Fighting Fraud with Tamas Kadar, Co-Founder & CEO of SEON | Episode 497

Fraud doesn’t usually announce itself with a flashing warning sign. It shows up as a chargeback, a fake account that looks “normal,” or an account takeover that slips through the exact same checkout flow your best customers use. Greg Myers sits down with Tamas Kadar, Co-Founder and CEO of SEON, to unpack how modern fraud actually works and how digital businesses can protect revenue without burying users under friction.

Tamas shares the origin story that started with a real loss: a crypto checkout experiment that got hit by fraud almost immediately. That experience turned into years of studying how fraudsters operate and, eventually, into SEON’s mission: help businesses prevent fraud, verify identities, and stay compliant in real time using the minimum data points companies already collect, like an email address or phone number, plus hard-to-fake device and digital footprint signals. We dig into when step-up verification makes sense, how to reduce false positives, and why trust and safety teams deserve to be seen as revenue drivers, not cost centers.

The conversation goes deep on AI in fraud prevention beyond the buzzwords. Tamas explains where classic machine learning helps, where it breaks, and how LLMs can speed up investigations by summarizing cases, surfacing patterns earlier, and reducing the “five tabs per investigation” problem. We also explore the shift toward headless software, where analysts can ask questions in natural language and get answers from the system of record without clicking through a UI, while still keeping decisions explainable with human-readable rules.

We close with what’s next: synthetic identities, deepfakes, account takeover, stablecoins and changing payment rails, plus the rise of agentic commerce where good agents and bad bots can blend into the same traffic.

Transcript

Welcome And Guest Introduction

SPEAKER_00 0:00

Welcome to the Leaders in Payments Podcast, where we talk to sea level leaders from across the payments landscape. We'll be discussing the products and services that impact the payment space today, as well as trends and predictions for the future of payments. We will also hear stories from our guests about their journeys to the top.

SPEAKER_02 0:18

Hello, everyone, and welcome to the Leaders in Payments Podcast. I'm your host, Greg Myers, and today's special guest is Tomas Kadar, the co-founder and CEO of Sion. Tomas, thank you so much for being here and welcome to the show.

SPEAKER_01 0:28

Thank you so much, Greg, for having me.

From Hungary To Austin

SPEAKER_02 0:30

Before we dive into your career and the company, can you give us a quick snapshot of your personal background, maybe where you grew up, where you call home today, a few things like that?

SPEAKER_01 0:37

Yeah, of course. Happy to. So I was born in Hungary and I grew up there. I have started Sion right at the last year of university. And then I was moving around in Europe. I lived in Malta and London for a couple of years. And recently, about 18 months ago, I have relocated to Austin, Texas. So it's hard to say what I'm calling home today because I do split my time between Europe and US. So maybe somewhere in the Atlantic, maybe?

SPEAKER_02 1:08

Well, I'm in Dallas, so we're both in Texas. So there you go.

SPEAKER_01 1:12

Yeah, not too far.

SPEAKER_02 1:13

Yeah. So if you don't mind, can you walk us through your professional journey and maybe how and why you started the company?

The Fraud Loss That Sparked Sion

SPEAKER_01 1:19

Of course. The reason why we have started the company is we had a negative experience related to fraud ourselves with my co-founder. Both him and myself were really interested in crypto. We just saw that every crypto exchange about 12 years ago, they were forcing customers to go through a frictionful KYC process. And also the only payment method that they were accepting was bank transfers, which took about three to four days back then to get processed. So we thought that, hey, why don't we just create a landing page, put a checkout button on it, and try to immediately charge cards and offer crypto in exchange for the funds, the fiat currency. So we did that. We have immediately lost within the first week half of our revenue, a couple of thousands of dollars. And we have never heard about fraud or cybercrime or chargebacks. So it was an eye-opening moment. And we've spent the entire summer investigating and really diving deep into different darknet forums to understand how forsters are operating, what tools, methods, schemes they are using, how they obtain personal information, how they actually use that information. And then we had realized that okay, it does seem to be a problem that can be solved. And when we were like trying to look at the solutions on the market, we saw that most of them are really enterprise focused. They were not able to cater to such an early stage business like ours. Also, they were missing a number of the product features that we thought could be pretty useful in order to combat fraudulent events. Eventually, we have started to build an in-house solution and later pivoted to launch Sion in 2017.

