In this episode of B2B Automation Spotlight, Michael Bernzweig sits down with Brad Levy, CEO of ThetaRay, to explore the transformative power of AI in compliance systems. Brad shares his journey from early fintech days to leading innovations in anti-money laundering technology. Dive into how ThetaRay is pioneering the shift from traditional rule-based systems to dynamic risk-based infrastructures. Gain insights into the future of financial crime detection and the importance of AI transparency and efficiency.
Introduction
Interview
Presentation
Panel Q&A
[MICHAEL]: You know, when Brad Levy began his journey in the fintech world over three decades ago, computers were a rarity on desks... and fax machines were the norm. Fast forward to today, and Brad is leading ThetaRay in using cutting-edge machine learning to combat financial crime. We'll delve into his transformation from a banker navigating early e-commerce challenges at Goldman Sachs to a CEO pioneering AI solutions for anti-money laundering. By the end of this episode, you'll understand exactly why embracing risk-based decision-making is crucial for modern finance. I am your host Michael Bernzweig and this is B2B Automation Spotlight.
--- INTERVIEW ---
[BRAD]: I know a lot of the individuals listening in and tuning in for this podcast episode this week have had their own journeys getting to where they are in in their organization, but yours has absolutely been a fascinating journey with a lot of twists and turns along the way.
[BRAD]: So I was hoping you could share a little bit with the audience your journey getting to where you are as the CEO and board member over at Theta Ray.
[BRAD]: Well, then thanks again, Michael, for the opportunity, and happy to talk about journey and backing into what we do today at Theta Ray.
[BRAD]: But I you know, I've been, in the, I guess, fintech financial services world for about thirty five years pretty directly now.
[BRAD]: Started in university doing research in the municipal finance world, public finance.
[BRAD]: And then I landed as a banker in public power project finance in the nineties, which was interesting because it was very active in the derivative space, and we were early adopters of technology and Lotus Excel.
[BRAD]: So I got a very early taste of, using technologies at work when people didn't have computers on desks.
[BRAD]: Assistance had IBM Selectrics.
[BRAD]: Dictation was done.
[BRAD]: People had letterhead.
[BRAD]: The Internet didn't exist.
[BRAD]: Computers were barely around.
[BRAD]: So, you know, that was fun to grow up in that investment banking world and doing pretty interesting, hard, quantitative type work.
[BRAD]: Again, on a relative basis, it wasn't quantitative, I guess, by today's standards.
[BRAD]: By the late nineties, I moved into the capital market side of that business onto the desk, trading desks, indiguit desk.
[BRAD]: And I took on an e commerce role as well at Goldman Sachs.
[BRAD]: So I did a line job and then I did a strategic role for ecommerce for the department.
[BRAD]: I worked on the y two k project.
[BRAD]: I worked on CRM systems for my department, you know, in addition to sort of sitting in the market and underwriting deals and taking risk and getting my head handed to me when long term capital blew up or some currency crisis.
[BRAD]: And so, yeah, I was an ecommerce person really fully by o one when the dot com bubble burst.
[BRAD]: And I had started working in what at Goldman was the FICC division across creating market infrastructures like exchanges, clearing houses, data companies, settlement platforms, working on standards like FICC and RIXML and just using the Internet or technology more as sort of two thousand one, two, three, five happened.
[BRAD]: I got very active in derivatives through that time and credit derivatives and the subprime markets and all those indices, and I developed a lot of product on the financial product side in addition to financial markets infrastructure, technologies, etcetera.
[BRAD]: And then in twelve, I left Goldman to join a portfolio company with market with an I.
[BRAD]: I was there for a few years.
[BRAD]: We went public and merged with IHS.
[BRAD]: It was a financial services software data company.
[BRAD]: It got quite large over time.
[BRAD]: Eventually, it was bought by S and P.
[BRAD]: In twenty nineteen, twenty twenty, I left that company and I joined another technology consortium involved in communication space called Symphony Collaboration Services.
[BRAD]: And I spent five years there, most of it as CEO where I was prior to Theta Ray.
[BRAD]: So I guess long way around, lots of compliance, lots of front middle back office, lots of global, lots of sort of trying to bring modern technology into a world that doesn't accept things that easily like all industry.
[BRAD]: I should think finance is pretty far forward, but, you know, can feel a bit antiquated in data at the times when you hear about mainframes and fax machines and things like that still roaming.
[BRAD]: But, yeah.
[BRAD]: So I've spent, you know, a lot of time in fintech before it was fintech, market infrastructures, and lots of compliance, related areas over time.
[BRAD]: Now I'm in a very, heavy compliance space with any money laundering and financial crime, etcetera, but it's not too far afield from where I've been for a few decades now.
[BRAD]: I love it.
[BRAD]: Yeah.
[BRAD]: And some of those early financial apps that you mentioned, Lotus and the rest, you know, ring a bell, I think, for a lot of listeners.
[BRAD]: So Yeah.
[BRAD]: Like the always add in for WYSIWYG to print if that rings a bell for the fifty five year olds out there.
[BRAD]: Exactly.
[BRAD]: Exactly.
[MICHAEL]: And, you know, before we get too too far into the episode, for any listeners that may not be familiar with either yourself or with Theta Ray, can can you give us the the quick elevator pitch of the solution and and the types of clients and organizations that Theta Ray is is helping?
[MICHAEL]: Sure.
[BRAD]: And I made up a word of three combination, which is a practo technovangelist, a practical technical evangelist.
[BRAD]: So that's what I think of myself as.
[BRAD]: And so I love the modern.
[BRAD]: I love the innovation.
[BRAD]: I love moving to the future, but the reality is it takes time.
[BRAD]: So I think we're a fourteen year old company, as Theta Ray.
[BRAD]: We started as a bit more of an academic technology company using machine learning technologies to find anomalies in something.
[BRAD]: So that was around, I would say, the Palantir DeepMind days when these technologies were really starting to come and we started to develop these algorithms as a company.
[BRAD]: Within a few years, they found the product market fit, which is applying these algorithms to financial crime and specifically any money laundering.
[BRAD]: So that was the first product market fit and really going into the transaction monitoring space.
[BRAD]: And I would argue we're one of the early users of machine learning algorithms to find anomalies in payment systems and banks, to find, obviously, money laundering, human trafficking, sex trafficking, arms dealing, part you know, all of those things drive You know, money laundering is illegal, but often it's a symptom of other, you know, criminal behavior.
[BRAD]: So, again, Israeli tech company that had built that technology over the years went more global.
[BRAD]: We have a significant presence in Europe, Spain and France in particular, Centender, Credit Agricole are big clients.
[BRAD]: And we have a number of clients that are banks, so traditional banks as you would think about down the street.
[BRAD]: That said, we have a lot of other types of clients like Travelex in the money moving space, obviously, or Shift4 in the point of sale.
[BRAD]: We have clients in Africa and UAE and Mexico.
[BRAD]: And even when we have a bank client, say, like, Ascentender or Credit Agricole, our deployment is across multiple countries and multiple banks in multiple regions.
[BRAD]: So one client looks like several or a bunch of implementations because all of this is done at the regulatory level, which is usually almost always country based.
[BRAD]: So where transaction monitoring, screening, customer risk assessments, that's sort of how people describe our core functional product.
[BRAD]: We now have our Ray Agent that is helping the investigator do the work that's coming out of an alert from a transaction monitoring system or screening it on a sanction list or a, you know, customer risk assessments that's flashing something and you have to go out, you know, do some bottoming out.
[BRAD]: So, again, model layer, software layer to get to the intelligence, and then now an agentic layer to help do the work, you know, more efficiently.
