC. I I think one of the surprises for someone who's been in the payment space for a long time with the Polish journey is that I I think you have great companies with great payments teams and they do things. Some are better than others, of course. Some are more resourced than others. Some have more complicated worlds to operate in than others. But then there are many, including very large companies, that do literally nothing. And then I I think there are a bunch of people that are only reactive. So the the last two basically management doesn't have expectations or demands on metrics and moving the metrics to something better. And and the the last two of those that is the majority of the market. So you would imagine that MRC could be 25 times larger in reality as a functional batch. What is it that drives that some companies have payments teams where they actually operate with payments as part of their core strategy? I think is it again that cost and fraud has been triggering them to take it serious or is it something else?

>> It's either the cost of fraud or the the customer experience or somebody needs to grow their business and they've identified payments is possibly one of those areas they can grow their business just by making some operational changes. I I think those are those are probably the biggest ones. I always say we have a handful every year of yo-yo members where a merchant says, "Oh, I've got everything sorted out. I don't need you anymore." and they don't always come back. Very, very rarely does the merchant actually leave the MRC and never come back because stuff is constantly changing in our industry. Fraud attacks are constantly changing. And if you're not staying current, you will have some bad either fee increase you didn't know about because of a of a regulatory change or a mandate that from the card networks that changed or you're going to get hit with a new fraud pattern that you didn't see coming. So that's that's kind of the trigger to keep people engaged is our industry is always changing.

>> And and on that note, I I maybe we say this every year, but I I truly believe it like it is getting more complicated. A lot more things are happening faster in payments than than it ever has. And even if you take something for example with fees, the fees are getting very complicated. I mean, I think a lot of people don't even realize like, "Hey, the day when you send and don't send to your payment processors actually drive the outcome of both cost and and sales for you and you don't take that serious." But I think the fee and the structure and what you actually need to do to optimize the fees are getting incredibly complicated here in the last couple years in a way that they were not before, even though the statement was true that you got to be on top of it. So,

>> yeah. No, absolutely. It's it I think one of our one of our members shared with us that there are thousands of card network updates every year and so you literally there's more than one a day. So how do you even read those, digest those and keep up with those and you're relying on your vendors to to integrate those and and support the new features or you're going to get dinged with fees. You know, there's been a lot of new features that have rolled out that everybody's paying for that nobody's using yet. So, it's just a lot to keep up with and understand. And and you are absolutely right, the pace is so fast. It's it's just as fast as AI changes are going.

Yeah, maybe some people who generate the the change is actually using AI to come up with sophisticated models for how things like cost are changing. Who knows?

Folks, we're coming towards the end. We covered a lot of ground. We got a fun round to close this out. I'm going to play the host. I'm going to ask you guys questions. You get to say real talk or hype and more than welcome to give a sentence afterwards and each one of you get to answer it. Okay? Doesn't matter who goes first. Some some controversial, some easy. Okay?

Most of what's being sold in AI and payments right now is just rules engine with better marketing. Hype or real talk?

We won't talk for sure. For sure. I think both the people are selling some of them are doing very basic things and doing something very basic is better than not doing it. So they're not wrong. But it is not sophisticated. It's often not really datadriven beyond the fact if A fails I'll I'll do B. But I also think a lot of the buyers are not data driven. So they take it on face value and no way of of understanding that. So very long sentence but yeah for sure that is true.

Yeah. Does that mean we say hype or does that mean we say real talk?

Real talk. That was real talk. If you agree with that.

Yeah. I I I I would say I think I would just add because I tend to agree. I think that the other question that a lot of people aren't asking is that for all of the perceived complexity is the value delivering too. So I think that's I think sometimes it's a little bit of what we were talking about. the tool can be quite seductive and you think complexity brings value but it doesn't always so I think I think being careful about that and that maybe some measure of efficiency it's like token maxing to token minimizing like something like that are you really maximizing is is the return of all of that worth it I think is worth investigating in that space right so

there's a lot of overstatements because we have been doing AI quote you know AI for for 20 plus years, but there's some really cool new innovative things emerging, too. So, so I think it just depends. I would say probably there's majority more fluff and Barney words right now, but there's definitely some cool new stuff coming out, too.

Cool. All right, next one. A payments professional who's great with AI, but mediocre at payments will outperform a payment expert who ignores AI.

I think that's real talk, unfortunately.

I think so, too. Real talk. Yeah.

but that's probably true about everything. The curious ones go and learn. Now they just learn a lot faster than the people are not curious and don't learn. And on that note, I think there a lot of people are great on talking themselves into payments why they don't need to do XYZ and they just delay it down the road. Card schemes come knock on the door, whatever it may be. Right.

Yeah. Yeah. Good thinking. I think they're finding that in general with, you know, thinking about LLMs. good thinking translates to many domains. So.

and Judy agree that the the models help. So, but yeah.

this one's the easy one. I saw this on uh Twitter formally. I'm just going to call it Twitter from here on out. Uh X [laughter] not working for me. Uh but AI will replace fraud analysts within three years.