What SEON Actually Does

SPEAKER_02 2:58

Well, let's talk about Sion. So tell our audience exactly what Sion does.

SPEAKER_01 3:02

Sion is building an AI command center to have businesses to prevent fraud, verify identities, and stay compliant real time.

SPEAKER_02 3:11

What would you say is the biggest challenge that your company is solving for your customers right now?

SPEAKER_01 3:16

The biggest challenge is that every online business is struggling with verifying identities based on the minimum and friction-free collectible data points, such as an email address or a phone number, which every business collects for contact information. Maybe that's all. That's all that they've got. They have to assess risk based on that. And they have to decide what they do about the customer or transaction and the specific action they take real time, right? So without an in-house solution, without using a tool like ours, they would be in the dark. They don't know whether these customers are bots or real people or maybe real people using stolen identities or stolen payment instruments. And we help them to actually assess the risk based on these minimum collectible data points. We provide a number of really high quality signals based on someone's digital footprint, which actually is done based on looking at the email address. And also we collect a number of signals from the device and then push it through a real-time algorithm, which then provides a score, which indicates how suspicious these transactions and customers could be. And also now we have them to prevent money laundering, and as well as, if in case needed, to do sort of like a step up verification, which can be document verification, selfie-based Lightness verification, or even two FA-based authentication methods. So this enables businesses to make the right decisions, the right time, ensure that customer experience remains the highest and churn will get as low as possible.

SPEAKER_02 4:54

Is there a certain kind of business or size of business that you focus on?

SPEAKER_01 4:59

The size of business who we usually work with is varying a lot. So I would say that most of our customers are mainly in the SMB sector. Some of them are larger enterprise type of clients, but almost all of them are nimble digital businesses. So in most cases, they don't really have a physical store or a brick and mortar shop. These businesses grew very fast during COVID and now are also being challenged due to the AI evolution that we all live through. So the commodity is really around like, hey, these businesses are onboarding customers online and letting them to transact through the platform. Size-wise, unless they are losing a couple of thousand every month, that might be not the best fit for us, just given like it does require some assistance from a human perspective to work with our product. But also at the same time, we work with businesses who are valued in the tens of billions ranges, or top 10 clients are almost all of them in that scale. We have some smaller clients, we have more than 5,000 businesses today. And in terms of the partnership levels, a couple of thousands of them are coming from partnerships, so like indirectly using or to like payment gateways or payment service providers. So not all of them are directly integrated to your API, some of them going through a

Competing In A Noisy Fraud Market

SPEAKER_01 6:21

middleman.

SPEAKER_02 6:21

What would you say differentiate you guys from your competitors out there?

SPEAKER_01 6:26

Our space is pretty noisy, I would say. There are certain pockets of different vendors. I would say some of them are so-called like data sources or data vendors. So they're usually only good at doing one thing, but they try to do it very well and try to be best in class. Let's say email analysis or device fingerprinting or EML data. Well, all of them have to be plugged into a decision maker, right? Like an orchestrator. Because they only provide certain attributes, they don't really provide the scoring. Many businesses are struggling to plug six, seven of them into an orchestrator which could be in-house or third party. There are, of course, orchestrators who might not be owning these data sources, so they go to these best-in-class data sources and they provide a decisioning layer, they provide orchestration engine. They usually are more expensive, so would be a better fit for SMB and smaller scale businesses. Large enterprises, large financial institutions are not really interested in using orchestrators because they can build that all in-house. And there are some of the niche players in our space. These players are usually focused on one use case or one vertical. This could be either just chargeback prevention, or could be onboarding as a keyvice vendor, or could be as well as beer biometrics provider for banks. No one really has emerged space as the category leader. Respectfully, we aim to be the first

What AI Fraud Fighting Means

SPEAKER_01 7:50

one.

SPEAKER_02 7:50

Most of the conversations that I have where we talk about AI and you've kind of mentioned it a couple of times, the immediate answer is we use it for fraud. And without double-clicking on that with them, I kind of leave it there. But tell me what that really means when they say, or when you say you're using AI to fight fraud, the first thing that comes to mind is people say, oh, the fraudsters are using it, so we're kind of always chasing the fraudsters. Maybe give us the inside story of AI and how you're using it. What does it really mean to fight fraud with AI?