[BRAD]: Quite global as a company, a couple of hundred people largely operating out of Israel, Madrid, New York, London, and a few other spots.
[BRAD]: So the original solution obviously had academic origins and and a a different model initially, but very quickly, it sounds like the problem with the two commercial origins driven by academics, if that makes sense.
[BRAD]: Okay.
[BRAD]: Alright.
[BRAD]: So so almost almost the exact opposite, but No.
[BRAD]: No.
[BRAD]: It was it was an academic group.
[BRAD]: Like, it was people that were more academic but trying to build a company.
[BRAD]: They were just looking for, like, where can we Got it.
[BRAD]: And build a business.
[MICHAEL]: So anyway So so that original idea, you know, if you were looking back to those early days, you know, when the company was founded, once the solution was to market, how how did how did the organization go from from zero to one, bringing aboard that that first organization that was, you know, rolling the the solution out and and there in the early years?
[MICHAEL]: How how did that happen?
[MICHAEL]: Yeah.
[BRAD]: Solution's an interesting word because I would say five or ten years ago, the solution meant working with a client to apply a model, tune it, and implement something.
[MICHAEL]: We would provide So it was very hands on at the end of the day?
[BRAD]: Very hands on, and I'll use the word services, where we provided a lot of services to help that client comply with our technology.
[BRAD]: We are now providing, I'd say, solutions based on software, a model, and an agent.
[BRAD]: It's more technically driven than project and human driven.
[BRAD]: So it's really come full circle to to where it is today.
[BRAD]: And again, ten, twenty years ago, a lot of businesses were services with tech inside of it.
[BRAD]: Now a lot of people are trying to really sell tech with some minimal services.
[MICHAEL]: Right?
[BRAD]: And whether it's ongoing services, managed services, professional services to implement or, you know, we all are obviously trying to move to more subscriptions on transactions or subscriptions on agents or tokenize something in consumption based, value based pricing.
[BRAD]: So I would say it was really working with clients that trusted us with sensitive information to take models and algorithms, apply them to data.
[BRAD]: They had already been trying to do this themselves and we could say, we're finding more alerts than you are and we're seeing less false positives than you do.
[BRAD]: And a lot of this comes down to model training and model tuning and the clustering you might look at.
[BRAD]: So we have eight algorithms that are unsupervised that we apply differently depending on the use cases.
[BRAD]: Then we go through processes and tunings and clusterings to learn.
[BRAD]: Over time, we don't share client information ever by the way, but our ability so there's, you know, there's knowledge and know how.
[BRAD]: Knowledge is just understanding something know how is how knowing how to apply it.
[BRAD]: We have know how.
[BRAD]: And I'd argue that's real intelligence.
[BRAD]: Our our models have intelligence and we have the know how and how to apply it without a client wasting a lot of time tuning models that probably don't make sense.
[BRAD]: So first five years was a lot of trial, error, iterate, and find the models that were best fit for a bank in Spain versus a payment system in Africa.
[MICHAEL]: Right?
[MICHAEL]: Africa, lots of small denomination movements, different culture, different models have different efficacy to find unknown unknowns and reduce false positives.
[MICHAEL]: Applying it to a consumer bank in Mexico or Brazil or Portugal at a centender is a different it's just different models are better like anything in life.
[BRAD]: So I would say you can't just wake up one day and say I'm intelligent, you have to go through looping learning processes to get there and we think that's one of our advantages is the we're ahead at the model layer.
[BRAD]: So it's really, really years of building and refining and tuning Like a hedge fund with no history of performance, you're just not gonna really bet on them unless they have at least three years of track record to say how are you doing versus the market.
[BRAD]: Yeah.
[BRAD]: So so for a lot of individual institutions that build versus buy conversation is really a no brainer.
[BRAD]: I mean, I think a lot of funny you say, I have my board meeting coming up literally writing my page on I have build versus buy, and then I strike versus and I put the word and.
[BRAD]: Because a lot of clients have figured out it's not a it's not a choice.
[BRAD]: It might be a blend.
[BRAD]: Just because I'm buying something from you doesn't mean I don't need to build something myself.
[BRAD]: Like, if I have an agent that I'm working with a client with, my guess is they'll need their own agent platform to do orchestration on their side as well because I'm not gonna do everything for them.
[BRAD]: So they may have three vendors providing agents, and they need to provide something in the middle.
[BRAD]: So I think build versus buy is a false choice and it's more of a and maybe it's more I'll buy you today but I'll build it tomorrow or I need to build something in the middle of what you and I do.
[BRAD]: So, I encourage clients to figure out what they should be building versus convincing them we could build everything.
[BRAD]: And I I would assume that that no organization is or no organization that that is looking for this type of solution is starting from from zero.
[BRAD]: So I'm sure a lot of organizations, when they first reach out to your team, either have something in place that that that's not working or understand that there's an opportunity to really, you know, step it up a notch.
[MICHAEL]: So what what are those first conversations look like when when the client comes aboard?
[MICHAEL]: Where are they at?
[MICHAEL]: You're really, triggering in a good way.
[BRAD]: So there's you know, I'm sure you've heard of rip and replace where you've got this big system Right.
[BRAD]: That you've had for three, five, twenty years, and you're gonna say, now I'm gonna rip and replace the whole thing.
[BRAD]: I see people fully moving away from that idea because it probably doesn't make sense to take your old cable in your house and wiring and doing you should put sonos at like, you're not gonna do it the same way, which we love because we're not able to provide the whole spectrum often to a rip and replace.
[BRAD]: So right now, people are saying it's not rip and replace.
[BRAD]: It's maybe let's take this piece.
[BRAD]: Let's put something on it.
[BRAD]: Let's degrade it, and let's evolve it with two other things we're doing in parallel.
[BRAD]: And in twelve months, we'll meet at something particular, specific.
[BRAD]: Sure.
[BRAD]: Of course, this I mean, these these a failure failed project to rip out the old system and put on some new thing you built or some other vendor.
[BRAD]: Yeah.
[BRAD]: These these are are significant undertakings, obviously.
[BRAD]: And, you know, I guess another question that comes to mind, can can you talk a little bit more obviously, you've worked with so many clients over the years, but it's a very well defined market.
[MICHAEL]: Can you talk about the types of organizations that see the most upside or the most benefit to the Theta Ray solution?
[BRAD]: Well, number one, the clients that have a good handle on their own data because half the battle is getting clients to get the data to a place where we can get to it, analyze it, and then put it back into some workflow loop for them.
[BRAD]: So I would say having a good handle on your own data makes the whole project just easier and more effective.
[BRAD]: It could be obvious, but, obviously, it's not easy to get information together to be ready.
[BRAD]: I would say number two, you know, your ability to take that and analyze it, which is even harder.
[BRAD]: But the reality and again, in my talk, that's how you move from a compliance mentality, check the box, to a risk based view.
[BRAD]: So compliance people don't think as much about analytics as much as maybe implementing rules where risk people always think of an analytic because they're trying to figure out the spectrum of risk.
[BRAD]: I'm not saying compliance people aren't good.
[BRAD]: They're just it's a different way of approaching how you might see a report.
[BRAD]: Compliance reports are like what stuck out, risk reports are there's some risk up or down and they're just different.
[BRAD]: So I would say clients number one that have a good handle on their data, number two, can really understand the output and number three, can work with us to iterate on problem output, and number three, can work with us to iterate on problem solving in a pretty progressive, almost non vendor dynamic like partnership.
[BRAD]: We know we're the client we're the provider and we're not equal to our client at all.
[BRAD]: But if you could work in like a in a development mentality with your clients, especially as things are moving so fast, it's it's kinda critical.
