Trusted Intelligence

In this episode, Yuval Giveon sits down with Brad  Levy , CEO and Board Member from ThetaRay to discuss one of the biggest challenges facing the payments industry today: how to manage risk in a world of rapidly growing transaction volumes and increasingly complex financial networks.

As cross-border payments transactions continue to expand, financial institutions are dealing with greater exposure to fraud, financial crime, and compliance risks.

 The conversation explores how ThetaRay leverages AI-driven transaction monitoring to help organizations detect suspicious activity, uncover hidden risks, and maintain compliance without slowing down business growth.

The episode also examines how payment providers and fintechs can balance innovation with security, and why advanced data analytics and intelligent risk management are becoming essential components of modern payment infrastructure.

Transcript

SPEAKER_00 0:06

Hello everyone and welcome to another episode of our podcast Crossing Borders. I'm very, very happy to have with me Brad Levi, the CEO of uh Tetaray. Um Brad, thank you so much for joining me. It's a pleasure to have you with us.

SPEAKER_01 0:23

Thank you for having me, Yuval. It's really a pleasure and looking forward to uh a good half hour with you.

SPEAKER_00 0:29

Sure thing. So um why don't before we dive in, why don't you tell our listeners a little bit about yourself? How did you find yourself um at the nice uh office um in New York and leading um Tetarray? And we'll take it from there.

SPEAKER_01 0:47

Sure. Yeah, so I'm 35 years into a Wall Street, I guess, fintech career. Uh I've been more outside of the banking world for about 15 years. Uh I've been a Theta Ray for coming on six months now. It's a financial crime uh technology company, uh, really in the reg tech space, uh, focused on very specifically anti-money laundering technology uh for banks and payments platforms and others in the financial system to really comply with regulations, number one, and do business with good actors uh in general. So it's a business-enabling platform that reduces risk and really tries to create operational leverage for people, you know, really dealing with quite a bit of hits on uh issues in uh the different financial supply chains of money laundering and sanctions and all of that behavior around the world that we're uh we're helping our clients manage.

SPEAKER_00 1:40

Amazing. So this is actually where um the two companies um started working together, NIMA, um as all of you probably know because you've been hearing me for the past few months now. Um we've been dealing with the cross-border payments. Um, the operation grew dramatically over the past few years, and the challenges of monitoring and understanding the volumes we process has become more challenging. Um both regulators but also um clients and customers demand not only um uh sanctioned screening like in the good old days, we now have AML monitoring and AML monitoring um is not enough because we now have AI and we now um both want and both need to also be able to monitor um transactions in real time. So, really um for us as a company, we've decided that we're not going to build this. We have a lot of things we're building, but um I really believe in the saying of do one thing well. And for us, we want to focus on moving money from one country to the other country, from one currency to the other, in real time, with perfect success rate. And we knew we want a partner that will go along with us, that will um keep up with our pace, and I think Tetray um is that partner for the past few years now. So um that's like uh the history of the relationship, so to speak. So, how how did the market change from your perspective, Brad? From the again, good old days of screening and maybe AML monitoring to what we've seen now, like rule-based and AI and uh real-time monitoring. Like, how do how do you see the industry change over the past few years, maybe?

SPEAKER_01 3:39

Well, philosophically, we align really nicely as we stick to our specializations. Uh, you know, there's plenty to do in cross-border payments, uh, plenty. Uh, and in what we do, there's real deep expertise. So I think that partnership combination comes from our natural philosophy to stick to our what we really do and do well. Um, I will tell you from my perspective, whatever challenges we have may just be getting started. If you think about technology and agents and drones and different types of synthetic level attacks, um, you know, we really are probably at the precipice of an explosion in volume. That said, data is always the challenge. Your own data, how you deal with data that you don't control but is away from you, but is clear, and then how you deal with data that's very dynamic and changing second to second or you know, nanoseconds. So I think they say now there's more data created sort of every day or every every year than has been created in the history of time. And that may be just getting started as well. So number one, how do you deal with huge amounts of structured, semi-structured, and unstructured data? Number two, how do you deal with the fact that it's exploding? Number three, how do you deal with the fact that all these technologies are both existing like mainframes and APIs and regular old tech and new agents and AI stacks and data centers? Uh and then the third leg, which we haven't said and we may not today, but this idea of quantum technologies, which is maybe five, ten, fifteen years away. So if we think it's fast, it's just getting started. If we think it's challenging, it's actually getting more complex. I would argue that a lot of that complexity can be dealt with though with the new technologies as well. So it's a double-edged sword, and you know, it's really good to have a partnership with such a deep payments, cross-border payments specialist, which is a very challenging and and very big deal. And we have a lot of new payments methods coming in the next few years, as we know, with stable coins, et cetera, that I'm sure we'll chat about.

