About NCFA Canada | Craig Asano | November 6, 2026

Featured Guest: CHRIS MARANIS, Deputy CEO, EXUS (LinkedIn)

Links

• EXUS Website | LinkedIn
• Chris Maranis on LinkedIn

Contact

General enquiries: m.legat@exus.co.uk

Headquarters: 4 Felstead Gardens, Cubitt Town, London E14 3BS, UK.

About this episode

EP67: Can AI Replace Debt Collectors?

Duration:  66 mins

Subscribe and tune in each Friday to check out the latest movers and shakers in fintech. Listen to more podcasts here:

Season 1 | Season 2 | Season 3 | Season 4 | and weekly newsletter

Fintech Friday Transcript of Episode 67:

CHRIS MARANIS, Deputy CEO, EXUS (LinkedIn)

Intro: Welcome to fintech Friday's a weekly podcast brought to you by the National Crowdfunding and Fintech Association of Canada and partners. Covering all things fintech, blockchain, AI and alternative finance.

[00:00:29] Craig Asano: Hello, everyone. Craig Asano here, the founder and CEO of NCFA Canada, welcoming you to season four, episode 67. Cannot believe we're already at episode 67 of Fintech Fridays. It's a weekly podcast brought to you by NCFA and partners where we feature conversations with leading voices across fintech, funding, and financial innovation around the globe. We talk with founders about their journeys, executives, companies. We talk about emerging trends, financial innovation, and what those experts are seeing in the market right now. So today we have another fantastic guest with us today. I'd like to introduce Chris Maranis. He's the deputy CEO at EXUS, a long-standing leader in the global debt collection tech space. Since joining EXUS in 2004, he's worked with financial institutions across more than 30 countries, helping to modernize collections, operations, all the way from pre-collection, pre-collection to recoveries. As part of EXUS' executive leadership team, Chris helped shape the company's strategic direction by connecting market needs with product vision, commercial growth, and operating execution.

[00:01:40] Craig Asano: He also works closely with customers, partners, stakeholders on bringing about more intelligent, ethical, and customer-centric approaches to collections. Drawing on more than two decades of international experience, his current focus includes AI, data-driven decisioning, and modern engagement strategies that can improve outcomes for both lenders and customers. Chris, thanks so much for joining the podcast today to share your experience and knowledge.

[00:02:05] Chris Maranis: Welcome to the show. Thank you very much, Craig. I'm very happy and excited to be with you here today. Well,

[00:02:10] Craig Asano: I think we're just gonna kick off with an opening question. I mean, you've spent over 20 years according to your bio, at EXUS working with lenders all around the world. What first drew you into the business, and what has kept you interested for so long? Yeah. That, takes me back, but to be honest from the start, it was not collections back then.

[00:02:32] Chris Maranis: So in 2004, I joined EXUS, the company I still work for. I, I think I just wanted to join a company which since then had a reputation that takes extra good care of its people, and it was growing fast. So that was my aim. And collections was, let's say, a side effect, an accident. So I joined EXUS, and I was assigned as my first assignment to be the project manager for the implementation of the debt collection solutions for a thriving bank back in Greece at that day. And I have to say that what, what really made me... And I think if I look think what has made me stay after, you know, 22 years in collections is that in contrast to what people think, people who are not in collections, collections touches people at a difficult point in their lives. And I saw the struggle a lot of collections teams were going through to, on the same time, achieve their business outcomes but help people in distress get out of it. And it was a shock for me. It, it wasn't what I was expecting collections is like. You know, when you're outside collections, you have this caricature of a collector in your mind, someone who is just, you know, making phone calls and pressing on people to pay.

[00:03:44] Chris Maranis: It's a completely different game. And when I got into it, I saw it's, it's challenging, it's complicated, it needs a lot of skills, and it does involve a lot of wonderful people who I still enjoy working with. So that's the thing that what has kept me all these years. Also, the fact that EXUS, the company I work for, is some kind of unique in, in the way that has a culture of constant change or growth. In these 22 years, I have, I have changed multiple times my role as a person. I have grown, and that's something that I think has kept me all

[00:04:18] Craig Asano: this time. Never had a boring day, Well, it's a remarkable journey since 2004, going from you know, project management really with, with the product on the operational side all the way to a deputy CEO, so congrats to, to that journey. You know, take us back a little bit to the time that you joined EXUS so we could get a sense for what collections were like back then and what EXUS as a company was like. We're gonna sort of provide some background because listeners might not really

[00:04:49] Chris Maranis: know much about Indeed. So back in 2004, collections was quite different than what it was today. To be honest most lenders, most organizations were considering collections as a back office operation. Even worse it was considered by some organizations as let's say, a place where you can exile your bad customers. Far from truth, of course, but that's how it was treated back then. And sometimes collection departments were located in some obscure locations or offices, so people would not know what was happening inside. Operationally collections was mostly a team of people in a call center handling large calling lists and a dialer and making, just trying to maximize the number of calls they make each day. There was some queuing, some filtering or prioritization, but very basic. And technology-wise since, you know, we are technology company the number one system I think that was used was the almighty Excel. Probably the best debt collection system of all times.

[00:05:59] Chris Maranis: Big organizations, huge organizations were... have had, you know, had tens of Excels and a lot of emails around no system a big high, a high turnover in the call center, so not an ideal situation. Back then in 2004 EXUS had already introduced its first version of the collection system, the ancestor of the system we currently have, and I remember the name was eCollect back then, and it kind of was rather, let's say, advanced for, for that age. But, you know, comparing with to what we are doing right now, you know, the multi-channels, the artificial intelligence, all the thing we have now, it was, rather old, cumbersome, and not as intelligent for sure as systems are today. So yeah, going from Excel to, you know, the,

[00:06:53] Craig Asano: the first iteration of eCollection software, the platform that naturally has evolved more than once or twice, you're probably talking numerous of, of times to commit to production new, new code and updates. But, you know, you had said that EXUS is dealing with company... countries all around the world. During the... Since 2004, during that growth phase, I mean, what... You must have been one of the very first employees. How, how many, how large is EXUS today? Okay. EXUS is not a large company. EXUS is a company employing

[00:07:25] Chris Maranis: around 170 Okay worldwide. What we do and what we feel distinguishes is the hyper-focus we have decided to put on one skill. Actually, it was 20 years ago, we decided that to be the best in the world and do something, you need to specialize. So we, we take a hard look in what we were doing. EXUS was doing other stuff back then, other enterprise business applications, having also some major customers, including Vodafone and other enterprises. But that when we decided that we actually want to be the best in the world and do something, we had to select what that would be, and that should combine three things. One is something you are really good at, and we saw that we were already really good at collections. Our first system was really bringing results. Then something you are good at, something you love doing, and that would sound weird, and we... But all at EXUS, we all love collections. We are so into collections. We have really love what we are doing. And the third is something that makes economic sense or fuels your engine. And combining all that, we came up that we decided we want to be the best in the world in debt collection software.