[BRAD]: And we we now have clients, I'd say, in the last three months even that have started to figure out that rhythm with us.
[BRAD]: So, you know, if I understand correctly, I mean, it sounds like banking and fintech are really the core of the portfolio of of clients and that that are seeing the upside to a lot of the Well, I would say banking.
[BRAD]: I think there's forty thousand banks in the world, so there's lots of flavors.
[MICHAEL]: Right?
[BRAD]: So like a bank, we have a great client in Kenya that has a five country presence with their banks.
[BRAD]: And, you know, it's harder to get people to do what they need to do just because it's a slightly less mature mentality across Kenya, Tanzania.
[BRAD]: Like, they're just not as evolved as markets and systems.
[BRAD]: So when you implement there, you know, the level of, knowledge of how to get these things done is limited.
[BRAD]: So you have to do a lot of education.
[BRAD]: In the more, I'd say, bigger financial markets, it's more pulling people away from what they've been doing for so many years already.
[BRAD]: So one is you're having to educate from near zero like a greenfield if you go more emerging market.
[BRAD]: If you're in US, Europe, you're more having to get people away from systems they've been running that way for twenty or forty years.
[BRAD]: So it is quite different depending on the region.
[BRAD]: And I'd say fintechs is too big a bubble.
[MICHAEL]: Right?
[BRAD]: Because everything that's not a bank is sort of a fintech, but we have, like Right.
[BRAD]: Travelex and Shift four as two clients, you know, different vintages, different businesses, and they're all sort of in our fintech.
[BRAD]: But that's and, again, Travelex is a more real time screening person at counter versus somebody that might be waiting a couple of days to open an account in a in an institutional way.
[BRAD]: Yeah.
[BRAD]: And obviously, velocity is is very important in in the space because, let's face it, you know, there's a certain amount of of volume that these organizations need to put through any specific, area.
[BRAD]: So if your organization, your bank is the weak link, you know, you're gonna see a lot of that volume heading in in your direction.
[BRAD]: So it's really important to to not be be exposed, I guess.
[BRAD]: No.
[BRAD]: And we we think I certainly do.
[BRAD]: We are in a supply chain.
[BRAD]: Financial services, you know, financial services people think they are financers of supply chains, but the the the the business that's dealt with supply chain.
[BRAD]: And then if you really believe that there's different forms of money and value that could move between networks, which there are now more than ever, everything is an interconnected supply chain including the financial system itself.
[BRAD]: So that's why so many potential clients exist for us because everybody's gonna have a requirement to monitor.
[MICHAEL]: Like, I'm starting to try to coin the term AVL, anti value laundering from money laundering because money is too limited in what people think of as what we might do where is it Bitcoin money?
[MICHAEL]: If you could move a barrel of oil for a certain amount of cattle, where's the money there if you're doing a direct deal?
[BRAD]: So that's a value transfer you we could certainly launder, or just move something outside of the system tax wise or whatever.
[BRAD]: All of those things need to be tracked now, and I think we figured that out slowly through pandemics and chip shortages and, you know, different energy.
[BRAD]: People figured out chips are needed, energy is needed, and data is needed, and financing is happening for all of those things.
[BRAD]: So it's all obviously interconnected, and the last five years has been pretty illuminating, I think, on that.
[BRAD]: Yeah.
[BRAD]: So as we've gone from machine learning to to AI that we know today, which keeps evolving faster than than anyone can can keep up with it.
[MICHAEL]: How has you know, obviously, the solution itself harnesses the power of AI, but as far as the threat, how has that evolved related to AI and how quickly the tech is moving along?
[BRAD]: Yeah, what I like to do is get, again, a little bit specific.
[BRAD]: So AI to me is the big wrapper of all.
[BRAD]: Machine learning is a field within it.
[BRAD]: It's more algo driving an intelligence kinda outcome.
[BRAD]: When I use the word agent personally, I think about it as the labor and the resource doing something.
[BRAD]: And then to me, the software or middleware is how you might move or an API between an agent action and an alert coming out of an algorithm.
[BRAD]: So Sure.
[MICHAEL]: I think, you know, AI, great, but what does your agent do?
[BRAD]: And I don't think agents have intelligence as agents need to interface with things that are intelligent.
[BRAD]: It's just it's the robot, it's the laborer and that's the cleanest view of the agentic platform we have called Ray.
[BRAD]: But without a model layer and software to move around, but differently depending on who you are as a user or how evolved you or how willing you are to go autonomous or not.
[MICHAEL]: Right?
[MICHAEL]: Human in the loop, human in control, agent in control, human in observation, oracles everywhere.
[MICHAEL]: Like, that's an evolution that you can see happening, but everybody's gonna do it in their own time and own pace.
[BRAD]: And I think Yeah.
[BRAD]: That that the right balance.
[BRAD]: That granularity helps understand the pieces maybe.
[MICHAEL]: So as the you know, when you're looking at a bank exec, for example, as a client, how has the the threat plane moved as AI has really come onto the scene in the last last year or two?
[MICHAEL]: Yeah.
[BRAD]: I would say two, three years ago, AI was a fear and people just didn't get it at all.
[BRAD]: That's changed a lot.
[BRAD]: I'd say nine months ago, the word agent was just not understood, and I'd say now it fully is.
[BRAD]: So just in my six months at the company, I've seen it go from, I get it, but it's too early.
[BRAD]: I get it, but I think I need to talk to I get it, and you've shown me something that is actually quite cool.
[BRAD]: And that's just been a three month arc, I'd say, back.
[BRAD]: First three months, there was a lot of it's a lot of hot air.
[BRAD]: It's a lot of vapor.
[BRAD]: None of this is gonna change my life anytime soon.
[BRAD]: And if you think huge change, you're right.
[BRAD]: But if you think I can do a swiping out of my false positive alerts, I can prioritize my key investigations, and I can drill down on something right now.
[BRAD]: All of that is actually possible today, and we have three, five clients going live now with our agent.
[BRAD]: And we've only been selling our new most modern agent we built, which is about four months in.
[BRAD]: We've already kind of thrown out Platform One, from nine months ago.
[BRAD]: So that's the pace of building and change.
[BRAD]: It it's unbelievable how quickly this is all moving.
[MICHAEL]: So if you were looking from where you sit within the the space, how a few years in terms of some of the the different areas that you're really excited about, what what are those areas that that really excite you as you look towards the next few years?
[BRAD]: You know, again, we take pride in our specialization and our narrowness, which is great.
[BRAD]: We also know we sit in within a much broader fabric of workflow that may go on at any institution, like a full know your client loop or, you know, we are more AML transaction screening, but then there's fraud players.
[BRAD]: And then away from them, there's identity systems.
[BRAD]: And away from them, there are cyber systems that play an identity and fraud.
[BRAD]: So I love these times where there's so many different pieces that look related.
[BRAD]: Nobody can predict exactly how it's gonna combine, you know, conflict or create.
[BRAD]: But it's like you're we know RadioShack was a advertisement physical twenty five things you would go and buy at the store.
[BRAD]: Now all twenty five of those objects are on your phone.
[BRAD]: Happened over thirty years.
[BRAD]: So I just think, you know, we are going to be broader in workflow, broader in case management, broader in customer due diligence, broader in ongoing due diligence, where today we're more transaction monitoring screening and customer risk.
[BRAD]: We're just gonna start to move up and down the supply chain, sometimes organically, sometimes through partnerships, sometimes through merger.
[BRAD]: But people are looking for that spectrum to develop, and I think we're one of the players that will start to move east west.
[BRAD]: And, you know, right now, we're quite north south, but as we go more east west, we just have to pick our spots and not leap too many hops and and sell vapor.