That's hype.

No way.

Yeah, I would agree. Like that seems very unrealistic.

That was the biggest hype I've seen in a long time. [laughter] I wanted to call him out on it, but uh just I think the fraud analyst role kind of gets redefined obviously with tooling and what the scope and mandate is especially if they're going to get merged into a different organization. But I think it might be they're doing more work not less work uh than they are today uh is where I land that.

This one I really think because all of us have lived through this uh and I think this last one will be a good one to forecast. So growth in aentic transactions will outpace tap and pay in terms of adoption.

On what time frame? [laughter]

Tap and pay has been gone.

I'm really like this is the one that I really like the consumer behavior side of tap and pay has been such an amazing evolution to watch and now we got this big rush of agentic coming in.

But is it just going to be the hype and adoption to the lags? like what's what's the reality? Tap and pay is owned by everybody these days. I mean, everybody just taps and pays.

I'm going to I'm going to go on a limb and say there's a lot of definition problem, I guess, but and things can change. I I'm probably going to lean towards hype a little bit just because of how shopping is an activity as well as a thing you're trying to. So, can it that doesn't mean the value won't potentially be higher in some it's not going to be even like it'll be lumpy in some kinds of shopping. I think Julia, your example is really an example of one that probably it would flip the other way. But overall, I don't know. I think about I think I only think about my kids when you ask that question, by the way, because I don't like my shopping's boring now, too. And I think about how they shop, and I think I don't think they're going to give up the act of shopping.

I mean, you can probably do a lot to uh discover. Sorry, Julie. So, I think that's also part of it. I think there's a lot of room to do a lot of things for discovery uh with AI and that's probably more exciting than payments in reality for the broader respect but then some places you just want to automate it so you don't have to deal with it but that could be true with subscriptions on Amazon too for certain products.

I think for the stuff that we don't like to shop for and I'm always just going to remember when my kids were little and I was a busy mom right so one of my neighbors called me the other day and goes oh my gosh I was just assigned snack mom for the soccer team. I got to do three day three games of snacks and it has to have a theme. I told her to come over. I said, "Shop." I told we told Chhat, "Shop on Target. Come up with the list. Come up with the themes and here's the allergies and it needs to be shelf stable. She needs to buy it today and package it up for the next three games." And it spit it all out within seconds, right? And she said, "Okay, where's the button to have it delivered?" And that isn't there yet. So, I think for things that the busy parent or or the busy person has to buy in a crunch, it's a no-brainer to outsource that to an agentic agent, but we have to have the security, the infrastructure, and the the retail storefronts need to step up, right? So, so you know, a couple of sites are using AI and making the shopping experience better, but most of them it's still super clunky and it doesn't help. So, we've got a ways to go. Probably a decade out.

I would imagine the like the small business like bakery, why would they want to go to Costco, too? Like that keeps coming to my mind. Like the B2B cases will also get programmed, right? Cuz it knows what your payments and your sales are coming in and what demand for what products. At the same time, it knows what ingredients you need like that. I imagine who wants to do that by hand or even in Excel, right? That will probably flip pretty quickly. I would imagine. But

oh, along those lines, do you think it's a threat to companies that have optimized keeping their shelves stocked properly and fulfilled like that Walmart's bestin-class of the shipping and distribution? So, does it start to level the playing field?

Are there boundaries that I'm not smart enough to know? I don't know. But I would imagine it they're also going to be getting smarter with those advantages today. So, I wonder if it compounds for them and then it's like that maybe [laughter] the manufacturers are squeezed into the last penny or something like that. But yeah, it's a good I think like small businesses might be more efficient faster. That might happen because you don't have you don't have as much waste, right? Because you know how much flour you need or coffee or like you're not guessing. But then again, we've seen the anthropic what model running a soda machine or there's that famous story. I'm sure it'll get better, but we'll see. Yeah.

Yeah. It's a good one.

Yeah. I I the Walmart logistics build out. I could only imagine what that's going to look like as they kind of try to time the shelf space and foot traffic and seasonality and product availability and disruption and shipping so much data.

God, this is an amazing problem to kind of uh hone in on. Well, folks, we covered a lot of ground. Julie, it's always great talking to you. Thanks for sharing. I think we could all agree payments and fraud. It's going to evolve. The roles are going to evolve. What we end up doing 3 years from now may look different, but the skill you build is going to be everything and what your job looks like from there. The thing I I really kind of think about is people building that muscle now are the ones who are going to be running payments in two years. So, thank you all. This has been a fun conversation.

and uh appreciate you all for keeping it real. This is Real Talk Payments. See you next time.

Thanks for tuning in to Real [music] Talk Payments. To learn more, visit pogos.ai.