[00:08:37] Chris Maranis: That's how we decided to actually stop other activities, even some profitable activities we had,

[00:08:44] Craig Asano: to focus on collections. And of course, since then and for these 20 years, there have been, you know, very big changes and many, know, turning points in history. Well, do you wanna touch on some of the maybe the business turning points as well as the technology turning points over... 20-plus years. A lot has changed, it sounds. 20 years, and actually,

[00:09:04] Chris Maranis: if thinking back of that, actually is a chronography of how the financial industry has changed these years. First of all, I remember 2006, more or less, we had our first implementation for a banking group, a group of companies implementing what we call a blueprint. A blueprint is, actually when you designed a, a master, let's say, implementation designed on the centralized level, and then you deploy in multiple entities of the group in different countries where you adjust, you configure it to adjust to the local regulation or cultural peculiarities. But in that sense you achieve through this blueprint approach economies of scale because they share a lot of stuff between them. And Also the transfer of know-how between different entities of the group, and you achieve what they call control autonomy. So that was our first installation for a banking group, setting the same collection system in seven or eight countries, Southeast Europe and North Africa. But I think what was the biggest change happened in the aftermath of the US subprime mortgage crisis and the Lehman bankruptcy back in 2008.

[00:10:15] Chris Maranis: You know, the crisis reached Europe after that, and then it also hit Greece when we were... back where we were located between 2011 and '13, and we experienced an unprecedented crisis. The GDP fell by 25%, unemployment rose to almost 30%. That was a natural stuff happening. Collection managers, the collection managers I was working with at that day, they almost overnight witnessed their NPL rate, the non-performing loans rates, going from something like 3% or 4% or 5%, which is, you know, an industry average, to 30% or 40%. Wow. Lending just froze, and in this environment, in this unprecedented conditions, together with some collection managers back then, we implemented the first collection self-service portal. This is what we practically say these last years as digital. This is the digital channels. It was back then in 2013, having to support this environment which we, we had never met before, we implementing, along with our customers, self-service portal to help customers in distress to find a solution out of the distress and the problem they were at.

[00:11:32] Chris Maranis: So that was a big change and a big experience for all of us this crisis. Then I think one of, one of another important milestone was a thousand... 2016. It was when the first report by an individual technology analyst, it was CEB Tower Group, North US company. They were then acquired by Gartner. They created the first report, independent report analyzing debt collection systems. Until then, and until recently, there was no technology report, there was no magic quadrant, no report from Gartner or anyone else about collection systems. It was the first in 2016, and we were included. It was a global report, so it was covering the whole world. And EXUS and another I think it was 14 companies in total, were included as the top system in this report. And more amazingly, we were already thrilled we were part of this report. We were still, you know, small and growing back then. But when the report actually came into the market, we were ranked as the top debt collections software vendor in the world with all the big guys there participating, which was for us, you know, a big surprise and I think a turning point for us, made, you know, a proof of concept that we can really working...

[00:12:53] Chris Maranis: we were really working on the path towards becoming the best debt collection software vendor in the world. So think that was another important milestone. Other than that, I can think our first attempts in machine learning in 2018, before AI became so popular as it is now due to the gen AI, mostly the LLMs. Machine learning is here for many years, so we got our first attempt in what we were calling self-adaptive strategies. That was 2018. 2018. Didn't work well. It didn't take off, to be honest, back then, and I think it was a combination of the technology not being very mature yet, and also more importantly, the unwillingness of the banking industry and the people to adapt to such a new, untested methodology. this is rapidly changing now, as we know. And I would, I would set as the last, let's say, important critical point around the pandemic because even before the pandemic hit EXUS was already working on a hybrid model, mostly remote.

[00:14:02] Chris Maranis: So we still had an office, but we were mostly working remote. We were traveling on our customer premises, and then we allowed our people to work from anywhere. But still the world was not let's say, accepting working remote in all situations. There was not a lot of trust that this would work. The pandemic helped helped us a lot to migrate to a fully remote model. People actually accepted that you can have a quality meeting remotely. That if you, of course, follow some ground rules, you can still have a productive meeting. So I think that was a turning point. We are completely remote company. We employ people from 16 countries around the world, think last time I checked, which that means it allows us to recruit the top talent from all over the world. We're not restricted in any geography, and we allow these people to work remotely, apart from the times where we actually travel for our, to our customer location for some specific sessions, which we still prefer happen on site. I think that's some major turning points in our history.

[00:15:12] Chris Maranis: Well, that's a lot. You touched upon some significant I would say

[00:15:17] Craig Asano: pivotal turning points just working with that early group of banks which helped you go global early on, and then being resilient to all these crises that we've experienced. And, you know, the comment about being truly a, a remote business definitely has advantages. But for, for listeners who don't know where the head office is

[00:15:39] Chris Maranis: and, and that... is, is it, is it in Greece or is it in the UK or...? It's in London. It's in London. Yes. Yeah,

[00:15:47] Craig Asano: that It's in London Okay. But you're, you're... We're, we're having this remote video interview, and you're based in Greece, though?

[00:15:55] Chris Maranis: I'm based in Greece. Yeah I recently moved to Greece. I used to live in London several years, but I've, have moved now back to Greece, which is my origin, where I was born and raised.

[00:16:07] Craig Asano: And the food and weather is better there, too, right?