[BRAD]: Well, I think this is really exciting.
[BRAD]: I really appreciate this deep dive this week on the niche within the space
[MICHAEL]: Brad just walked us through his fascinating journey from the early days of fintech to leading a company at the forefront of AI-driven compliance. Now, he's going to dive deeper into how ThetaRay is shifting from legacy rules to a risk-based decisioning infrastructure. It's a vital shift for tackling financial crime, aligning perfectly with our episode's focus on transforming compliance.
--- PRESENTATION ---
[BRAD]: just on the title page here, from legacy rules to risk decisioning infrastructure, just to maybe double click a bit on that.
[BRAD]: So the idea of a rules based system is a bit more, of the legacy of the market and, you know, we tend to move more with unsupervised models.
[BRAD]: I'll explain a bit about that as we go.
[BRAD]: And I would say there's a bigger shift from a a compliance view of the world to a risk based view where you are compliant.
[BRAD]: And then for us, you know, we are traditionally a software and data company.
[BRAD]: I would say those are more generic words, but we really, you know, we are a vertical AI technology company that is infrastructure broadly and hyper specialized.
[BRAD]: So it's a more of a vertical front to back capability that really, enables a lot of, risk management for our clients.
[BRAD]: If you think about, you know, the the flow of capital in the world and, you know, the complexification, I guess, or the in increasing complexity, you have these systems which I would say are software network data people in process.
[BRAD]: This spans many jurisdictions with real granularity, many more products or assets moving around these different flows and, many types of entities that are engaging with those flows.
[BRAD]: So and many more types of entities coming.
[BRAD]: So you have a very complex flow of capital around the world and, it's getting more complex by the second.
[BRAD]: And I would even argue that the word cap the word capital is suspect and it's probably a broader meaning than just money.
[BRAD]: And, there's many, many forms of payments that are evolving into the, into the industry.
[BRAD]: So, you know, what is the root cause of what I would say is the problem in our markets?
[BRAD]: You know, there's a a heavily siloed infrastructure.
[BRAD]: Maybe finance is not alone in having that.
[BRAD]: The compliance world is especially siloed and and, could be quite fragmented.
[BRAD]: Obviously, there's a quite a analog history and a very linear nature to the technologies, and we're moving away from that.
[BRAD]: Number two, there's a volume explosion that's happening, will continue to happen, and will likely accelerate.
[BRAD]: And then number three, there's quite a bit of fatigue.
[BRAD]: And I would say both fatigue at the system level at all those levels I described previously as well as at the human level.
[BRAD]: So plenty of fatigue.
[BRAD]: And, you know, the idea is how do we move into a more digital circular versus analog linear mode as we all do work.
[BRAD]: So just in terms of thinking about, you know, the words infrastructure and decisioning, you know, around all of that, and this word is is certainly out there quite often these days, the word context.
[BRAD]: And having context is key.
[BRAD]: So really, it's it's some maybe raw data that becomes information and you marry that with behavior.
[BRAD]: As you do that and you do that more real time, you can then really start to understand these links across networks.
[BRAD]: And we do that in a very modern, again, more unsupervised way where we're just trying to understand what is going on in a dataset based on behavior, you know, where we try to come at that with very little bias.
[BRAD]: Then what we do is we bring more rules or supervision around those outputs, what we call semi supervised machine learning, where we took a much more unsupervised approach to really go after the unknown unknowns to then move into the more rules based systems where you do have to ultimately implement rules because rules exist, especially in spaces like any money laundering.
[BRAD]: And then beyond that, you then start to move into agentic AI is where you take unsupervised to semi supervised and you really then have something you want to do and you or an agent or you and your agent, can move on that.
[BRAD]: So and that's where we really move to more, I would say, risk based decisioning and decision models versus just, you know, trying to comply with a rule of a threshold barrier of a certain amount, going through a system that, you know, say a certain level, dollars ten thousand or not.
[BRAD]: You could take a much more liberal broad view of of what, of what something nefarious might look like.
[BRAD]: And it really ultimately, this word decisioning is how we describe, you know, having a an insight that you're taking action on.
[BRAD]: So analytics and insights are great, but once you take an action on it, it's not terribly meaningful.
[BRAD]: So we we've sort of used that word decisioning to describe.
[BRAD]: It's just multilayered, multi pronged approach which ultimately gets, into something that's, you know, a real decision making process that is risk based, and and very dynamic, or or, more real time.
[BRAD]: So, and then ultimately, you know, there's a lot of raw information out there, just alerts that are hitting you, in life and this industry is is certainly has that challenge.
[BRAD]: You know, there's analysts that wanna become more of a catalyst and there's, you know, how do you move from detection to actual inspection?
[BRAD]: So for us, we wanna get to decisioning that is quite fast but accurate.
[BRAD]: So you really need to be explainable, predictable, and ultimately controllable where that's where you could be quite on point, quite on target, quite clear, but in a much more compressed time.
[BRAD]: But you can still have that audit ready mentality, for for anybody that, you know, maybe in their loop.
[BRAD]: So, you know, just to maybe, repeat and and bring it, you know, full circle.
[BRAD]: There's, just, you know, a super heavily siloed industry in finance and financial services and whether you look at retail banking, corporate banking, investment banking, or the broader world of payments, and all of those can be, you know, payments inside of many systems, that are not actual banks at all, that are more maybe money movers, payment service providers, point of sale systems, but even, you know, many movements of of value and capital go through marketplaces.
[BRAD]: And we're starting to see, you know, a lot of folks, you know, thinking about how you deploy these solutions for compliance and money laundering observation and lots of different corporates that, again, are not banks or, you know, considered traditional money movers.
[BRAD]: But all of those areas traditionally are heavily siloed and then there's obviously, you know, new disruptors that are more digital first coming in.
[BRAD]: There's quite a few static rules out there and typologies that have been understood, put in some system, and are are quite stale and static almost the moment they exist.
[BRAD]: There's quite about a black quite a lot of black boxing.
[BRAD]: If you think of the explosion in LLMs and what's really going on in there and how do you make it explainable and predictable, it's quite challenging.
[BRAD]: So there's quite a bit of opacity in any of the technologies out there.
[BRAD]: And then again, linear cost scaling, meaning it's just very expensive.
[BRAD]: The more you put on a a system, traditionally, you can reduce your cost and increase your margin.
[BRAD]: But a lot of our industry, especially in the compliance space, is quite linear in its ability to support it.
[BRAD]: So how do you take a compliance but make it more risk adjusted and move to really a unified view?
[BRAD]: So if you're going to talk about risk versus compliance, you need a really holistic view, and you need it to be much more dynamic in real time.
[BRAD]: So when we say we're moving to a risk based decisioning infrastructure, it really is that real time unified nature of needing to know where your risk is at all times.
[BRAD]: So it's, it's a real, unified risk view.
[BRAD]: It's a dynamic and flexible number or system or analytic.
[BRAD]: It's very explainable and clear.
[BRAD]: And the more volume you put through it, actually, it gets smarter.
[BRAD]: So again, as you understand and take more risk or see more transaction, your actual management can improve because your intelligence of risk improves, your confidence improves, and therefore your business velocity can increase.
[BRAD]: So the more data you see, the more you know the more you know, the more risk you can take.
[BRAD]: And if we move from a compliant rules based, more static siloed view of the world to a more dynamic risk based dynamic view, we're going to be able to do more business more quickly, on effectively the same or less or take more risk just because we understand it better.
[BRAD]: So if I, you know, just think about how to how to bring this all together, and play off of a few words, There's certainly some real blind spots out there, and our industry talks about known knowns, known unknowns, and then the more elusive unknown unknowns.