SPEAKER_01 8:21

A subset of AI is machine learning. So from day one, we have built machine learning-based algorithms, supervised and unsupervised types, but now with LLMs, the game has changed. Machine learning is really good to actually assess and classify certain outcomes based on historical events. But forsters are also smart enough not to try for too long with the same patterns. If they see that they are blocked and their attempt is not going through, they will just change. And if that surfaces a new pattern which wasn't seen before, and almost entirely in most cases, that's what's happening, mushrooming is not going to be good enough. That's why the human intuition was always pretty important in these processes. Humans have been reviewing transactions, assessing the genuinity of the customers. They've been spending long hours analyzing these customer transactions and finding matching patterns. And then they were always trying to make the most accurate changes in their algorithm, which could be a rule set or some sort of data science-based model. And humans were leading those processes. Now with LLMs, what we have achieved and what we actually use it for, multiple layers of certain productivity gains. So on one hand, we have a UI, right? And on this UI, our customers or end users, we have 7,000 daily active users who are all day spending their time analyzing transactions, finding those common patterns, and then tailoring and fine-tuning the rule sets in order to have the right algorithm in place. Now, with some of the product improvements, we have achieved 70% gain in terms of how much time less I have to spend on one review. We have started to surface insights much earlier. In the flow, we have been shadowing our clients actually for years now to see okay, which pages they spend the most time on, what kind of repetitive actions they take, how they conduct an investigation, what can lead to making changes in the algorithm, what kind of insights I pull in. So we have some improvements to make their life easier, essentially, just on the UI. LMs are very good in summarizing long text or essentially providing some shortcuts in order to make them more efficient and effective in their daily operations. The second layer to it, and this is going from like manual processes to semi-autonomous and then eventually like fully autonomous processes, which still would be supervised by humans. But second layer is there is this evolution also in how tech workers and people who are using softwares are interacting with the tools they're using on a daily basis. More and more people are spending more and more time on using one of these AI tools, let it be a cloud, ChatGPT, Gemini 9. What we have seen is that they've been exporting data from the system and then sending to one of these tools, doing some analysis, going back to your tool, and maybe they're using another tool, another tab, and they might end up using like an AI two and five different tabs for one investigation, right? Because maybe they have data in their CRM, maybe data in another backend tool, maybe data in Sion, maybe data in the AI tool they've been using. So the idea is really to turn Sion into what's being called headless now. And I think this is the future of software. How you can talk to the software's brain without actually opening an interface, right? Like as you were like prompting your AI tool, you should be able to pull the data from your system of a record and then ask the questions without clicking through the UI. So you might be able to click through the UI to get an answer for your question, but it might take five to ten minutes in some cases when you conduct complex investigations. So we are turning that database, the knowledge, and the certain access pathways through the UI into immediately answerable responses through human-based questions. And that's why we have decided to turn CON into a headless solution, which means that we are opening up MCPs, so modal context protocols, in order to plug our software into your AI tool and then simulate the work a human analyst will do to a large extent, which would save a ton of time. It doesn't mean that the human work will go away, but it means that they can be elevated. We can help them to 5x their productivity, which means that if it took 20 minutes to conduct an investigation, it should be done in five minutes. If they know that what kind of certain steps I'm taking, what are the if-then questions of the logic of a branch of a tree would be like, and what questions as part of the notes