[00:16:10] Chris Maranis: Food is great. Food is great. Okay. Well, no, that, that's covered quite,

[00:16:14] Craig Asano: quite a range, so, so thanks for, for that. So let's fast-forward then a little bit into sort of the modern version of, of EXUS and with, with AI and its impact on collections. You know, talk a little bit about, from your perspective and EXUS's perspective, what the technology is allowing you to do today that you couldn't really do, you know, a few years ago,

[00:16:39] Chris Maranis: even a couple of years ago. It's changing so rapidly. Indeed. answering that, I would by repeating something I said earlier, distinguishing the advancements we all witness now in AI, in AI in regards to gen AI, the LLM, which is something rather new. I mean, say it's new. It's already a couple of years now we're all using them, but still, you know, in the fintech environment it's something new. In contrast to machine learning and the predictions based on machine learning, which are here for several for many many years. So machine learning has existed for many years. Before that most of organizations were using behavioral scorecards which is, let's say, a different technology, working towards the same end which is predictions. Now with the advancement of machine learning, of course, you get more accurate predictions. The accuracy you get with predictions is amazing. You get most fast predictions and more adapting predictions. You know, when the pandemic hit a lot of organizations were already using some behavioral scorecards models that helped them predict the behavior of people in order to adjust their strategies accordingly. But these models, let's say this previous era, let's call them models they were designed and put to practice and they would be recalibrated every, I don't know, six months or a year.

[00:18:10] Chris Maranis: The moment the pandemic people started to behave differently because of the pandemic. This change of the behavior that people started demonstrating actually made these previous models completely obsolete. They did not, they did not have the capacity to predict behavior of people because the behavior of people had completely changed of... due to this, you know, life-changing event global event we all witnessed. In contrast now with our machine learning models, we have the ability to retrain them in... and have them adapt to new behaviors, if not instantly, in a much more automatic and quick way. So they are much more adaptive to the models we used to have previously. And I use this, you know, pandemic event as just to make the point and explain how that affects. Other than that, we have to say that the gen AI and the LLMs are a true disrupter. They have brought now capabilities That were not here five years ago, we were not even thinking about them. Of course, we all know now that you need to have guardrails, you need to have governance, you need to have regulations, and I'm sure we're gonna talk about them.

[00:19:25] Chris Maranis: But still, these agentic AI concept and the LLMs are truly making, you know, a difference in how collections and other businesses, I can only talk about collections,

[00:19:39] Craig Asano: are changing right now. I think it's fascinating that the, because of these global crises, that the behavioral it makes total sense, but if you're not actually thinking about that and how it impacts your, your business. And so when, when AI is helping lenders make those key decisions today in, in 2026, maybe, you know, you can talk a little bit about, you know, how... Like, is it... You know, we, we definitely wanna talk about how AI is helping lenders but, you know, clearly there's a discussion around the, how autonomous those decisions will be. And, you know, maybe you can talk about the industry, what you've seen a little bit in terms of how other companies are employing AI and, you know, how it's used by lenders today in 2026. Or, or just Axis. it's really, you know, up to you to provide some information on AI and its impact on lending.

[00:20:36] Chris Maranis: Okay. The impact of AI on lending, and of course, again, I'm gonna focus mostly on the collections part. Yes, course. Yeah. There's a large part on on boarding approving new loans, servicing the loans. But in the collections part, we have already seen a lot of changes. Mind you, I think the, the, the aspiration and the ambition of the usage of AI versus what is actually right now, today, put in production operation, the gap between the two is big. So there's a lot of discussion, there's a lot of new things being piloted. Still there is a way before we see that deployed in actual operations. And I was just reading just the other day about a survey. It was a survey on 1,000, almost 1,000 tech leaders or risk leaders and only 2.5% of them actually confirmed that they have a customer-facing agentic AI tool in production operation. 2.5 of them.

[00:21:44] Chris Maranis: The same percentage regarding, you know, running pilots around it. That will, will be huge, very big. A lot of things are happening, still there is a way to go to actually embed that in the routine and to the operations. Now, having said that there are multiple areas where AI now comes to help, and the first one is about helping lenders decide who needs attention and what kind of attention and what to do next. So to combine the payment behavior that the customer had before, the contact history you have with this customer, and also the current context in order to decide what to do. I'll give an example. We have a prediction model which we call lazy payer. What is that? There are some people who are perfectly willing to, able and willing to pay, they just pay late. eventually they pay. So if you can identify people in your portfolio, you would probably do a very soft outreach. You don't need to go hard on them.

[00:22:53] Chris Maranis: You don't need to spend resource on them. The guys will eventually pay, so probably a very kind reminder through, you know, an SMS or a WhatsApp, depending on their preferred channel would work. There's also some prediction models that help identify the people that we call them self-cure. So there are some people who will eventually self-cure, as the word says. They will pay the debt without needing any, you know, more resource allocated to them, without needing to even irritate because they will eventually pay. But there's also some other prediction models that can predict if you call someone, whether this person is going to give a promise to pay. In the collections world, this PTP, the promise to pay, is one of the important concepts, is actually calling someone, and this someone, first of all identifies as the person you want to talk with, and this is very important, you have to do the authentication, and then acknowledges the debt, an important step there also. And then stating that they're willing to pay, but just want to let's say given some grace period or just get some help and gives a promise to pay in a week or two.

