The End of Account-Centric Banking: Jouk Pleiter on Backbase's AI-Native Banking OS

Host: Foreign. Kills intelligence yoke. You have spent over 20 years helping banks in their transformation towards digital mobile. The experience economy with layers that sit on top of their course. But now you're telling them that adding AI to their fragmented stack, or like I like to say, spraying the AI some outma index in different departments, that's not going to help them towards becoming an AI native business. Why?

Jouk Pleiter: Well, I think the. To your point, if I. Thanks for having me. I think the big backbase has been in business for already 20 years and we always have had a, let's say an omnichannel vision, like how do we unify channels? But it was just channel facing. I think the problem we've already been fighting for the last, uh, 20 years is fragmentation. Channel fragmentation. And I think to a degree you can solve that. But now we see with AI that basically the operating model of the bank is basically really the underlying problem. And what do I mean with that? Uh, you will have basically different. The bank is organized around products or around M channels or around operations. So basically people are operating in their own budgets and in their own, let's say, organizational logic. They're investing in systems over there, but nobody is really connecting the dots. So that's a larger vision we see is how do we move to a unified frontline? So we go from a fragmented frontline. And the problem in a fragmented frontline is there is no shared truth. Everybody's acting on their own local truth, some sort of. Right. And there's also no shared outcome. So the meta orchestration, something is happening in the channel with the customer. They have an intent. Ideally we do that straight in the mobile app and there's immediately fulfillment. But in a lot of edge cases, you have to go into the bank, you have to go to the branch, you have to go into the call center and then the drama starts because there we see thousands of people that are basically on swivel chairs between different systems and they're basically. So in a way you could argue banks don't need more channels, banks don't need more systems because you basically increase the fragmentation problem. We need to now unify these systems and make sure they can work as one. So this is already true without AI, Right? You could already argue without AI. This is a very important direction of travel. Now imagine you're going to add the power of AI to this fragmentation. I think agents don't have the right context. If you cannot give them the total context, how can they reason? How can they make a decision? Uh, if you do not have like More central, let's say governance and like how do you rule them and how do you give them the guardrails? So you can actually every action, every step is explainable to the end customer, to the employee, but also to the regulator and the risk and compliance team. Yeah. So I think really where maybe the last 10 years the battle was about user experience and having a beautiful mobile app like Revolut. I think the next 10 years is all about the operating model and how to make sure that humans and agents can really collaborate together, not in one isolated app, but across the frontline.

Host: Yeah. In a nutshell we've done very well with the experience economy, the customer experience, that's what really changed in the last, let's say two decades. But problems like servicing a client, uh, that may touch like 10 different areas of the bank behind the scenes. Right. Disputes might touch, I don't know, five areas or onboarding. How do we solve that? And how do we move from just data feedback loops that were created in the digital and mobile era, but not the contextual intelligence that we need. So let me dive in and introduce you first because we hit the conversation immediately. It's a very exciting times. You are the founder and CEO of uh, Backbase. And this is a company that's 22 years old serving over 120 financial services institutions, be it retail banks, commercial banks, wealth managers across 50 countries. And recently you announced this new operating system, an AI native banking operating system. You were awarded just a week later, sort of you were characterized or awarded the leader position from the Forrester Wave report on digital banking engagement platforms. And specifically one layer of your operating system, Nexus was called out as uh, a ah, very competitive and significant part of what you're offering. Now I hope we'll have the opportunity to talk about these layers, but before we do, let's discuss what your vision is. Yaok of an AI native bank. Not the architecture, the tech stack. Really? Yeah. What does it mean? You know, 20, 33, five years down the road from both the enterprise perspective, the customer perspective, just the ecosystem perspective, how does this look like? Where are we aiming? What is the North Star?