[BRAD]: And that's the real challenge is how do you really understand the blind spots, see around the corners when, to be honest, we're not very good at being predictive, we're quite good at looking back and doing something in the future that understands what we did yesterday.
[BRAD]: But we definitely have to move to really more technologies that allow unknown unknowns to exist.
[BRAD]: We find them, we deal, and then ultimately we resolve again on a risk adjusted basis.
[BRAD]: So part of this is how do we get there before regulator and effectively meet the criminal or the financial crime network at the line of scrimmage.
[BRAD]: So you may not think you can get ahead of these folks to some degree because they're untethered and, you know, they're looking to do bad things and get around systems.
[BRAD]: But at a minimum, if you use the same technologies, you can sort of meet there and, help us all, including the regulators, you know, not be caught off guard.
[BRAD]: Ultimately, the word audibility and board transparency, you know, it sounds very high level executive c suite into board, but I would argue this is auditability and transparency from the line worker doing the investigation, seeing the alerts, to the boss and the manager of that line, to the executives that oversee that area of the business, to the ultimate board.
[BRAD]: So number two is really a full auditability and board transparency, you know, for all in an organization front to back and the confidence creator.
[BRAD]: And then ultimately, driving a lot more operational velocity.
[BRAD]: And again, I would add the word business velocity to the operational velocity because the whole point is to obviously, to some degree, manage and control your cost.
[BRAD]: More importantly, understand your risk, and then really, more interestingly maybe, is is really manage your foot forward.
[BRAD]: So cost management, risk management, growth management, the compliance industry, the financial crime space is starting to move from a cost center thinking about compliance to a revenue generator thinking about risk.
[BRAD]: So just to close-up here a bit, you know, as a specialized risk infrastructure focused on the financial crime space, we really believe that our model layer brings a unique level of intelligence.
[BRAD]: Our software layer brings a level of efficiency to getting at that intelligence.
[BRAD]: And more recently, Ray platform, our agentic layer, makes you very effective.
[BRAD]: So intelligent, efficient, and effective where we are focused on, you know, the main client outcome is allowing them to do more business at less cost with a really clear view of risk.
[BRAD]: So I will leave it there in terms of my formal presentation.
[BRAD]: Happy to jump back on with, Michael and chat about.
[BRAD]: Fantastic.
[BRAD]: Well, I think, a lot of the different, executives that are at the event today will, take take note of a lot of the details in your presentation.
[BRAD]: And one of the things I know coming up in just a few minutes, you're joining us on the next, live q and a session.
[BRAD]: So for everyone on the live call today, you have a q and a box in front of you.
[BRAD]: Type in any questions that you have for Brad in that q and a box, and we'll be doing our best to get to as many of those as we can during the live session.
[BRAD]: But I'll be honest with you, Brad.
[BRAD]: One question that comes to mind on this end.
[BRAD]: I know that, overall, you know, compliance is an important part for a lot of organizations.
[BRAD]: But at at an executive level, I think a lot of the different individuals attending the event today may not even be sure where to start.
[BRAD]: You know, this is a very broad topic in terms of something that that impacts a lot of organizations.
[BRAD]: But what is that that first area of of engagement that that you see for a lot of a lot of clients?
[BRAD]: Well, I would, maybe suggest, again, a cute statement.
[BRAD]: You know, where's the best place to start maybe at the beginning?
[BRAD]: And by that, I mean, you know, we are a provider to people that have clients.
[BRAD]: We have clients ourselves.
[BRAD]: We all speak more on behalf of maybe a bank.
[BRAD]: So if you're a new bank with no clients, the first thing you wanna do is onboard a client.
[BRAD]: So the best place to start is what is your process for that initial customer due diligence, that initial onboarding that you're doing.
[BRAD]: So it's a great place to start to say, let's make that really tight and robust.
[BRAD]: Number two is sort of what's your ongoing relationship with that client and how do you observe what they're doing.
[BRAD]: And then over time, you still have to refresh and collect what they're doing.
[BRAD]: I'd even argue sort of from birth to death, you know, as you may off board a client, and we have requirement all over the world now like Dora and GDPR that we have to cleanse and say we never knew you ever even though you've been with us forever.
[BRAD]: So I think it's like, you know, from a compliance perspective, instead of thinking about it as a moment in time that you go away and maybe called in later, it's a upfront ongoing and ending engagement of how you think about this.
[BRAD]: So if you think about that, again, less linear, more circular, you're not gonna allow data you collected upfront to drop or for another part of the organization that could share that to benefit from it.
[BRAD]: Now sometimes it's not easy to share information across parts of your organization, but because we're so siloed and our clients are siloed, it's not easy to actually get that information across.
[BRAD]: So upfront, we collect information many times that may not make sense.
[BRAD]: We then look at the ongoing due diligence as a separate process from the upfront due diligence.
[BRAD]: And then we think of off boarding as annoying because sometimes a client goes away or fires us, and we don't love to comply as easily when people go away from us.
[BRAD]: So I think it's like thinking about all of those as loops that link versus linear moments that never connect.
[BRAD]: Well, I'm sure we'll we'll hear a lot of lot of questions coming in from the audience, so we'll get to those in just a few minutes, but I think that, gives us a very good, starting point.
[BRAD]: So thank you, Brad.
[MICHAEL]: Brad's presentation really underscored the shift from static compliance to dynamic risk-based decisioning. Now, let's hear from the live panel at the Software Oasis AI Summit, where you'll get answers to real audience questions that explore this transformation further.
[MICHAEL — CHECKPOINT]: Brad's presentation made it clear that moving from static compliance to dynamic risk-based decision-making is not just beneficial, but necessary for modern finance. I think many of us are now wondering how to practically implement these changes and what challenges might arise. And that's exactly what this audience brought to the table during the panel.
--- Q&A PANEL ---
[BRAD]: okay for everyone that's been with us live since first thing this morning this is our second panel of the event uh with us we have all of the speakers that you've heard over the past little bit and uh we will do our best to get to as many of the questions that have come in to this point uh in the event during this panel time but first i want to start off by introducing everyone on the panel.
[BRAD]: We have joining us from Amdocs, Paul Soleil, who is the head of AI growth.
[BRAD]: Welcome, Paul.
[BRAD]: Thank you, Michael.
[BRAD]: Next from LinkSquares, Bill Hewitt, who is the CEO over there.
[BRAD]: Welcome, Bill.
[BRAD]: Thanks, Michael.
[BRAD]: Good to be here.
[BRAD]: And finally, Louis Pinedo.
[BRAD]: He's the chief strategic customers officer over at ThetaRite.
[BRAD]: So welcome, Louis.
[BRAD]: Thank you, Michael.
[BRAD]: Happy to be here.
[BRAD]: Pleasure to have everyone.
[BRAD]: And we have had a ton of questions come in.
[BRAD]: So I want to do my best to get to as many of these as we can in the limited time that we have.
[BRAD]: I do want to point out up at the top of the site, you'll see that big orange button.
[BRAD]: For anyone that would like to reach out to any of the panelists here or any of the speakers you've heard throughout the rest of the event just hit that button we'll get you connected or if you'd like to reach out to anybody on the software oasis team that button will get you there so why don't we get started uh paul it looks like the first questions came in for you so i will start with uh tessa who is in richmond Virginia here in the US.
[MICHAEL]: And Tessa's question is, how do you prevent shadow AI, shadow IT, and shadow data from undermining governance?
[MICHAEL]: Oh, hi, Tessa.
[MICHAEL]: Good question.
[MICHAEL]: Absolutely.
[BRAD]: Look, shadow AI, shadow IT, and yeah, shadow data, in my experience, they tend to be a product of organizations that are trying to hold the offerings to the business very tightly.