[00:24:06] Chris Maranis: That's what we call a PTP, a promise to pay. Mm-hmm. And based on the models we have created right now, lenders, when they have a list, I don't know, 100,000 or 200 debtors to call, these prediction models could, with very high accuracy, identify the people who are more probably to give a promise to pay. And therefore, you would prioritize your calling list, calling these people first until going to the next. So this is some areas where prediction help. But let's not forget that prediction doesn't actually recover the money. This is halfway there. This is the prediction. Then you need to engage. And when engaging, then we have other AI tools, and this is also where the gen AI and the agentic AI comes into play with either taking control of the communication, so the AI directly talking with the debtor, or in other cases, assisting the human agent to make an efficient and thoughtful communication with the customer. So these are, let's say, very broad categories where AI has started to be applied with, of course,

[00:25:19] Craig Asano: other areas also come into play if we get into more detail. I mean, it seems that, you know, as these pilots mature and in theory become more successful, they're saving time, they're saving money they're becoming more accurate in their predictions of, of behaviors, it, it's quicker. You know, is there an emerging sort of line that you see with the lenders on, on, and debt collections, that side of the lender's business that, you know, you could see, well, we think AI or, or you think AI is gonna be beneficial for all of these tasks, all of these decisions, but we're gonna draw the line here. Or do you believe that, you know, in the very near future, you know, AI will be having that call with that potential customer trying to solicit that, you know, promise to pay? You know, how far will AI go or, you know, how far is it today in making those decisions? It sounds like it, it's, it's a... I, I'm not sure. I guess I need the answer. It, it's, it sounds, it sounds like there's a lot of activity but are there any fully autonomous AI debt collectors, or is that a, a, a space that you don't think it, it's, it's going to come to fruition?

[00:26:37] Craig Asano: Okay. First of all, I will start with a quote I heard from someone,

[00:26:41] Chris Maranis: and I like very much, that it's very difficult to predict, especially the future. So, especially when we talk about AI, and with the rate that things are changing I would not risk any predictions for, for, you know, more than two or three years. Mm-hmm. But I can, I can say what I see already and where I think or feel things are going to move. So one of your questions well, several questions here, but one of them and I think it's a very important one, is whether lenders draw a line between what autonomous AI could do and where a human intervention should take place. And I would, in a very high level of course, draw the line here where the impact of the decision is or becomes bigger or critical or more important for the debtor. So, talking with someone, just letting them know that they owe an amount, just kindly reminding them or negotiating some repayment plan, that's perfect. And I, to be honest, I expect that AI autonomous, as you say agents are going to take over this part.

[00:27:51] Chris Maranis: That's what we call mostly in early collections or soft collections where automation is expected, where you can have a call center instead of having people calling and needing to adjust your calling list or schedule to the time that people are actually working at the call center. You can take off these restrictions. You can have your AI agents calling at the preferred time of the customer through the preferred channel of the customer, and having these discussions. However, some point when a customer, and this is often the case, fell into distress when there is some major event, so a person must has lost their job they're facing some major health issue. There's any something happening in their life that has put them in a difficult situation. I don't expect, and I don't hope, and I hope humans are still going to intervene there and handle these situations showing the level of empathy that is required for these situations. also due to the fact that there's going to be, and I'm sure there going to be although they're not there already, regulations around Because if things go wrong, you need to know what who made the decision, where this decision tracks back to, and being able to explain where we found ourselves into this decision.

[00:29:18] Chris Maranis: So I, I, appreciate my answer is not straightforward,

[00:29:22] Craig Asano: it's not black and white, but this is how I think things are going to be evolving. I have an interesting question here on my research notes around, you know, we tend to assume that people, like humans, have are safer for, for judgment, like in the example that you're talking about an emotional crisis, somebody lost their job, they're more empathetic. But is there a situation that a computer, a, a bot, an AI is more consistent or they're, they could be fairer for the kinds of decisions in, in debt collection outcomes because a, a person, as you say, they could be tired, they're, they're working remotely around the globe on different files. Do, do you see that eventually that they could, they could be the preferred choice for debt collections even in the complex cases because they're more consistent and fair, and they have what I've learned this new word, this institutional memory because they don't forget. They can go back and read the entire file instantly. And as long as they're programmed to be, let's say, fair, and these are the outcomes,

[00:30:28] Chris Maranis: that it could be more effective than humans? Okay. Allow me, Craig, to make a distinction here between consistency and fairness. Okay. Because consistency is not debatable. A consistent consistency is much easier secured through the use of automated tools, including the AI agent. So if consistency is what we are discussing not using humans, who by the way, even humans hallucinate as AI does. real. I mean, they're, they're only human. When it comes to consistency automated tool and AI tool far better. However, that doesn't make it fair. If, if there is a process, strategy, a process designed right now, which is, let's say, unsuitable for the specific situation of the customer, it's not fair. AI will not make it more fair. Fairness is a different thing, and it's mostly applied by, first of all, the regulators that need to safeguard that, and by the policy makers, the strategy maker of the organization, of the bank, who designs a fair strategy for the

[00:31:43] Craig Asano: Yeah. That, that, that's fair to to use the word that the, the actual decision logic would be, you know, baked in and, and applied, whether it's, it's a human running or executing that file per those rules or, or an agentic computer or, or bot. So one thing you mentioned about, you know, when something goes wrong you know, who is at fault. Do you have a view on where you think the regulatory and the legals will eventually... You know, they're studying it on every file, particular you know, in financial services. It, it, it's, it's a very serious sector. So what, what are your thoughts on regulations where that's gonna

[00:32:25] Chris Maranis: that's a rather complicated issue you touched Craig. And it's a question that is yet to be answered by not only the lenders, the regulators, the vendors, the whole market. Generally speaking, again, that's generally speaking the lender is the one who always retains accountability for what happens. So just saying that, "It's not me, it's the bot that you know, made this wrongful decision to fast-track this person to a legal office," won't cut it. The lender will retain accountability. However, the technology providers will have their own responsibilities. Now, now that becomes complicated because different parts of the world are in different, let's say, levels of maturity, and eventually they're gonna deal with this differently. In Europe GDPR exists for many years now. And there's this Article 22, if I'm not mistaken, which already provides safeguards against decisions which are based solely on automated processing that eventually produce legal effects on a individual. That's already in place.