Jouk Pleiter: Yeah, yeah. So maybe we can then go back to. Okay, if we talk about five years or ten years from now, uh, what is changing? What is the game changer? I think basically the biggest game changer is that our agents are joining the workforce. Okay, so that, that means that basically we now need to think about thousands of agents that will be operating inside the bank. And these agents will help customers that can act as Their guides, their coach, complex discussions about do I refinance my mortgage? Like all of those kind of very advanced advisory type of conversations. There's a lot of that stuff going to happen towards the client. You're going to have digital twins of the client that basically can negotiate the best rates. I think banks are at the risk that maybe you don't need the mobile interface anymore because people will start banking in GPT and basically you're not owning the experience layer. So I would say there's a lot of at stake in the next five to 10 years, one how to keep ownership of the user experience layer. So that basically means it has to be conversational, it has to have that new form factor. So I think historically form factor online, then we've got the mobile form factor, we're moving our fingers on a smaller screen and now we have text and voice to interact with these apps. So I would say that's really the continuation of the battle for the customer experience. But agents are driving that. You can also now envision, okay, if these agents are going to have all that freedom, how do we govern that? But we're going to, we're going to touch on that later. Sure. On the other side, I think it's going to have a massive impact on the internal workings of the bank, basically operations. So I think what you will see there, instead of hiring additional new employees in underwriting, in kyc, in ongoing customer diligence, basically both in the front office, mid office and back office, all these roles, you basically hire people and then you train them on SOP, standard operating procedures, guidelines, etc. I think we're going to see a massive explosion of how do we train agents in these particular roles and within these skills and how do we make sure these agents over time become really good. Right. So I think that is roughly what is the game changer. I think we, we now have the power and the technology to automate everything that was non deterministic. With non deterministic is basically what you put in workflows and what you put in in business code and large software code. We're going to go now cross that frontier at scale. So the problem is not the models. The models are more than powerful enough. So I think that is already more than proven and if you can just extrapolate, that's totally fine. I think it's now more about um, the operating model, the guardrails, the governance, creating a shared outcome model, creating a shared understanding of what is happening in the front line. I think that is really the big challenge. So 10 years from now, I think one, we have to change the operating model from a fragmentation model with different product groups different. We have to do probably the most difficult thing in banking because the move to a unified frontline I think is basically also fighting the org structure of the bank. That makes sense. People can spend and execute within their org structure. So a group can optimize the online channel, but holistically they cannot control end to end. So I think we need to have a very strong alliance between the chief digital officer, the chief operating officer and the head of the business units to start to think more holistic. Most of the customer intent is in the mobile app, right? Customers are, they want to do something. Pretty much every intent is basically online. But some of these intents we can close within a second and you wire the money and it's done. So it's great. But then all these other edge cases, I think they are the ones that are very problematic and I think that is the ch. So it's not only where do we gonna BE I think 10 years from now, uh, it is completely agentic. Agents are doing most of the work and agents have to be orchestrated in end to end workflow. So starting with the customer intents, executing it throughout the bank and then basically doing the fulfillment. And ideally most of that is happening in seconds or milliseconds.

Host: And this is quite complex, especially given platform business models where you have ecosystem collaborations and partnerships and maybe we will see new products. But definitely what I'm hearing from you, we are talking about the transformation at the organizational chart level that yeah, maybe,

Jouk Pleiter: maybe this is the most difficult thing. And that's also I've been wondering, like, why do banks have so many fragmentation over the past 50 years? And I think to a very large degree it's the Conley Conway's law. You get an your operational output, your software architecture pretty much is dictated by the org chart. That's the thing in software companies, that's the thing you see now in banks as well. So Yeah, I think 10 years from now everything will be agentic. Every task, every role will be executed by agents. You need to govern that, you need to control that, you need to make it safe. Secondly, these agents are more powerful not only in context of a little siloed application, but especially if they can see the whole picture. So what is the total picture? End to end? That's the unified frontline, that is the operating model change. If you believe that ultimately this goes into an operating model change, then the change management question is how do you mobilize different power units in the org structure to collaborate and start to think more end to end. So I think you don't want more systems. There's enough systems that's not the reflects to add more systems I think really should stop because you're basically increasing the problem. I think now it's the time for a meta overlay layer that is starting to basically acting as a control plane that can go into 500 downstream systems to get the right data and to execute, but also that can start upstream, coordinate work to customers, employees and agents. So that's also I think where banking OS comes in. It's that control plane for governed execution. How do we make sure that humans, customers, employees and agents can collaborate end to end and then how do we make sure that we can orchestrate the work and the data, uh, below to 500 underlying systems over time? I think you can also simplify and clean up and simplify your IT landscape and basically reduce that to a smaller number of anchor platforms.

Host: Yeah, yeah, but there is already a core transformation that you're a new operating system hides if you want or is worth highlighting in my opinion, which is the following forever and ever. Uh, banks have. And when I say banks, I mean in general financial services, right. Have been operating in an account centric way. There's the ledger with the accounts and so on. Now the architecture that you're suggesting that is behind the this layer is really a customer centric truth layer, meaning that this layer that, that you. One of the layers that you have created has the customer with everything, the account, the products, the policies that are relevant to this customer hanging from it versus the other way around. This is really a deep transformative way of looking at, at the business and extracting content and intelligence from this. Can you talk to us a little bit about uh, this from m perspective of how did you come up with this concept and then build the engineering stack? Why is it important to create this customer centric truth layer? And do you believe that this is really core to an AI native transformation, at least for existing businesses? We're not talking about new ones that are uh, built maybe from scratch.