you would ask as if-then, right? So that's what we are doing now. We have launched our MCPs earlier this month, and you're seeing excessive usage, really positive feedback from our clients. Of course, it's an evolution. Like many businesses out there have done the same. You look at the CRM tools, you look at other software solutions. I think it's the future, right? Like maybe in two years, none of the actual end users might want to interact with a UI. They might just want to do everything within their own cloud or JGBT or Gemini instance, right? So that's the second layer. The third layer is how we can actually provide certain proposals to the humans, because there's an interest, and I think forever will be, especially in risk and compliance, to have a human as the final decision maker, as the supervisor of an algorithm. So no one would want to let their AI agents to go in and make changes which might go against their policies, might be biased, might be influenced with bad historical data, or might be not tailored the risk appetite or the direction of the risk and compliance procedures as a business, right? So humans will be the designers, will be the supervisors, and will be the controllers of their agentic workflows. And that's a combination of using agents, but also designing the paths of these agents, right? So what we are working on right now is turning our algorithm into an elevated version of it, which actually, on its own, not just providing new rule options, but also suggesting ongoing changes in the rule sets based on emerging patterns, not just from that specific client, but from our own network. So let's say you might have, you know, 120 rules as a client. I'm talking about rules because that's the only way it can remain explainable and supervisable. So it might not be a rule set of what people believe would be like simple, it can be like complex, but we are really against introducing any black box algorithm elements because if you turn to black books that the human would not be able to understand and change and influence, then they might not like the outcome, like the compliance team, and maybe like outcomes wouldn't be the best. So you have to keep it white box, which means that you know it will be actually like a human-readable, human-adjustable rule set. And now in this rule set, let's say if you have 200, 300 rules, over time, you know, the set will grow, get more complex. So what we hear is what our clients are calling the rocket science part is that hey, if I were to make a change to one part of the algorithm, like if I'm going to change one rule, how it will affect my false positive rates, my precision recall, F1 metrics, right? So we believe that there is a potential new way of doing it by the system, you know, re-evaluating the outcomes of the authentic workflows and also offering changes on the anti-rule sets based on what could lead to the most accurate decisions over time. But all of this should be suggestions, options to the clients. Like the clients, the end users will set the direction, the risk appetite, and the system should be able to offer option A, B, or C, offer the trade-offs, be able to back up the trade-offs with data, how it would impact false positive rates, monetary, not just in terms of transaction number-wise. And in that case, humans can make the decisions. They wouldn't have to spend the time on figuring out those changes, but the system can offer all the potential changes and they can decide what makes sense for their business, what doesn't make sense. And then this way they can switch to more like a proactive approach. Right now, machine learning is really reactive because something bad needs to happen in order for the training model to be able to predict the outcome. And again, as I said, like fraudsters won't forever try with the same methods if they are being stopped, right? So they always will look for the new loopholes, the new exploits. And then human intuition needs to be applied on actually selecting the right changes in their algorithm. But it's just very challenging because you know, if you have like a couple of hundred rules, then they will have overlaps. You might have changes in your customer journey in the UX, which might again be quite complex to solve. So in that case, AI can be helpful to actually let humans to the right outcomes, but still offer solutions, not removing the humans, but tell them like, hey, these are the options, and this is what believe is the best option based on the data, the evidence that they can surface to.