[BRAD]: And what ends up happening is that the business rebels, goes out, does their own thing.
[BRAD]: They either put it on their corporate Amex or they just start doing their own thing.
[BRAD]: So in order for organizations to really have governance wrapped around the totality of their organization, they actually need to be more liberal with the services and offerings that they're giving.
[BRAD]: So they need to be able to understand the business need and be able to be responsive to that business need.
[BRAD]: Because when they're not responsive to the business need is when those lines of business or divisions go out and start doing their own thing, breaking that governance model.
[BRAD]: You know, it's very difficult once those groups start doing that to actually bring them back in.
[BRAD]: And it ends up being, frankly, a large migration or a large change activity to bring them back into the overall model.
[BRAD]: So in my opinion, the IT departments and the technology departments need to be more flexible with what they're offering.
[BRAD]: Obviously have the proper governance, have the proper security controls, but be able to be responsive to the business and what they're trying to achieve.
[BRAD]: Makes a lot of sense.
[BRAD]: The next question came in to us for Bill.
[BRAD]: And Bill, this question comes to us from Mason, who is just outside of Charlotte, North Carolina.
[MICHAEL]: Mason is asking, what is the best first contract workflow to automate with agents?
[MICHAEL]: Yeah, great question, Mason, and thanks for asking that.
[MICHAEL]: You know, if you think about contracts, contracts touch every single department and every piece of data within an enterprise.
[BRAD]: And what we always recommend for our clients is get a good handle on your contract data first.
[BRAD]: get your repository in place so you have a place for all your contract intelligence.
[BRAD]: Once that's done, then you can start building workflow out to the stakeholders within the organization.
[BRAD]: But if that foundation isn't there first, it makes all the adoption in the rest of the business very difficult.
[BRAD]: So get everything in one place, make sure you have your current contracts, amendments, schedules, exhibits all in one place.
[BRAD]: And then it makes pulling information out of those contracts a lot easier.
[BRAD]: Louis, this next question came in for you, and this is related to the Theta Ray presentation.
[BRAD]: And I want to try to get to some of our international audience.
[BRAD]: So let me grab this question.
[BRAD]: This is from Felix, who is in Tel Aviv, Israel.
[MICHAEL]: And Felix is asking, How can agentic AI improve investigations while preserving human control?
[MICHAEL]: Great question.
[BRAD]: So basically, the way that our agents is taking into consideration what is the existing process that our customers, our banks or fintechs are processing.
[BRAD]: So mainly what we do is to train our engine in the way that what is the protocol that has been defined by the customer.
[BRAD]: So as you can imagine, nobody's training is allowing us to really test that the machine is not going to fail.
[BRAD]: So the main benefits that we see anytime that we are improving this, we are proving this to basically.
[BRAD]: First of all, we are reducing a lot at the time of the investigation, which is always something quite important for the customer.
[BRAD]: But at the same time, and even more important from my perspective, I'm coming from a compliance background, we can make sure that the process that we are doing in these investigations is exactly what the bank is expecting.
[BRAD]: You know, normally when you are working with people, it could be like a fatigue or people that is not properly trained, could not follow in exactly what we are looking for.
[BRAD]: With this approach, after iteration and iteration with the hundreds or even thousands, we can make sure that the result that we are getting are what we are expecting.
[BRAD]: And what we are seeing with our customers is that the external auditor, internal auditors are checking that the object we are getting is much better from a quality perspective than what we have just a human in the loop.
[BRAD]: By the way, always we are working with ideas, but the last decision maker will be the human.
[BRAD]: Sounds like a very smart workflow.
[BRAD]: Bill, the next question came in for you, and this is from Callum, who is in Manchester over in the UK.
[MICHAEL]: And Callum is asking, how do you identify overlapping systems before deploying Agenda workflows?
[BRAD]: That's a really, really good question, Callum, because right now we like to say with AI, everyone's doing more with more.
[BRAD]: They're putting AI on top of their existing systems instead of looking at what systems they can replace.
[BRAD]: We're actually in the process of looking at all 163 systems we have in link squares we use to run the business to figure out which ones are doing similar functions, but which ones can be eliminated through AI.
[BRAD]: So we're starting first with information management because today we pull information from lots of systems, use ETL to pull it into a data lake, then use BI tools to access that data.
[BRAD]: If we can pull data directly from the source systems, those are the first systems that we can eliminate.
[BRAD]: And then it just works from there throughout the business process.
[BRAD]: And we're actually redesigning our business processes at the same time.
[BRAD]: So as we go through that, we think we can get the number of packages we use down to below 50.
[BRAD]: Fantastic And obviously a lot of these answers come from personal know working with customers working with internal systems and best practices The next question comes to us from Brooke who is in Houston Texas And this question is for Paul.
[MICHAEL]: Paul, the question is, what controls should be in place before moving an AI pilot into production?
[MICHAEL]: Yeah, thank you for the question, Brooke.
[MICHAEL]: The main control that needs to be in place is the observability.
[BRAD]: So observability, tracing, because in my view, and you would have seen from my presentation, one of the biggest things around adopting AI at scale is trust.
[BRAD]: So if you don't trust the answers of the AI, you don't trust the actions that it's taking, there's no way to scale it from pilot to production.
[BRAD]: So the ability to observe those actions, to observe those decisions, observe why it's making those decisions based on the input data it's taking in and what actions it's taking is the key governance piece.
[BRAD]: Makes a lot of sense.
[BRAD]: And the next question came in for Lewis.
[BRAD]: And Lewis, this question comes to us from Gavin, who's here in the U.S.
[BRAD]: and just outside of Philadelphia, Pennsylvania.
[MICHAEL]: And Gavin's question is, how can firms find financial crime risks that static rules miss?
[MICHAEL]: Well, you know, the industry is working with rules from a long time ago.
[BRAD]: And actually, what we brought from TETARAY is a different approach.
[BRAD]: different approach.
[BRAD]: I think the difference that we have with the other players that we have in the industry is that our detection is based on machine learning, particularly unsupervised and semi-supervised.
[BRAD]: And the best what you can get when you are using this approach is basically that you are looking for anomalies.
[BRAD]: In a rule-based system, you are defining you know what is the rule what is the threshold and everything that is uh above the threshold is alert what is below the line you are not getting anything so the approach that we are following is not uh we are not uh let's say have any kind of bias we are not having uh necessarily different scenarios per type of risk we are seeing the customer like a 360 degree analysis with different features controlling different ways particularly we are not using threshold so so basically that means us to to to identify the behavior of the customer one by one in order to avoid expectation and these kind of things and at the end what we are seeing in the implementation that we we complete with our customers and could be very big customers big banks is that as a result we are able to catch these kind of behaviors that normally are moving below below the line and we can reduce dramatically the number of others that we are we are so i'm sure that still you know the regulation is is not easy to to to affect this kind of approaches but i am seeing how the industry and working in this world for 16 years in in this particular world now are moving to the to the right direction and this is the way to to create a more or let's say secure environment for our customers and for the individuals that actually are using the banking services.
[BRAD]: The next question comes to us from Chloe, and this is for Bill.
[BRAD]: This is Chloe, who's in Salt Lake City, Utah.
[MICHAEL]: And Chloe is asking, how do you move from configurable agents agents to custom agents safely?
[MICHAEL]: Yeah, it's a great question.
[MICHAEL]: And AI is making everybody sort of think about the guardrails that they have in their applications in their business.
[BRAD]: And we actually believe this is a way to help people understand how best to make that move.
[BRAD]: And you start with a Configurable Agent where you really can't hurt yourself.