[00:33:37] Chris Maranis: And right now, EU is formulating its AI Act, which is going to be introduced in the coming period which is going to further add more controls around that. That's Europe On, on the other hand, US is having a different approach. In Canada, as far as I know, OSFI is, is... There's this guideline E23, I'm not mistaken, which says that financial institutions are accountable for the use of the models they use, and that includes also externally sourced models. I think this is under discussion, and it's going to go in effect in 2027. But, you see, there's, there's a lot discussion to be made. And I expect different, as it has been up to now, different parts of the world are going to handle that in a different way. Yeah, I think it's, it... I guess it's early days. The, in, in most countries, unless it's,

[00:34:36] Craig Asano: it's very harmful, it's, it's difficult to put those regulations in market today because they don't wanna restrict on, on the innovation side and, and fall behind That's also important think yeah, we'll have to come back and do this podcast in another three years with you and see what happened. That, that's what we'll have to do. I hope we'll do one sooner than that. Well, I, I think, I mean, it's, it's fascinating, AI's role, and it, it's really, you know... I wanna try something different here for this next segment 'cause we talked about the decisions that AI can help for on the debt collection side versus a human. And so I wanna play a little game, we'll say, that, you Sure I'm gonna have you respond whether, you know, this example is suitable from your perspective or EXUS's perspective to fully automate it, it's okay. Or let AI calculate and, and support a human's decision, but they're recommending. So, you know, the first option is fully autonomous. Second is, we'll make the recommendation. And, and the third is a human, this is best for a human. Got it. Got and, one of those three decisions based on, you know, all your 22 years of experience here.

[00:35:50] Craig Asano: Got you. So for a Ready payment a customer misses a payment. It's, it's a routine payment. They've got no previous problems. The, the debt collection system identifies, you know, here is a suitable repayment plan. Do you allow that scenario to be fully automated,

[00:36:07] Chris Maranis: or just the decision, or it has to be a human? Misses one payment, has a good history, no previous problem. We identify the problem, automate it. Certainly automate the offer. This is, can be structured automation based on the policy by the bank. Offer something to the customer. The customer chooses to accept or not.

[00:36:24] Craig Asano: And then we have a different, you know, have a fork on the road. We have different scenarios, but certainly automated that's based on the bank's policy. That, okay. So it has policies, give it to the robots. So how about a financial hardship example where the customer says, lost my job. I can't make these payments." Mm is that a situation where, again, well, you know the three outcomes. How would you handle it? Yeah. I would certainly pause the standard path,

[00:36:51] Chris Maranis: the standard negotiation path, the automated path when somebody claims, at least states they are in hardship. And I would bring bring a specialist in the picture. Somebody stating there is a problem doesn't necessarily make it a fact. certainly warrants to stop the automated process and bring someone, a specialist, give time and authority to that person to manage that. Of course, I expect AI would support the handover, would give all the information. They might later follow up,

[00:37:25] Craig Asano: but I would bring in a, a specialist case. Specialist in financial hardship How about what I've indicated here as sort of a, a detected vulnerability? So AI detects a, a pattern that they think this customer is vulnerable, but the customer hasn't disclosed they've lost their job or anything. Should the lender or EXUS, you know, act based on what they found and,

[00:37:49] Chris Maranis: and the customer, the borrower has not mentioned anything? One of the reasons we have put these prediction models in place is to provide this kind of inference, this kind of signals. So we should certainly act on that signal, certainly. If it's, again, it's a suspected hardship, I would route it to a human expert. After the signal is generated by the predictive model, I would route it to a, a human expert who would need to do the diagnosis, so we don't still know what it is, establish the facts, do the check-in with the customer, offer voluntarily to help. And depending on the result, we will follow up. if I may say here Craig, it's important when these things happen to close the loop, meaning to feedback to the process the outcome of what happened. Because the AI model needs to improve its prediction capabilities, and this is will only happens if you close the loop. And that's one of the cases where AI fails to be useful. When people take it on the first steps of actually producing something, and they're happy with that, so they're producing some signals, but not closing the loop, feeding back what actually happened to further train the model and improve. That's a very important step.

[00:39:08] Chris Maranis: Absolutely. And then the,

[00:39:09] Craig Asano: the model only gets stronger and, for, for your particular cases. So I have another one here, a debt dispute. So a customer says, "The amount's wrong. I don't, I don't owe that amount of debt." Should AI Is that simple enough for AI to handle or do you think you have to involve a human? Yeah. I would say banks do have specific processes for disputes,

[00:39:29] Chris Maranis: so this is a standard workflow. When when a debt is disputed, it needs to go through a process. Usually, this process involves a person. So a person needs to call up, verify, again, identity of the person talking, verify the balance, and discuss the grounds on which this dispute came, log all the history, and then feedback. So for disputes, again I would go off the automated and use an

[00:39:59] Craig Asano: expert on disputes to handle that. Seems to be a lot of combinations right now with the, the, the complex cases. But so the, the last case or example I have is, is what I'll call legal escalation. It's, you know, customer has just missed payment after payment, and next they wanna, you know, technically move them to what I'll just call legal recover me. So it's, it's a legal decision. Do you think the AI should be escalating those decisions, which has tremendous impact, adverse impact on, on the borrower? Well, as you rightfully say, this is a, a very impactful decision, and I think it's going to be part also of regulatory,

[00:40:38] Chris Maranis: let's say, attention. It is in most of the cases already. So before moving someone to litigation I think again, a human expert needs to intervene. Although all the evidence are gathered by the AI, let's say, agent verifying that, going through affordability reviewing the situation, and taking the final decision, and being accountable for this decision should be a human being because that gives me a good opportunity to, to add something here. Introducing AI agents does not remove accountability from the organization. In contrast and that's something which I heard I, I was, I was in a in a FinTech event in Frankfurt last week about AI. And there was something said, so that's not me, I'm just quoting someone, but I thought it was spot on. So we all talk about human in the loop. So we have someone in the loop, as I'm talking right now.

[00:41:45] Chris Maranis: There's a person to make the ultimate decision. But you need to have someone on the loop, meaning someone monitoring the whole loop, the whole journey, making sure that it works as it should work, being accountable for the decision at the end of the day. So if things go wrong, there's a human accountable, not a bot accountable. There's a human accountable, and also has a kill switch, being able to pause that. So further to the human in the loop, the human on the loop, and especially for this, what we call special workflows, which for bankruptcy, for litigation, for insolvency you need to have this let's say, design in your organization. Yeah, absolutely. The, you know, AI can really perform incredibly,

[00:42:37] Craig Asano: but in terms of governance and accountability the, the humans surely should take precedent and, and priority. No, that's fantastic. So I, I think these cases provide a lot of insight into the decisioning and how AI interacts with humans or what they could potentially automate in the future versus all the way to, to the governance and the legal escalation and the more complex cases. So thanks for sharing those. So moving on here. I wanna You know, when I was researching this conversation, I came across something online that you had spoken to before. I thought it was interesting. You talked about the business case for caring in collections. So what, you know, what do you mean, and, and does Do lenders think like that? Is this, is something that, you know, only EXUS thinks about or, or you? Or what is the business case for caring?