SPEAKER_02 17:25

Well, thanks for sharing that. It brings it home, it makes it more real for people, I think, to understand that.

Synthetic IDs And New Threat Vectors

SPEAKER_02 17:30

So let's talk a little bit about the future. So, where do you see the biggest growth opportunity in your segment?

SPEAKER_01 17:35

Fraud is growing. You know, even AI is fueling it. So I don't believe looking at an image or watching a video is actually something that people can take for granted as deciding whether if it's real or not. Like looking at certain defects and images of synthetic IDs, humans even cannot tell the difference between a fake video or a real video anymore, right? And that has been changing out the last two years. So I think this will move even more into what SEO was historically very good at, and that was remote, is really capturing those invisible signals of someone's device or contact information that will be really hard to replicate. So AI-powered synthetic identities are growing. You can use AI agents to create millions of accounts if you want on a certain platform, a certain app. You can also instruct them without knowing how to code, to do certain actions on your behalf or on someone else's account, but on your behalf if you are a fraudster. So I do believe that these invisible signals will be more important to be captured, to be analyzed, to be assessed real time. And I think the more data, the better, but you have to know what data you have to be after. So many companies who we talk to, they believe that they might have all the data, right? Which is true in terms of having your own data warehouse and knowing what customers are doing on your platform, right? But if your core business is not fraud detection, then you might be not the best in class at. And then there are certain vendors out there who can help you to really provide additional signals. You know, if you have a data science team, they can actually implement those signals and see how big of an impact they would make on the algorithm outcomes. And on a product and engineering level, I do think that since AI is fueling all these new threat vectors and helping process to scale the operation, then new loopholes will be opened and exploited. What we're hearing even from our own clientele is that they see more and more account decorate text. Onboarding fraud has been steady, but it hasn't decreased, it just hasn't been growing. But there are more and more attempts of account hacks. They have to be ensure that when they do a step of verification, when they use 2FA or MFA in their flow, it will ensure the highest conversion, the best level of security, for the best conversion metrics. And in some cases, when there's some real suspicious event that's happening, then you make sure that it would only disrupt the customer experience for those customers who are actually in this group, in this population of being somewhat suspicious, being far from the baseline that you would create, right? So on the RD side, I think that more data will be needed to make better decisions, which doesn't really impact customer experience, actually improve customer experience. And then it can turn the payment, fraud, risk, trust and safety teams to be seen as revenue drivers and not cost centers. Because still that's the case. Like most of these teams are seen as cost centers. Leadership have a hard time to justify increased investment or headcantal budgeting decisions because they do see that, hey, this department might be slowing or growth. No one really wants that, right? On an industry level, many companies are trying to be super apps now, right? If you are an e-commerce business, you want to offer maybe credit cards, you want to offer an EI chatbot to have to streamline refund and promo cases, right? And that's opening new loops, also new vectors, because then people can really exploit it if they know that I'm not going to talk with a human. So I will just try my best to really get that refund done, really get that promo code transmitted to me, or as well as maybe using stolen identities to open up credit cards and spend the money on those credit cards instead of using stolen credit cards, right? So e-commerce definitely changing, and even e-commerce businesses are turning into super apps and financial service providers. In the fintech space, we do see that a lot of emerging markets, BNPL, online lending services are growing. But in most markets outside of the US, credit bureaus don't have the right data. So those companies who are dealing with building the right credit scores or credit decisions, they have to rely on alternative data, which takes us back to the point of what kind of data they can take on, which doesn't really disrupt customer experience, but can still be pretty valuable. In the US, that's sort of Solved with the big credit bureaus, but they don't really exist or are not very useful outside of the states. As well as in digital banking, the big players now are capturing more and more markets. Revolute new bank, both of our clients are trying to enter the US market to win the largest market in the world in terms of economic power. And more and more smaller fintechs are popping up to solve certain challenges and use cases in better and smarter ways, versus incumbents to challenge banks, challenge financial institutions. We are also seeing iGaming space, prediction markets are affecting some of the primarily US-based population. Now they have a chance to actually gamble in non-standard gambling ways, which then again like drives us back to the point of okay, just besides regulation, how much they care about keyc compliance, are they really focused in on trust and safety? We'll be either scrutiny from the regulatory side, or will they be not caring so much? Because you know, like 15 years ago, since the US gambling market got regulated, it has changed. It used to be the largest online gambling and sportsbook market in the world. It's actually like smaller than presently the big production markets came up. So a lot of changes on that front, too. And then if you look in different industries, fake accounts, scams, especially romance crypto scams, are becoming more and more sophisticated. Now it's AI voice cloning with again deep fakes. There are definitely new ways of how fraudsters are using some of these technologies in order to social engineer, business leaders, employees, the elderly, is just becoming harder and harder to tell reality from non-reality. It's changing rapidly.

Defining Success For SEON

SPEAKER_02 23:47

So specifically to Sejan, what does success look like for you in the next say three to five years?

SPEAKER_01 23:53

I'll say that for us, success, the vision and mission we have set for the business when we started about eight years ago hasn't changed. Our mission is to create a safer place for businesses online. So if for more businesses we can create a safer place to transact and onboard customers, then we are getting closer to this mission. So I would say that our goal is to control our own destiny, which means that we are actually not interested in being acquired by a large player. We are growing organically very well. They're looking at certain acquisition targets. But let's say if we were able to 5x the business size, you know, revenue headcount, global presents, then that would be a desirable goal for me personally as well. I would like to sign up more Fortune 500 businesses as part of our clientele. Beyond that, beyond these kind of business goals, if we have a chance to become a household name in terms of having Sion as a keyword on a CV that, hey, I have experience using Sion, and your business would be hiring for that sort of experience. That would be a very positive moment for me that we made an impact in this space and we weren't just one of the many high-level similar tools. We would like to be a category leader, would like to become the 600-pound gorilla in this space who is not just providing the highest quality first-party signals or just tapping to cover the entire workflow for clients, but also providing the best decisions, having the most accurate decisions at a significantly lower cost than what it would be possible today.