[BRAD]: You can have outcomes that you don't want, but you'll see the ability to roll things back the ability to make sure that you have a two-stage commit before you do anything that might affect the system in a negative way.
[BRAD]: That's just really good training for understanding how agents behave.
[BRAD]: And then customizable agent is something like a workflow where I'm not actually changing the agent itself, but I'm changing what the agent does in a very controlled way.
[BRAD]: And again, having gone through the configurable stage, you learn more about what works, what doesn't.
[BRAD]: And we provide training along the way as well, things to watch out for.
[BRAD]: But we also put guardrails in the system where somebody does something that we know is going to have a negative outcome, some kind of negative outcome.
[BRAD]: We'll stop them, ask them to refine what they're trying to do.
[BRAD]: So we make sure they get it as close to perfect the first time.
[BRAD]: When we get into custom agents, that's a whole new ballgame.
[BRAD]: And that's where we want to have experts either at the customer who have been trained and certified or within link squares, make sure that those agents are deployed by us to ensure that they provide the maximum value for the customer.
[BRAD]: The next question came in for Paul.
[BRAD]: This is, there were actually a few questions similar, so I'm just going to pick one.
[BRAD]: This is from Ian, who's in San Diego, California, right here in the U.S.
[MICHAEL]: And Ian is asking, how can enterprises create a common AI foundation across separate business units?
[MICHAEL]: Thank you for the question, Ian.
[BRAD]: This is actually something that I deal with day in and day out.
[BRAD]: So within our organization, we have our own framework that does just that.
[BRAD]: So having the common elements of governance, of agent management, of MCP management, of all of your observability, your reporting, and so forth.
[BRAD]: And I actually think it's the right way for enterprises to do it, leaning on a particular technology stack, stitching a few different ones together to get the right outcomes for them, but then expanding the use of it.
[BRAD]: So very much like if I can relate it to software development, you know, in the past, there were multiple teams that were using their own CICD pipelines, their own different tool sets, their own everything.
[BRAD]: But you're only getting the right outcome when you're centralizing that CICD pipeline, you're centralizing that SELC in the deployment process, because at that point, you can govern it all.
[BRAD]: If you're still distributing that platform across multiple teams, there's no central governance.
[BRAD]: There's no standards that could be held.
[BRAD]: So we're doing the same thing with AI platforms now, where once you centralize that platform, then you're having that hub and spoke model where the teams are consumers of that platform and they're designing their own agents and even having an agent to agent framework to go across multiple domains within the organization in a safe way.
[BRAD]: But you can only have those things in place with that centralized platform, the centralized governance and the centralized controls, obviously at a level that's not too restrictive on the business, because then you get the first question I answered, which is you start getting your shadow AI and your shadow data and shadow IT.
[BRAD]: but it has to be an enabler for the business to move very quickly, reuse agents across multiple areas, changing configurations, changing the knowledge base, changing the context that they consume, but having that reusability and having that common platform.
[BRAD]: So, yeah, that's the best way I've seen AI developed at that enterprise scale.
[BRAD]: Lewis, the next question comes to us from Alina, who is in Las Vegas, Nevada, here in the U.S.
[MICHAEL]: And Alina's asking, what makes risk decisioning explainable enough for auditors?
[BRAD]: That a great question particularly here I think in the US And this is probably one of the the always the big questions that uh are coming from from our customers because you know when the the customers that can come from a rule-based system super simple to explain a rule which is basically a plus b is greater than c then i get an alert we uh the company uh has spent a lot of time and deploying a methodology where we are explaining in a risk language how our system is working so basically our approach is starting to to listen in what the regulator are expecting that the banks are will be executing us in the control environment we identify those documents we identify the red flags that are included in those documents and we are going to start to to speak in in these languages measuring what is the inner risk considering the products and services that our customers are running also volumes and then we are going to set up different features and starting to speak in a matching learning language different features that are going to be mitigating different uh risk as a result of that then we we have our own uh elbows uh and then we will get an alert and the alert is actually telling us by the trigger feature which is the the most important risk that is related to the alert so once the analyst is actually in front of the of the of the alert he said we'll really understand why the other other is raising whatever could be related to cars to international payments to high-risk countries etc etc on top of that then of course We have been passed through different model validation governance exams in some of the tier one banks that are doing business with us and also external internal auditor has been passed through this process.
[BRAD]: So it's a journey, of course it's a journey, but there is an explainability that we are able to complete successfully in different tier one banks that we are doing business.
[BRAD]: It looks like this next question came in for Paul, and this is from Omar, who's just outside of Cleveland, Ohio.
[MICHAEL]: And Paul is asking, what should executives look for when selecting an AI implementation partner?
[MICHAEL]: Oh, Omar, thank you for the question.
[MICHAEL]: That's a big one.
[BRAD]: that's actually going to take me a minute so what are the important things about selecting an AI partner these days look I've had a lot of conversations with enterprise executives around AI and the feedback that I've gotten from them and I think I'll use their words rather than my own for this is you're really looking for a partner that understands your business because AI is very contextual.
[BRAD]: And while your business brings, like your business users and your SMEs bring a lot of context to the AI, you want a partner who can understand your business as well and understand the challenges that your particular business and your industry has.
[BRAD]: The other one is you want a partner that can actually dive deep into how AI works and what the roadblocks are and what the challenges are.
[BRAD]: And what I've heard from a lot of executives I've spoken to is that there's a lot of slideware in the industry right now.
[BRAD]: So a lot of people talking about how AI can be done or potentially, yeah, how it can be done.
[BRAD]: But the hands-on knowledge and the scars that you get from continuous delivery and actually doing this stuff in practice isn't talked about a lot in the industry yet because there aren't a lot of organizations that have actually deployed AI at scale.
[BRAD]: So definitely as an executive looking for the individuals that understand not just the benefits of AI and what the potential of AI is, but also what the drawbacks are, what the challenges are, and strategies for overcoming those challenges.
[BRAD]: So those are the two areas that I would really look at as an executive looking for a partner.
[BRAD]: The next question comes to us.
[BRAD]: This is for Bill, and this comes to us from Andre, who is in Brunswick, New Jersey.
[MICHAEL]: And Andre is asking, which metrics prove that agentic contract management is delivering real ROI to an organization?
[MICHAEL]: Yeah, it's a great question because like many technologies, people become enamored with the technology and don't always look at the ROI.
[MICHAEL]: The great news is everything about contract lifecycle management is process oriented.
[MICHAEL]: So I can look and do time in motion to say, how long does it take me to draft a contract?
[MICHAEL]: How long does it take Agentec to draft a contract?
[MICHAEL]: How long does it take me to negotiate a contract?
[MICHAEL]: And what it typically does is reduces the throughput to outcome versus necessarily speeding up any one part of it.
[BRAD]: So, for example, inside of LinkSquares, we use our own product.
[BRAD]: Over 60 percent of our contracts never touch a lawyer inside the company because we have good clause libraries, good playbooks.
[BRAD]: We can actually negotiate fully a contract with a counterparty, and that's all managed by the sales team, not by the legal team.
[BRAD]: So we've reduced the amount of time legal spends on contracts over, I'd say, the last two years from over nine hours per contract to 4.2 hours per contract.
[BRAD]: So we see direct and immediate cost savings to doing that.
[BRAD]: But more importantly, we're more responsive to customers and we can drive more opportunities through the business.
[BRAD]: Now, this next question is an interesting one.
[BRAD]: And this is for Louis.
[BRAD]: And this comes to us from Connor, who is located just outside of Milwaukee, Wisconsin, right here in the U.S.
[MICHAEL]: And Louis, what Connor is asking is, how do you unify siloed data without delaying fraud investigations?