[00:43:30] Chris Maranis: Yeah. Good point. Actually this was what you read, this article was sparked by a personal experience of mine some, I don't know, seven years ago, I think that experience brought forth for me something that was always there but it was made clear. To, to cut a long story short, I was in a management meeting with my management team, my peers, and I was challenged by my coach to set a breakthrough goal. Something that I love to do, but it was very, very difficult to actually do. So I did commit to play guitar electric guitar with my 16-year-old son in, in front of the whole company in the Christmas event, which you might say is not a big deal, but it was for me because I had never touched a guitar in my life. I don't know how to play guitar. And it was what, was June when I said that, and, you know, Christmas was just, you know, a couple of months down the road. Anyhow, the, the interesting thing was that the, my team did disregard or mock me for that. In contrast they supported me. I got a lot of support from my team. They kept me accountable, and I did it. I actually played Sweet Dreams by Marilyn Manson, one of my favorite Okay with my son in the Christmas event.

[00:44:42] Chris Maranis: what that uncovered for me was that being in a great workplace it's means helping people become more than they think they can be. And you know, there's a myth that caring for people is something soft. While in business, what really counts is, you know, KPIs, performance indicators, pressure. What, what I truly believe and strongly believe is that caring for people is not separate from performance. It's actually one of the drivers of performance, and that's what we do at EXUS. We really genuinely care for our people. Caring for people doesn't mean going easy or soft on them. That's not happening. We ask a lot from our people, but we really deeply care for them, and we have made our modus operandi. We call it our flywheel. So we take extra good care of our people, and our people, in return, they do the same with our customers, which are the collection teams, the people working at the collection departments. They take extra good care of our customers. Mind you that our customers, further to being, you know, large enterprises, at the end of the day, they are people, people that have to make decisions under pressures pressure that need to you know, achieve their goals, but at the same time they have to show the same care for their customers who are the debtors.

[00:46:03] Chris Maranis: And this is, this is a flywheel that actually works. We take extra good care of our people. Our people take extra good care of our customers, the collections teams. And these people are in, are capable to provide this extra good service to the debtors, and that actually works. And when I mean it actually works, this is not just, you know, fluffy stuff. EXUS has been consistently growing in an double digits, I may say, for almost a decade now.

[00:46:33] Craig Asano: So this is actually working in business terms, not just having a good time at work. It, it's the it's known as the golden rule. I, I think it's, it's treating people how you want to be treated and, you know, what you're touching upon, you know, the culture, it's, it's, it's a great leadership lesson for small and, and big companies as well. And it's good to hear that. We don't hear that enough, and I, I'm, I'm glad that you, you that article fleshed that out and we, we saw that. I wasn't exactly certain what that was, so I wanted to get your opinion, and I absolutely agree, so that's fantastic. So one of the reasons that, you know, brought you to NCFA's Fintech Fridays podcast here is that you've... EXUS has an expanding presence in Canada,

[00:47:18] Chris Maranis: and it's now Working with a major Canadian bank on its collections side with the technology. Can you tell us a little bit about, you know, your, your, your footprint here in, Okay Canada and, and the relationship and, you know, your views on the Canadian market? So indeed, we have recently expanded our presence in Canada, working with one of the major in the country. And Canada for us is, I would call it a long-term commitment. We strongly feel that we can add value by bringing the expertise we have from another 40 countries in the world we have worked with these two last decades, but also this unique service level approach we have to the local market. We have already established a trusted relationship with our first customer, and plan to, of course, build on it. And as we have done in all other markets, we have, let's say, increased our footprint. We will allow results and word of mouth do the rest because this is actually how EXUS has grown, by word of mouth, to be honest.

[00:48:26] Chris Maranis: You know, the collections industry, the credit industry, let's say, it's not, it's not that big as you can expect. I have connected with collection managers in all these countries, and we have worked with, and I've seen that the market's not that big. And people talk, and people move from organization to organization. And the best marketing we get is actually from successful implementations, and that's why we also plan to expand in the Canadian market, by being successful, by making, making our first installation in the country being a center of excellence, something that other organizations would, let's say, be envy of and would like to also put forth for their own organization. Now, I, I'm assuming here you,

[00:49:13] Craig Asano: you can't say too much about the implementation but maybe, you know, from a any large bank's perspective, when they go through modernizing collections Mm-hmm sort of integrating that technology, what... You know, can you talk to us little bit about those experiences? What, what is the most challenging hurdles in those implementations that any large bank or even a mid-sized bank has Mm deal with, at least when integrating collections platform technology? Greg, if you allow me here, I will, I will draw from the experience I have, not from Canada, but from multiple tens of installations around the world because some patterns are clear.

[00:49:57] Chris Maranis: And it doesn't matter if it's a auto finance company a universal bank, a consumer finance, a multi finance, a leasing. We have all these kind of lenders, and it may be in Indonesia, in the Philippines, in Saudi Arabia, in the UK, Netherlands or Canada. There are some things which I always see are the key challenges in implementing a modern debt collection solution. I would say the first one is, the hardest part, is to getting the data right. Getting on time and on quality the data. Everybody's saying that. I'm not new to that. Garbage in, garbage out. If you don't have the data stored and be able to gather, consolidate, and feed the collection system, then the quality of the output you will get would be low. I've seen organizations, huge organizations, very advanced and well-respected not having the quality of data they needed even to uniquely identify customers with multiple products in their organization. You know, one of the most important things you can do in collections is to work on customer level. So for example, Greg, if you have a mortgage your bank and you have a credit card, and you are paying your mortgage regularly every month, and you just miss one payment, just a small, you know, $50, $100 payment on your credit card, you wouldn't expect them to call you first thing on Monday morning and tell you, "Why are you not paying your credit card when you have been so regularly paying on your mortgage?"