SPEAKER_00 5:36

So how do you really um deal with that? I mean, um you serve big clients like um banks and the dynamic and smaller, maybe fintechs and and companies like Nima. So, how do you create a machine, so to speak, that makes sure um the volume is legit, that you know who's at one end and who's at the other end. Um how do you um process really huge amounts of data and flag transactions? Like this transaction should be flagged, this transaction should be investigated. I know from our domain and our challenges, we process money in like over 100 countries, and each country has its own specific and unique payment methods, like you have widget and alite in China, you have GKSh in the Philippines, you have UPI in India, and you have obviously Fedwire and the ACH in the US and SEPA and the SEPA instant in Europe. So, how how do you manage to take all this data and all these payment methods and say, hey, something's not right with this transaction? What's the secret sauce?

SPEAKER_01 6:52

Obviously, AI, but but I mean Yeah, no, I would say it really comes from this flexible philosophy, right? You just need to know that whatever you're building is dated the moment you ship it. It's always been true, but it's just getting more true every day. You know, this idea of elasticity of software, low code, workflow development kit, everything needs to be able to be picked up pretty easily by almost any reasonably technical person, and that movement is on big time. So I'd say that flexibility is critical always, but more so now than ever. Um But really, if you look, you said it well, like the nature of payments is very granular and detailed. It gets to cultural things within a country, how types of things that happen in a country in terms of human trafficking, sex trafficking, arms dealing, money dealing, drug dealing. Like it could be very country, not neighborhood specific. So you really need what we call our model layer to be very unbiased and virtually unassisted in somehow getting to unknown unknowns. Like it is very tough to find unknown unknowns. It's because they're unknown. Now, it's hard enough to find the known knowns, just the hits that you have to hit on the sanctioned screen and the not false positive alerts. Then there's the known unknowns, which is hard but not impossible, or not impossibly hard with very high-end tech. But we really started with a company over 10 years ago that came at it with unbiased, unassisted, real frontier-level machine learning for anomaly detection. We have literally these algorithms that form the basis of, I'd say, our core learning loop intelligence. That said, we're not building an LLM, we're not competing at that level, and we're leveraging LLMs for a lot of the workflow, I'd say, now the more agenc layer than the true algorithmic anomaly detection. So the model layer is part of our intelligence, the software layer is how people get to alerts and action them, and then the agentic layer will now be how you sweat and leverage the resources you have around the world. So the increment, it's like any athlete or any learner, it's incremental, it's over time, and the more curious and flexible you are, the better you're gonna be. But you also have to know these things are rigid and you have your rules and you can't be wildly flexible either. You have to be a lot of the time.

SPEAKER_00 9:10

I think that people really tend to forget how powerful um traditional deep learning and machine learning the technology and technologies are.

SPEAKER_01 9:19

I'll even tell I'll yes and you on that. I think people forget how powerful human intelligence is in front of them. Right.

SPEAKER_00 9:25

Right.

SPEAKER_01 9:26

A lot of what these technologies are leverage for humans that make good judgments. Humans that make bad judgments, the technology is actually gonna not help them. It will exacerbate their bad judgment. I've seen it in 08, I've seen it in 94, I've seen it in 98 in the financial systems. The technology is generic, but some people could get wildly levered without having a clue at that time, just like these technologies have a double-edged sword. So I really think human intelligence plus really deep learning, unassisted, kind of how do you really find the thing that's a true anomaly in a sea of, but I don't think doing it for defense or biotech is the same thing for doing it for financial crime.

unknown 10:05

Right.

SPEAKER_01 10:06

Which is why we have chosen financial crime as our specialization.