Jouk Pleiter: I think they are. It is essential for old and new. And the reason it started for us two years ago to basically think how do we reinvent backbase just from a digital banking platform into an AI native platform? That was the north style. Like how do we reinvent ourselves? If you do first principle thinking, you basically start to realize, okay, ultimately agents will initially start with human in the loop, but there must Be a future where agents will become a little bit more autonomous, right? And okay, if you have agents, how do you. Agents can only be as good as the input you can get, right? So we can train them on SOPs and knowledge bases etc and we can give them skills. But also agents need to be able to reason about, okay, what is actually going on around the customer. What we then start to realize, like the first principle thinking, the core actor in any banking operating is the customer. Ultimately it starts with the customer and it ends with the customer. So you can almost argue customer intent and then resolution loop, right? Call it intent loop, call it resolution loop, but that's it, right? And I think with mobile banking we already have done a great job. For instance, you can instantly, you see your account balance, right? You can instantly wire money, you can block your card. So I think there's a subset of capabilities where the intent loop is pretty much perfect, it's done. But there's, then there's the long tail where it is still a drama. And that's basically the frontier that has never been automated. If you go there, the agents, if they want to operate the data and the truth is scattered in let's say 50, 200 different backend systems. So they all have a subset of the truth, they all have different pieces, right? So that's the first problem you. So we basically said we have to do a paradigm shift. Every architect in the last 20 years explained us like you do not duplicate code or data, right? It has to stay in the system of record. So most digital banking platforms, if not all basically had a pass through architecture. You get the data, uh, you combine it from multiple five sources. But like you do not persist it or you do minimum persistency because it was an anti pattern, you just print it on the screen and you bring it back to the system of record. The whole industry for the last 20, 30 years basically being educated in that paradigm, do not duplicate data. It's a sin. We basically like, hey, we need to break that if we want to unify the frontline. And uh, then we said, okay, we got inspiration from a company called Palantir where they say they are not the system of record, they are not the data lake. So also backbase is not the system of record, we are not the data lake. You have plenty of data stuff, right? The thing is that how do we bring it together into what we call an ontology? That is a common business language. So in five different backend systems, they all call the customer entity little different. So how does the agent figure out like is this customer? It's called client. What is this? Right?

Host: Yes.

Jouk Pleiter: So within the ontology that's part of the truth layer, you basically help the bank to come up. Let's basically build in business language what is a person. That person can be a customer. In retail context that person can be an employee. So you start to basically build up a common language and about what are the different actors in the front line from customers to employees to products, the relationships. And you can start to model that once you have that ontology, you can then start to bind data from 50 different legacy systems without replacing them, keep them in place, keep your data lake with snowflake data bricks, it's all fine but connected to the ontology in a customer state graph. And that is highly efficient because now any for the client mobile app, for the employee workspace, for the agents, you can just query that customer state graph and uh, it has instantly at the right. And if the customer state graph is not complete because you're adding new products or you're modernizing another part of the bank, you're basically building that ontology and you're extending it and you're building your state graph. So that's one. You basically have a single shared of truth. There's one more thing which is like gold and we've been wasting it in the industry. That is in every journey you have decisions like an onboarding journey, you have decisions about kyc, you know, address verification, like dozens. We typically uh, at the end of the day when the customer is approved, we just onboard them in the core banking system. The only thing we continue to know that there's this customer and this is the address and this is the account. But that all the previous historical decisioning data, uh, in that whole journey, we don't store it, it's lost. So we have no historic recollection about like why did we onboard it, what was the credit score, etc. Imagine you could have a ledger that you can basically start to track all your decisions. And you start.

Host: So it's a dynamic, it's a dynamic kind of time series. It's got all the events in there.

Jouk Pleiter: Exactly. That's it. And that is gold because uh, backbase in a digital platform sits in the in on the crossroads of pretty much every decision you make around your customer, from onboarding to Re KYC to fraud detection, to making an offer for xyz, all of that. And also exactly why did we make the decision at the time? Why did we make that offer? That whole historical trail is super relevant to make future decisions. So in, in summary, this new truth layer or customer through layer, we basically said the core artifact is basically the customer and everything. How do we make sure that we have a common business language? So almost like the Tower of Babel where everybody's speaking different data models, et cetera, let's unify the model in a single ontology. Let's then use that ontology which you can tailor per bank to basically build the state graph so that whenever we need to query something about the customer, we have just one location. We can go there, super efficient and we keep that always in sync. And then uh, thirdly, very importantly with the context graph, we're basically building this historical trail of all the decisions and the decision timestamps. So we basically are building a very concrete picture of the customer in the moment customer state breath, but also the historical decisioning trail or evidence trail for that particular customer.