[00:51:34] Chris Maranis: You would expect they would, you know, see the big picture and you know, treat you as the customer you are. Sometimes this is hard for some organizations because they fail to consolidate information that comes from legacy systems, back office systems that haven't been let's say, consolidated. So that's one thing that affects a lot of projects, the quality of the data in the organization. The second one is the coordination between the IT and the business. So the business teams usually have a clear view on what they need. They have the urgency, they have the requirements, they have the vision. On the same time, the IT department has a portfolio of projects, and in those there's tens, if not hundreds of projects that need to run on the same time. And these two, let's say, teams, organizations sometimes fail to sync or find their common ground to make, you know, to move let's say in unison. And we have found that one of the services we have offered, which is not in the core services but this is something required, is to be the glue, let's say, between the IT and the business, to be the API, say, that translates, you know, business world with IT world, and help them, you know, proceed with a, a common milestone, so with the project schedules we have.

[00:53:01] Chris Maranis: That's the second thing. And the third one is when an organization is trying on the same time to change their infrastructure, their technology, when on the same time changing their strategies, their policies. So if you, if you're changing both at the same time, both infrastructure and the strategy, then there is a big challenge on actually testing if what you have generated is according to what your expectations are. You don't have to check against something. So that's why a lot of organizations sometimes do the let's say the lift and shift type, changing their technology without changing the processes, or changing their processes if the technology is sound. Or if the technology is not sound, first doing the technology upgrade, and then building on that technology, doing all the changes that this new technology enables. So I would say these are the common

[00:53:58] Craig Asano: challenges I've seen around the in different lenders. They certainly sound like challenges a large Canadian bank might just have. Why should they be different than the rest of the world? Yeah. Well, you know, you often hear legacy data. We've all experienced if working for a large company the business versus the IT teams, you know, trying to control and, and just having that alignment and the political will from the bottom and the top to, to do it right and, and properly. But it's exciting times for sure. I was just having a look at the time here. We're gonna move to sort of the future of collections discussion a little bit as we move to, to wrapping up here. You know, we talked a little bit about agentic AI. And from your perspective, where do you see in the few years all the innovation happening? And, you know, how And then after that, we wanna talk a little bit about the actual innovation AI or otherwise. You know, what, what's the future for EXUS? But first, let's talk about the agentic AI and, and its impact on collections in Where, where do you see it going from So the,

[00:55:10] Chris Maranis: we already discussed the part of let's say prediction and decisioning, which is already advanced. And I think more and more being more and more accurate a lot of the decisioning is going to shift towards prediction machine learning models. But I think the biggest disruption right now that's happening on the gen AI part, which is what most people relate as replacing the human agent with a bot. Again, it's difficult to predict the future. I don't think we're gonna get to what's called the complete agentless environment. Of course, the role of the agent is going to substantially change. They're going to be less agents, and the work the job description is going to be completely different. It's going to be more challenging, which is good, it's great. But more and more contact is going to happen automatically and through agentic AI as they can select the best time, the best channel, and handle more and more discussions. This, let's not forget, it also reduces the operational costs of the organization, which one of the major thing and the after.

[00:56:20] Chris Maranis: And also the consistency you talked about earlier comes into play here, and gives another point on automating contacts. What remains to be solved, and for the time being, I think it's going to halt a bit, this migration, is the requirement for accountability and for explainability. It's not only regulations. It's, it's regulations, it's ethical, it's business. For many reasons, you need to be able to explain why something happened. Why was this customer contacted or not contacted on that day? Why was this message specifically on that day sent to this customer? So this is something the whole industry is working on, and I think with the regulations that are going to come in play the next years, we're gonna see more and more of this business being automated and getting over to the AI Yeah, absolutely. Well, as you say, it's tough to predict, but we're living through some fantastical times, I, I will say, on, on the innovation front. So from XS's, you know innovation vision,

[00:57:29] Craig Asano: you know, can you share a few projects that you might be excited about? What excites you? What's coming down the pipeline? Okay. So there are, there are tools that have already been developed, have been piloted. Some are already in operation,

[00:57:43] Chris Maranis: some under pilot operation. That includes, let's say, the AI negotiator. This is a tool that actually handles communication with the customer directly, with the debtor. And we also have the AI coach, which is something like, a co-pilot for agent that helps the agents in their communication, but it also helps the supervisors to manage their teams. If I give a quick example when you are a supervisor in a call center, one of the tasks you have to do is to actually make sure, do, you know, the, your QA, your quality assurance. Make sure you that all policies are being followed, and that includes also eavesdropping on some of the communications that your agent have with people to make sure that all regulations are followed, that the bank's ethical and strategic guidelines are followed. Imagine having 200 people in your team, and they make something like 200 calls per day. Can you actually eavesdrop on all these calls? No way. You can just, you know, sample one or two.

[00:58:51] Chris Maranis: With this AI coach we have developed, all communications are audited, and the supervisor gets summary of the insights and the results of all these communications. So they can very quickly do their performance manager on their teams, they know which ones are performing well, and also tune their messaging in all the discussions they're having. So this is something which already happening, and I think it's a big, big change for collections operations. And if I had to choose one of the things that, you know, are in the pipeline to come in the future, I would say something that our innovation lab is currently still, you know, fleshing out. It's the regulatory compliance agent. And, you know, we, we, we operate under a very regulated environment, and these regulations change constantly. And collection operations need to remain compliant at any times. So this is a big task, being aware of all the regulations and making sure that are, at any time, they are followed. So this regulatory compliance agent is a tool who is going to monitor all changes in regulations and make sure that all internal processes abide by these regulations for the bank.

[01:00:06] Chris Maranis: This is still in design. It's something, you know, it's still in a let's say prototyping phase. And hopefully we will have seen it, you know, come to market later next year. Yeah, it's super interesting because,

[01:00:18] Craig Asano: it's, as you said, it's, it's a, it's a massive task, but it cuts across and into or maybe through all the other agent processes. So in a multi-agent platform and, stack. But it, its impact is, is critical to, to all, any operations and, you know, it's gonna have impact on costs and just making sure that these transactions are, are legit and, and legal. So fantastic. I can't believe that the time has come on here. We're, we're going to go to our rapid fire questions. These are, you know, just expecting short you know, short answers just to have some fun.