SPEAKER_00 10:09

Right. That's interesting. For me, when I look at the future, future of payments, maybe, or or future, which is probably not too far away. Um I see that every one of us um has an agent. It might be supercharged Siri or a Google Assistant or whatever phone you have, that obviously it can help me um make a reservation to a restaurant, or help me um with a reminder, or provide information. But it is crystal clear for me that the day is not far where I could um ask Siri to pay the rent or send money again, like I sent last month, or stuff like that, stuff that's related to money, because again, not being too um buzzworthy, like AI agents and MCP, that's already happening in payments, in like retail, and it is most definitely going to happen for personal banking.

SPEAKER_01 11:11

Like I'm going to have my way, Yuval, I would argue to pile on there is like the open banking technologies that came over 20 years. Exactly, exactly. On top of a so there's computers, the internet, open banking, and now agentic, all of those layered over time and incrementally moved. The the traditional capital markets institutional world has not gone to an open banking kind of back-end model yet. So, but now those movements will happen across the whole financial system, but they'll happen in a bit shorter time, but not still in weeks or months. It will still take a few years for that core infrastructure to exist so that connectivity can happen. Um I think that um and just to make the uh you know, if we think about these technologies, uh you will not have one assistant. You will have numerous assistants that help you with your travel, your banking. They'll be just like in your human life, there'll be personas you won't have Siri, you'll have 50 different agents that do everything from advise you on I'm not gonna want the same voice or specialization that helps me be healthy. That would also maybe help me um, you know, fix my Wi-Fi in my home. That's like Craigslist, right? You're going to want TaskRabbit or Angie or Match.com. You're gonna want your specialist.

SPEAKER_00 12:39

Right.

SPEAKER_01 12:40

It it will just be different voices and different models, but it will still be a sea of people helping you. It will just be more easy to access.

SPEAKER_00 12:48

Yeah, that that's it's really interesting for us as a payments company. And again, I'm partly thinking out loud like, what does this mean? Do we need to uh make our services available not only via API but also via MCP? Am I going to build a layer that's going to allow all of our agents to talk to our system and to move money? So it's really interesting as the as a payment company, but like how do you look at it? How do you fit in that brave new world, so to speak?

SPEAKER_01 13:23

So I think it's um I wrote a piece in last October called Ledgers and Tokens and Bots OMI, because I believe programmable ledgers and digital tokens and smart bots are convergent. And but they'll still happen at a very granular specialized level. So if I think about us in reg tech, we're not covering compliance, we're not even covering all of a KYC process. And you might argue a KYC process is really a result of knowing your client. And if you can move certain features functions up to more real time and then just indicate that they happened, as opposed to running a process up or downstream. So I think of all of these things in terms of supply chains, and you're a payments cross-border platform. But payments only exist if something else happens in front of that in a supply chain normally. Right? I bought oil, I bought bread, I bought grain, I bought something, now I have to procure something. So and I don't think of AML even anymore. I think of A VL like anti-value laundering. Because if you take money and buy a good, and then there's use that good to barter for another good, you actually want to track the money to good to good from an AVL perspective.

SPEAKER_00 14:38

Right.

SPEAKER_01 14:39

And when stable coins and all these, you know, new digital assets, whether it's a building that's digital twin, that's telling you it needs more fuel and how you might get that there efficiently if you rearrange some things in the twin. Um, all the way down to just the bond market and equity market supply chain, which you're seeing play out in real time with SpaceX, and people are piling into that, so siphoning money off of some other things or selling a house to buy some SpaceX stock. But it's all much more connected and in a value chain and a supply chain than and you're kind of responsible for these up and down, you know, blood diamond challenges that we all have.

SPEAKER_00 15:20

No doubt that real time was and still is like the key because you want to know to alert in real time if this payment or this transaction or this barter, like you mentioned, is legit or not. But um, because again, AI agents talking to one another and making decisions.

SPEAKER_01 15:44

Are you born? You're born once, right? Just once in theory, right? And then you die. That's a pretty linear thing.

SPEAKER_00 15:51

Correct, correct.

SPEAKER_01 15:52

I don't mean to be deathy, but but the reality is while that's happening, this is actually happening.

SPEAKER_00 16:00

Right.