Payments Trends Stablecoins And Fewer Middlemen

SPEAKER_02 25:26

Well, when you step back and look at the payments industry as a whole, I mean, we've talked about AI a lot. Obviously, that's one of the biggest trends in the industry. But what else do you think are the trends that are reshaping the payments industry as a whole?

SPEAKER_01 25:40

I would say that stable coins are one area which is chaining on the payment trails more rapidly than ever before. So USDT, USDC, and how crypto assets can move into the standard payment trails, how it can move out, what kind of compliance challenges it might create for businesses to be able to accept those sort of payments is definitely a big area. The second thing is a lot of the challenger digital banks are offering their own schemes, which means that the big car networks might be less and less relevant in the distant future, not in the short term. But we do see that the payment chain and rails, you know, starting from the issuers through the acquires to PSPs and merchants, and it's looking for ways to skip the middlemans in this chain. So how we can bring the consumers who are owning the wallet into the merchant's pocket as seamlessly as possible, right? And right now, a lot of middlemans are using technologies that were created 30, 40 years ago. And they might be outdated, they might be improved and change some bits of it, but it's essentially the same, right? So what crypto has helped to understand that you can own your own wallet. You can trust the institution with it, you know, if you were to hold your crypto or your money in an exchange or in a bank, but you can just hold your own wallet on your desk. But also the same with like USD CNT, like with the stable coins, right? Like the currency really doesn't have an impact on them. So I do see that there will be a tendency for these challenger banks to really try to find ways to connect to the merchants without relying on these middlemen's, such as the PSPs, orchestrators, or the acquires. And also maybe consumers will be on interest of doing so because they might get some discounts. You know, if you wouldn't have to pay car networks their fees, then it could be a mean situation. You know, if they are not able to innovate like some of the other players on the side of the spectrum are able to innovate, then they have to find ways, you know, in order to remain relevant in 25, 30 years out. I do see that the reason why they are, you know, like acquiring a lot of business in the space because they want to be relevant, right, in 25, 30 years. But also like consumers are smart, right? So they just know for a fact that if there are alternative options without using their credit card, which helps them to process payments for lower cost, which means that having some discounts on certain services or products, then they will just go for it. Like it's not hard, right? But consumers have a trust issue, a safety issue with some of these challenges banks. So they wouldn't really want to keep all of their earnings and income in a digital place, right? If they cannot go to a big and mortal place, then they might not trust it. I think it's changing. So more and more people are actually open to that. It's not a switch of a button, it won't happen immediately. But as you know, people are being more comfortable of just relying on some of these digital financial institutions to hold all of their savings and money, then eventually those financial institutions will eat up some middlemans of this payment chain and can go directly to the merchants. And then you know, merchants might cooperate with those banks. So in the end, hopefully the customers will be on the winning end. I hope it will happen. People are expecting a much faster change. I would say like five, six years ago when COVID has started. Now it's changing, but significantly slower than some people thought it might. But yeah, it's happening. So like in 25 years, probably we will be paying and shopping, especially now with AI agents capable of browsing and selecting services and goods for you and maybe like transacting on your behalf. So I imagine a world where I just open up whatever will be like the Siri on my phone and say, like, hey, buy me XYZ, and then it will just do it. Knows my address, knows a preferred place to go, has access to my wallet, whether if it's a wallet from one of the large technology companies or somewhere else, they will have infinite options to make that purchase, but it will be definitely tailored to their historic behavior and preferences.

SPEAKER_02 29:32

You hit the nail on the head. This is an industry full of a lot of middlemen. I don't know if in 25 years maybe that changes. Everyone thinks it'll change faster than it is. And I think the other one is the agentic commerce, like the agents buying. That comes up in a lot of conversation as well. As we come towards the end, two final

Career Advice US First Talent Always

SPEAKER_02 29:49

questions. One, if you could go back and give yourself some advice at the very beginning of your career, what would that advice be?