Host: So tell us a little bit about the actual layers. I know you've presented three layers on this operating system without obviously, yeah, uh,

Jouk Pleiter: we'll keep it relatively so maybe it's good for people to remember, uh, in our heritage, the banking OS is basically the new control plane of what we call the unified frontline. So banking OS is basically the platform that will help the bank to run your digital channels. But it doesn't stop there. It can also you can repurpose it for your front office, for the relationship manager, for the branch. So basically it's very logic. If you can do it online or mobile, why not power your call center employees with that? Why not give it to your relationship manager so they can see the same thing, the customer can see the same data, they can act on behalf of entitlements. Okay, so we went from digital channels into front office and then in the front office you see a lot of edge cases and exceptions where people have to go to operations to get together an expert opinion. So our vision is basically really with one banking operating system to streamline the front to back workflow from the digital channel to the front office to operations and back and forward. We basically want to do that because we want to close the intent resolution loop. Does that make sense? Something starts with the customer and then how use case by use case, how do you basically utilize the platform? If you then look at the platform, what does it need to do? The layers you ask is first it needs to have a connectivity layer. We need to be able to integrate to any legacy system where remember we're not a system of record. There are dozens of systems of record. Let's make sure that Nexus, the customer state graph, the state, the truth layer, it needs to be connected. So the connectivity layer is bread and butter to legacy systems, to CRMs, to FinTechs, anything.

Host: Right.

Jouk Pleiter: Then the next layer is basically that truth layer we just described. Instead of list, uh, being a pass through architecture, how can we persist everything that is essential for that customer state graph and the context graph with the historical trail? I think ultimately that is really gold for any bank to progressively build that and have that understanding. It's almost like the holy grail in the industry. Let's see if we can get there. And I think we're now maturity is enough to get there. Then another layer on top of this, basically we're going to see workflows and we're going to see agents. So how do we basically execute the work and these workflows? And these agents will consult the customer truth, right? And then you need something we call Sentinel, which is basically the authority inside the platform to make sure every agent, whenever they want to do something they will call an API. They want to make a decision, they have to basically make an envelope with the total package of hey, this is my context, this is who I am, this is my identity. I'm um, acting on behalf of this, this client. I have this authority level by the way. This is the data I have from Nexus, the state graph. And I want to do this, I want to go to the backend and I want to execute xyz. Sentinel is basically the central control tower. It's a critical layer to make sure that agents can only do that within deterministic policies, entitlements and rules. So basically this is where the risk and compliance teams can come in and say, okay, how do we help initiatives in the bank to move from pilots into real work? You need to clear that, right? And especially if they become more autonomous. So in a nutshell, the layers are we need to do connectivity, we need to have the truth layer, we need to have the decision basically the police layer. And then we need to have the workflow, um, agentic layer. And these layers are basically working together every interaction. You basically go through all the layers.

Host: All the layers, yeah. And you made it clear that this is not about helping banks build a data lake or touching their data. These are uh, layers that connect to the systems and the data. What about LLMs foundation models that each um, like I'm thinking we have a bank in Singapore, we have a um, wealth manager in Europe, ah, another commercial bank in Boston. And they have different model choices or what they are allowed to touch on, how do these connect to to these layers?

Jouk Pleiter: If you want to operate as a wide label platform in banking you have to be interoperable by, by, by nature, otherwise it doesn't work. So of course we basically have an LLM gateway that can route to the right so but also even more interestingly, for instance in conversational banking we now have a semantic router that is basically just detecting on the question the intent. Okay, I'm going to solve this thing with a very expensive premium model and this one. Oh, I can use my own local small language model that I have optimized for this task. So I'm convinced we're going to see of course everything will be multimodal anyhow. So between the large labs, between open source and also between smaller dedicated, I think we're going to see an explosion in various smaller language models that are just optimized for specific subdomains and that are hyper efficient. So you can do the cost arbitrage between expensive commercial models, free open source models and in house tailored to small language models.

Host: Because the cost, cost is becoming a major factor these days. Uh, absolutely.