SPEAKER_01 16:00

So the fact that you are linear is kind of irrelevant. Everything is moving real time around you. So we have to move linearly just to kind of wake up and go to bed. But the reality is the world doesn't shut off, bitcoins don't shut off, it's a 24-7 market. Uh yeah, yeah. Oil supply chains can't settle. So this idea that things shut off and is linear is silly, right? Binary code and linear doesn't really exist. So everything is real time.

SPEAKER_00 16:30

Cut off time, end of business day. This doesn't exist anymore.

SPEAKER_01 16:34

Why don't we just take a snapshot at the moment of real time and call it the moment, as opposed to close the books or try to slow the world down? Because that's worked for 50 or 70 years with tech, but now you cannot follow the sun and you can't throw bodies at it in a low-cost region to pick up on your induction management. It just doesn't we're we're we're now we can't do that physically. It's we're out of we're out of that.

SPEAKER_00 16:58

That's super interesting. And also like as if uh, you know, um stable coins alone aren't enough and and the MCP and AI agents aren't enough. Like I keep hearing more and more about like stable coins. We've been hearing a lot about stable coins. Obviously, you're in the industry, so you know this. But I think like what I've been hearing in the past few months is that AI agents are finally going to be able to utilize stable coins-based um treasury or liquidity management to finally find a usage to these um rails. Because PEM blockchain and stablecoins, it's rails. We're not going to go to the grocery store and pay for the bread or milk with USDC or Bitcoin, at least not at this moment. Maybe in the future we will. But um like you mentioned, the world doesn't slip, the world doesn't stop. Like there are no cutoff times, there are no um, we want the money to move, to move and work. Also in weekends, like I don't want to be thinking about business days. So AI agents working 24-7, utilizing Rails that can move money 24-7. If you add on top of that, on and off ramp, like money is not going to sleep. So that's a huge opportunity for me as a payment company. Obviously, it brings a lot of challenges in terms of compliance and risk and regulation and whatever. So, so yeah, it's it's really interesting, it's frightening, and we are going to be um needing partners like yourself to help us adapt and make sure we bring the trust that we brought until this day also to you know payments 2.0. Because at the end of the day, every bank, every financial um company, every fintech, we sell trust. Why would you trust us with your money? Why would you trust us to make this transaction? How will you trust us when we say the money has been paid at China or Africa or the US or whatever? So we sell trust, and the more complex the ecosystem is becoming, we're going to keep up and keep on selling this trust, again, by ourselves and uh with partners, obviously.

SPEAKER_01 19:33

No, and trust comes through time. I believe that deeply, and it comes from I wrote a formula around it. It's having good and bad experiences with entities or people or things through time, right? And the more you have a good experience, the more trust.

SPEAKER_00 19:46

Right.

SPEAKER_01 19:46

You can actually destroy quite a bit of trust in a moment, um, especially when it's really challenging. Meaning, like in a bad time, you can take 20 years of trust built and destroy it in a minute. So I think, you know, to your point, the you know, these technologies that are old don't go away. They get leveraged. Like we didn't build new railroad tracks everywhere on the planet when we got faster trains. Um, we leveraged and we combined those things. Mainframes are still around and they actually run like at the bottom of the ocean, like quite resilient. They don't have great API, you know, kit connections, but they do run, and there's probably a place for that. So I think it's this idea that you build things up over time, trust builds over time, and then there's catalysts, like the fact that the Clarity Act has not passed in the US yet. That to me is the thing that's probably stopping things from really developing. Part of it is it just needs to happen. Part of it is the US is the biggest financial system in the world. So until that goes somewhat digital, crypto, it's hard to say that the global system can go there. But it seems like it's happening this year. Just at the moment where MCP model context protocol environments and agent-to-agent, and maybe the thing we need to solve for most quickly around trust is how you identify people uh digitally, identify entities digitally, and identify objects digitally. I wrote again a piece on this in 18 called uh the weapon of mass solution, the trident, the triangle of identity management across entities, objects, and individuals. Now that we, you know, we won't have one social security number, we'll have many ways to identify and authorize us, as will agents, as will MCPs, as will, you know, the firewall is dead, and every object will be blocked and protected and hopefully crypto proof or quantum proof at the crypto level. But yeah, all these things are um happening pretty quickly, relative. They're converging. And any one of them is not interesting, but three, five, ten of them together is is complete change of society.