SPEAKER_01 29:56

That's a very good question. One of the advices would be go to the US market as fast as possible. So I think we made that decision maybe two years late. We're in a position already being an early mover in the European market, but the US market is the largest software market in the world. Every second dollar is being spent in the US for any software in the world. And also they have like larger budgets and shorter sales cycles too. So it requires a different approach. It's more personal relationship-based, but also like they just buy from whoever they like in the States. In Europe, it's more about like head-to-head comparison, you know, pricing and negotiation. Just it's different. But I would say like all of our big competitors were and still are in the US. You don't really have someone big outside of the US. So I would have challenged those players in their own market, in their own courts, probably earlier. The second thing is don't compromise on talent. You know, sometimes we make decisions because some people were too expensive. You know, we thought that, hey, we are like too early to afford someone so expensive. But talent and skill and experience will come with a price tag. You know, you cannot compromise. And I see also many other funders are struggling with these decisions is that they do think that they look at the price tag, but they don't look at the value. They don't really see that, okay, well, if I spend on someone's salary this amount, then they might get like three, four, five, ten, fifty, hundred X back, right? But you can only make as good as a business as the talent you have in the business. If you have no good talent, your business will go nowhere. If you're able to access talent, if it cannot be the price tag, that's what you should spend money on. You should just spend money on anything else that doesn't really improve on your business in terms of the value that you're creating for the world and for your customers. That would be the two main things I would point out.

SPEAKER_02 31:41

Final question: what's the one thing that payment listeners that are listening to the show today, what should they be thinking about right

Agentic Commerce Good Bots Bad Bots

SPEAKER_02 31:48

now?

SPEAKER_01 31:48

So, as you have mentioned, like the agentic commerce is a big thing, right? So what I'm hearing, what I'm seeing is that let's say about eight, nine months ago, it was a huge storm, right? Everyone thought that it would happen probably in the next four, five, six months. Like it's much slower, but it's happening. What you see is that agentic commerce traffic is actually growing. It's not growing as fast as expected, but it's growing. Thrusters and bots have a very easy time to blend in with that traffic. It's very hard to tell whether they are good or bad bots, or good or bad agents. But I would say that you know, try to quantify the impact of those type of customers and transactions who are being actually instructed by a human through an agent to conduct agent e commerce and look from like the revenue perspective because the more revenue you can accept even from bots, the faster your business will grow. But also, thrusters are smart, so they know that if they can blend in with the traffic, that's the exploit they will be after. And we actually see growth in that. So that's why we've been working on our dynamic fiction approach, is how we could have businesses to verify those agents, determine whether they are bad or good. And then once it's done and you trust the agent, you wouldn't want to verify them as long as it's still the same agent instructed by the same human. So, what I'm surprised of is some businesses are capable of telling that whether you are actually accepting a transaction from an agent or not. And tools like Xeon can do that. But most of the e-commerce engines are not capable of doing that. So if you are a business owner, you're accepting payments, try to think of some tools that can help you to see that, hey, how this traffic looks like today and how it has been growing, how it will grow maybe in the future, and how it will have an impact on my processes. How can I actually set up my business in a way to be open? I do see when I move to the US and start to use some sites, when I come back to Europe, many of these sites are not actually visible outside of the US visitor. So I have to use a VPN to see some of these sites. I understand what's the reason, but I will go back and maybe purchase something so I don't want to use my VPN to just browse the product. So I'm thinking like, hey, the decision was just because Timbe had like a couple of bad and transactional events or or some attacks from like outside the US, just block everything from outside the US. So like instead of export control, I don't think it makes sense. I think the whole market should be global. And that includes, even now with the entropy craze, I'm not sure if you are following it, but you know, the export controls don't make any sense. Like commerce should be global. All businesses should be able to accept customers globally in order to grow their business as fast as possible, if that's what they want. If they impose restrictions, even on a gentic commerce traffic, they will struggle. Like I think the idea is to be sure that you wouldn't turn away good customers. And now that more and more tools and methods are coming up to actually ensure that it's possible for business owners out there, but it's not happening as fast as I

Final Takeaways And Closing

SPEAKER_01 34:44

would like.

SPEAKER_02 34:44

Well, Tomas, I think that's a great way to wrap up the show. So thank you so much for being here. I know your time is very valuable, so I really appreciate you being on the show today.

SPEAKER_01 34:52

My pleasure, greatly appreciate it.

SPEAKER_02 34:53

And to all you listeners out there, I thank you for your time as well.

SPEAKER_00 34:56

And until the next story.com, where you can subscribe to the show and where you'll find our show notes. If you enjoyed listening, please share on your social channels as well.