[01:00:58] Chris Maranis: Rapid fire. Sometimes it's a bit of business, sometimes it's a bit of personal. So we're just Let's go short answers. So the first rapid fire question for you, Chris, is the biggest misconception about collections. That's easy. The biggest misconception is that collection starts

[01:01:12] Craig Asano: after the relationship with the customer ends. Totally after the end. I see. Good answer. Next rapid fire. What's one thing that the collections

[01:01:21] Chris Maranis: industry needs to stop doing? Thinking of applying or thinking of applying AI

[01:01:26] Craig Asano: to do more of something that does not work. Yeah. You can't automate something that's, that's broken. It's broken for a good reason, right? Next question. One lesson that you'd give yourself when you joined XS way back in 2004, 22 years ago.

[01:01:42] Chris Maranis: Oh, to my much younger self. To your much younger self. Yeah, that's, that's I think,

[01:01:47] Craig Asano: I think a good advice is talk less, listen more. Well, you've done a good, good job today. Next rapid fire, what still excites you most about the industry after 22 years?

[01:01:59] Chris Maranis: I think the best moments I experience when we get, me and my team get the thank you notes from our end users mentioning that how we

[01:02:08] Craig Asano: somehow, even through technology, positively affected their lives. That is the business case for caring in full circle right there. So last rapid fire question, what's on your personal bucket list that you haven't done yet, but you will do it? That's personal, so not business. Personal It's a personal bucket list.

[01:02:26] Chris Maranis: If you wanna That's easy. Yeah. That's easy. Yeah. It's been for many years, probably gonna stay for many years. That's to read James Joyce Ulysses. Oh. You, you could just pick that book up, no time? I've tried in my 30s, I've tried in my 40s, now I've just entered my 50s. I'm gonna try again, but this is a bucket

[01:02:44] Craig Asano: Well, you get that in, in your retirement plan the, the golden years to come. Well I just, you know, wanna thank you, Chris for sharing your, your knowledge. It, I, I've actually learned a ton. One thing, you know, as we wrap up this podcast how can people who get in touch with you or, you know, if they have more, they wanna learn more about EXUS or potentially,

[01:03:10] Chris Maranis: you know, discuss the platform or how do they contact you? I would say that's rather easy. You can find EXUS online, E-X-U-S. It's rather simple. And of course, I'm on LinkedIn. And another way to find about EXUS, just search and find about making collection heroes. That's our trademark. So making collections heroes is what we want to do. Search for that and you'll find

[01:03:34] Craig Asano: Sounds like an amazing company to, to work for. Thank you. Chris, at, at this point, we're just gonna, to wrap up, and is there any last message that you wanna say to our listeners here before we wrap? Well, first of all, I would like to thank you, Greg. It's,

[01:03:51] Chris Maranis: I look at the clock now and I see it's one full, one hour we have been talking. That was a very interesting discussion. Thank you for prompting these questions. And what I would like to say is that it's a chance for me to say that collections on a, is much more complicated than what people think, and it's much more impactful than what people would think. And I would like to just say kudos to all the people that are working in collections all these years and have been serving the debtors, helping them, while at the same time getting, you know, some negative spotlight

[01:04:28] Craig Asano: on their job. I just want to say I really appreciate them, and feeling great working with them. It's a very positive, you know, spin or, or perspective, Chris. It's, it's fantastic. So wanna wish you and, you know, the entire 170 plus team at EXUS, you know, all the success in, in the future here in Canada as well as you expand and, and grow your, your footprint. Thank you very much, Greg. you know, welcome for that, and well, thank you for sharing your insights today. And with that, we'll just wrap up here. So to everyone and our listeners, if you're new to Fintech Fridays, please check out some of the incredible past episodes on the site. I think you'll be surprised with what you find. We look forward to seeing you next Friday for another episode of Fintech Fridays. Have a great weekend, everybody. Take care.

Outro : You've been listening to Fintech Fridays brought to you by NCFA and partners. Tune in weekly for the latest fintech Friday podcast by subscribing to this channel. The National crowdfunding and Fintech Association of Canada is a non-profit actively engaged with social and investment fintech sectors around the globe and provide education research industry stewardship services and networking opportunities to thousands of members and subscribers. For more information please visit ncfacanada.org.

End of Podcast

Subscribe and Listen to more Fintech Fridays podcasts here

Join NCFA's weekly Podcast series 'FINTECH FRIDAYS' where we sit down with the incredible people in the Fintech community and talk about leading fintech products innovations developments and challenges!

Interested in getting involved as a partner or participant? info@ncfacanada.org

The National Crowdfunding & Fintech Association (NCFA Canada) is a financial innovation ecosystem that provides education, market intelligence, industry stewardship, networking and funding opportunities and services to thousands of community members and works closely with industry, government, partners and affiliates to create a vibrant and innovative fintech and funding industry in Canada. Decentralized and distributed, NCFA is engaged with global stakeholders and helps incubate projects and investment in fintech, alternative finance, crowdfunding, peer-to-peer finance, payments, digital assets and tokens, blockchain, cryptocurrency, regtech, and insurtech sectors. Join Canada's Fintech & Funding Community today FREE! Or become a contributing member and get perks. For more information, please visit: www.ncfacanada.org

Buying a Canadian MSB and Reviewing FINTRAC RPAA and Financial Due Diligence

Why MSCI Is Removing Strategy From Its Indexes

How Is Crypto Market Cap Calculated? What a Change in the Top Coins Really Tells You

What Happens When AI Agents Start Trading Together?

Canada’s VC Totals Are Reliant On Mega Deals

Why Canada Is Making It Cheaper to Leave Your Bank

CSA Tokenization Report Shows Where Canada May Scale First

Creative Ways to Boost Team Morale

Webull Adds Free Data as Canadian Brokers Compete Beyond Fees

BBVA Deploys Canadian AI for Derivatives Pricing