SPEAKER_00 21:52

Right, right. Right. So um we've been talking a lot about um technology.

SPEAKER_01 22:00

about fintech about moving money and I'd really love to get your take about um building company as a relatively new CEO of Tetaray like in 2026 like what changed like in leading a company like what's different today that wasn't um like like like this like five years ago ten years ago maybe two years ago yeah I would say I have a I have some and and you know like Tetra is growing very quickly very timely so I have a thing where I alliterate and I have a model of rule of three so people process and product I've always sort of believed in that order um no matter how many robots come no matter how much artificial intelligence comes people will still be the most important factor in these recipes um so I believe that's still the case uh this company has had great people for many years and always does at the time it needs it I believe that uh but the reality is being able to do more having more range as a skilled individual truly running globally with those local nuances that you described earlier you know that's a lot to deal with and I think um you know there's different types of people not age based that will figure out a way to deal with these new technologies and make it their own and really lever it up so I think you know in the last six months there have been a number of changes here on the people front um some of them by design and a lot of them I would say all of them for me at least geared towards what does the future look like and how do we build it versus how do we consume the future from others. So that's just a philosophy. I think if you go from batch to real time in life as a company and how everything runs that's a very big change. So a lot of our processes that could run by kind of a daily system now need a continuous system, whatever that is shipping software, making decisions at management level having fun events as a team and then the product like we're probably turning from five, eight years ago to more of a service company based on tech to a product solution software company based on agent. And that's a huge sift where a lot of our you're much more of a client of ours on the product and the stack and the algo where some rely on us even as a service provider as individuals to leverage this on their behalf. So I think you know we're moving a bit away from the services model into a software model and even into a consumption based value-based pricing of how we determine what our worth is you know on the field in the market. So a lot of changes around token usage in cloud, bodies to build bodies to implement bodies to support or tech across all of those. So again a lot of what we're doing is based on the macro trends the exploding problem the legacy stacks that need shining up but the reality is it's an incremental world and there's going to be very few big bang rollouts of these systems in the future in a good way not a dull moment not a dull moment.

SPEAKER_00 25:18

I couldn't agree more about what he said at the end of the day the company the organization is is it's people okay so even for us and even like we have new technologies and AI and the product is changing and the business is changing and the industry is changing at the end of the day like if you have a team that you trust on to build and deliver and to adapt and at the end of the day believe in the same cause and want to solve the same pain and so that that's what you need like that's the the winning team.

SPEAKER_01 25:54

Just to prove it the first day or week I was here at the firm we redid our values from maybe six values that were maybe I think 19 or so words to five values that are 10 words and it's purpose driven right at the top it's a different purpose. We're in financial crime it's kind of a really high order interesting purpose I believe bold and creative fast and smart uh trusted accountability and one theta ray. So we've got these 10 words now one theta and purpose driven as the book ends. And then the point on bold you got to take risks and be bold, be creative, you got to be fast and smart and you have to have trusted but also you know how do you build trust by being accountable but as a specialist so I really love what the values we came up with. It could be just corporate values on stickers around the office but it really feels right. It feels like we're living it and if we're not we're becoming those values and uh it's probably a really important time to be doing that. One thing I like to say is like artificial intelligence is exactly that. I like HI, human intelligence wrapped symbiotically with AI but wrapped and then I'm not I don't sell FOMO I sell Joji the joy of joining in AI like there's a real awesomeness to what's going on but there is a darkness to some of these technologies and we have to actually use the awesomeness to to deal with some of that dark. But it's happening inevitably which is why we have no choice which is kind of cool.

SPEAKER_00 27:25

Right right very interesting um a lot of things to think about I think um but the future really sounds promising um for you for us for the industry and hopefully for humanity and um Brad thank you very much for uh for joining us and and and uh that's it was a true pleasure to talk to you today.

SPEAKER_01 27:52

No Nival thank you for the word partnership right up front where you you're our client and we take that seriously but we've really been developing a lot of interesting things from our specialization seats and I think it's just begun and we appreciate that. So thank you and for this opportunity to speak with you today.

SPEAKER_00 28:09

Sure thing bye bye