[MICHAEL]: Yeah, that's a good one.
[MICHAEL]: That's a good question.
[MICHAEL]: So, yeah, right.
[MICHAEL]: coming there although you know AML or money laundering activities and fraud are well connected okay so you know although the data rates super focusing money laundering activities and not too much in fraud but yes particularly the US both both areas are working together and are well connected.
[BRAD]: What we have seen in different implementations that we completed is that our solution is also really connected with the information that potentially our customer could bring from front.
[BRAD]: So our for instance approach is in some of the detection engines that we are setting up for our customers, for instance in the customer due diligence process and the ongoing due diligence process, we are taking into consideration also information that is coming from fraud could be alerts that could be you know false positive alerts or just what means that could be related to control at any time we are taking this into consideration and actually it's true that the industry not only in the us it's also it's also moving in this direction in europe more and more now the banks are starting to see this like these two topics that some years ago could be just totally different and starting to see like more or less the same the same thing Fantastic.
[BRAD]: I know we have a few minutes left.
[BRAD]: I want to try to get to as many of the questions as we can.
[BRAD]: A lot of individuals that are at the event today will have questions that we may not get to during the session At the top of the site you see that big orange button That get you connected to any of the speakers that you heard across this panel or during the event I know a lot of individuals like to follow along in the series from one event to the next.
[BRAD]: So we have several really exciting events coming up over the next.
[BRAD]: I hate to say it.
[BRAD]: Here we are.
[BRAD]: It's right up on Q4.
[BRAD]: So we have quite a few events that are coming up over these next three months in Q4.
[BRAD]: And I'll just give everybody a quick overview of what's coming up so that people can register for any of the events that they like.
[BRAD]: The very next event that's coming up in October is the AI Talent Acquisition Summit that is scheduled to be live the first Thursday in October, which is October 1st.
[BRAD]: Right after that, we'll have our Enterprise Security Boot Camp, which is coming up in November.
[BRAD]: the November event.
[BRAD]: We have quite a lineup for this, and it's an event that we've touched on some of these different areas that we'll be covering during this, but it's a full-day expert-led event for security leaders focused on AI risk within their organization.
[BRAD]: And finally, coming up in December.
[BRAD]: I hate to say it, but rounding out 2026, we'll have our most awaited topic.
[BRAD]: We touched on this at some earlier events and had a lot of interest.
[BRAD]: So this is a full day where we're discussing AI identity and security.
[BRAD]: So we're really covering agents and human identity and all of the intersection in between and a lot of really important topics on this event.
[BRAD]: So that gives everybody a feel for what's coming up over these next few months.
[BRAD]: And I'd like to jump into some of the final questions that came through that we'll have time for.
[BRAD]: I will toss this next question out to Bill.
[BRAD]: This is from Natalie, who is in Kansas City, Missouri, right in the city there.
[MICHAEL]: And Natalie is asking, how can legal teams automate NDAs without losing control of risk?
[MICHAEL]: It's a great question.
[MICHAEL]: And the NDA is the most drafted, negotiated, and signed contract in the world.
[MICHAEL]: Also, the least litigated, if that's of any interest.
[BRAD]: But it's one of these things where I think you really have to make sure that you understand what you're trying to control.
[BRAD]: NDAs can get very complicated and they don't necessarily need to be.
[BRAD]: The reality is that if you've got a good system that keeps the sender or the writer within the confines of the system, so they can't take it out into a Word document and make changes and send it that way, that's the easiest way.
[BRAD]: That's not always possible.
[BRAD]: there's got to be a review process before it goes through signature to make sure that matches all the current clauses that exist in the playbooks of the company.
[BRAD]: But it is, as I said, the most negotiated document, also the one that gets changed the most often without legal input.
[BRAD]: Interesting.
[BRAD]: Interesting.
[BRAD]: A few, few interesting details there.
[BRAD]: Paul, the next question came in for you.
[MICHAEL]: This is from Aisha who's joining us from Singapore and Aisha is asking, How can organizations prepare securely for agents that work across company boundaries?
[MICHAEL]: Thank you, Aisha.
[MICHAEL]: That's a very big one.
[MICHAEL]: Yeah, so the industry is not quite ready for that yet.
[BRAD]: I'll be completely honest.
[BRAD]: What's going to end up happening is a couple technologies that I've seen in publication, in research in patent right now are going to help with that.
[BRAD]: And it's going to be a mix of being able to distribute private and public keys and be able to validate that through across organizations.
[BRAD]: So I know this is a challenge I'm sure all of us in contract and finance, especially AML, are facing into as well, because there's potentially a lot of fraud that happens when you start having agents communicate across different organizations, especially when you start, you know, passing transactions around or negotiating on contracts.
[BRAD]: So it's definitely something the industry is now looking at.
[BRAD]: I don't have a great clean answer in the couple of minutes I had to respond, but it's something that we can discuss further if you want to reach out.
[BRAD]: Fantastic.
[BRAD]: And as we're rounding out this Q&A time, Lewis, I'll get the last question to you.
[BRAD]: This comes to us from Jasmine, who's in Sacramento, California.
[MICHAEL]: And Jasmine is asking, where should a financial institution begin with continuous due diligence?
[MICHAEL]: This is a great question.
[BRAD]: So I think this is probably one of the toughest challenges that the banks are facing, particularly the big banks, because you know the ongoing due diligence is something that is uh if you continue working in let's say in the old-fashioned way it's it's it's taking a lot of people so we have a lot of people of corporations taking care of uh of the customers so one of the things that we brought uh in in tetherray is the concept of the perpetual kwec which basically means that uh our implementation allowing the banks to have an environment that is controlling the customer 360 views which basically anything that could happen related to the customer could be a broader layer it could be a transaction one to another could be something related to a post media could be changes that even the customer is providing to me anytime that this is happening we have an environment or a perpetual camera the environment actually what it does is automatically take this data and also reconsider the risk of the customer not only taking into consideration the static data that the customer is provided at the beginning of the process but to actually consider the information that the banks are observing in the way that the customer is is acting which is super smart because probably in the past the risk of the customer was based on the information that the customer is providing in the process of the onboarding or in the ongoing.
[BRAD]: Now with this approach, it's just probably 60-70% of the risk assessment of the customer will be based on what the bank is observing in the way that the customer's behavior with the bank, with the channels, in the branches, etc.
[BRAD]: So I think this is the next move.
[BRAD]: And we will see that a lot of banks that are moving in this direction are getting significant reduction in the cost and also a better, let's say, tracking in the process that any time that something happens is able to track that this change has been recorded and there is an action from the bank.
[BRAD]: Well I would like to say Paul, Bill, Lewis, thank you so much for all of your time preparing for the event today and for your thoughts shared on this panel.
[BRAD]: Some very important topics for many executives within the community and governance, risk and compliance are certainly areas that every executive and every organization is working to get their hands around and it's not an easy ask.
[BRAD]: So thank you so much for all of your time and insights on the panel today.
[BRAD]: Thanks for having us.
[BRAD]: Thank you, Michael.
[BRAD]: Take care.
[MICHAEL]: So, a few key takeaways: first, embracing AI and risk-based decision-making can dramatically enhance financial crime detection. Second, it's crucial to move beyond traditional, siloed compliance systems. And third, transparency in AI systems builds trust and efficiency. By the end of our talk today, I hope you've seen why these shifts are vital for the future. Brad, thank you so much for sharing your insights. For anyone wanting to connect with Brad, you can find him on LinkedIn. And if you haven't already, come join us over at softwareoasis.com — that's where all of this really comes together. We'd love for you to subscribe to the podcast and leave a review if you found this valuable. Take care, and talk soon.