Jouk Pleiter: Yeah, yeah, you need to, yeah, you need to have your finops almost like you're built into the banking OS just to make sure that all the token utilization, you basically protect the bank. You really show them on how they can optimize the ah, cost. I think initially it's more important to get the agent's life and to get the governance in place and get the truth layer in place. But from, from the moment you really go into production. Yeah, I think the second phase is definitely having the infrastructure to do cost optimization.

Host: Yao, you have been in beta uh, with some of your clients over the past couple of months. What have you learned? What can you share with us from these beta? Uh, what is hard, what is maybe some anecdotes. Anything would be great once you see it live.

Jouk Pleiter: You don't want to go back, that is for sure. You really don't want to go back. So that is amazing. Nobody wants to go back. It is so efficient. From Nexus for instance, the truth layer, from the moment you have that in. Normally we had to create 500 different microservices to get data from let's say wealth management. You had to go to the portfolio management system, you had to go to the core, you had to go to the risk assess like you have to go to the CRM. So you had to manually for every kind of customer journey. You basically had to get stuff together somehow in the session and then print it on the screen. I will spare you the details, but it is very labor intense. I think from the moment you basically just like in an ERP or in a CRM, you basically now have a database table where you can say, give me everything about this customer. The amount of innovation you can now do on top of it is just way more productive and way faster. So having a source of truth is one thing. The other one is people. I had a blind, a lot of people had initially a blind spot I think for Sentinel, the decision authority. But I think it's the most strategic piece in the whole bunch because I cannot see a future where you're going to utilize agents at scale and the platform is not doing the reporting to the risk and the regulator teams. Right. You have to maybe. I'm not sure if everybody in the audience is deeply, but there's this open source initiative, it's called Open Claw. There was a lot of hype around it. For those of you who don't know it, you basically have some sort of a, uh, sandbox. You throw in agents and they can just figure out, they just go wild. Actually there's the Wild west without a police. And uh, of course that's very interesting early stage. But you would never do that in your personal desktop and you never ever

Host: would do in business.

Jouk Pleiter: You would never do it in a business. However, everybody does feel like, hey, agents that start to become more autonomous and they can collaborate and they can reason if you can put the right guardrails around it. If you can give them context so they can reason, read Nexus. If you can give them a police function so they can operate. So what, you basically want to harness the power of not only the large language model, but the power of autonomous agents reasoning and collaborating. I think if the banking industry is not able to jump on that train, uh, that's a major issue. If you, if that would be blocked by, hey, we cannot explain this to the regulator, so we cannot do it. Uh, I think that would be a massive disadvantage. So I think that's the first principle thinking that's really the problem you need to solve. How do we give them a shared source of truth so agents can reason? And how do we bring the police function?

Host: Sure. And the reporting even before the regulations and the policies are in place. Because for sure we will have to have an audit trail and full explaining

Jouk Pleiter: and maybe you recognize it. But what I also see is that in a lot of cases, people invite a risk and compliance team a little bit too late. Uh, and then they have to basically approve it. I think I would do it the other way around. I think with banking OS we basically your risk and regulator experts are part of the project team. Here's your login, you are a dedicated team member and let's figure out what are the right policies, what are the right guardrails. So you basically give them a graphical user interface to configure the platform so they can have full control about how they define and model the guardrails for agents to operate. They can connect to internal policy systems, entitlement systems. They can really utilize the infrastructure they already have and we help them to bind that uh, to create the agents. So yeah, yeah.

Host: What would you say to banks that look at this banking OS system and are a bit intimidated if you want because this is effectively a new core. Maybe they're not API first. Uh, maybe it's really too much for them and they do nothing which is the riskiest thing one can do these days. How would you suggest that they could approach this transformation in a small steps way?

Jouk Pleiter: Yeah, I think maybe for the record, I really think backwards is not the making of is is not a new core. I really think and maybe I'm maybe too sensitive to the semantics here but I really think it is an overlay on um. So it's not. We're, we have zero intent to be the system of record. We're not. There are systems of record. Are there? And back B is completely agnostic of the system of record. And actually it basically helps to solve the problem that you have multiple systems of record uh and you have multiple data sources that are relevant. We basically help you as the. I think back OS is the meta layer that sits on top of it. Think of this as the control plane that is one layer up. It's a separate layer that is helping you to orchestrate the work and the work is basically customer wants to do something. How do we make sure that we can do that as fast like that's also the core problem is that if the bank would double the number of customers you basically have to hire a double amount of office space and people because it was just linear more customers, more linear scaling. Because unfortunately the good thing is have more customers. We also get a lot of more manual work and a lot of more exceptions and blah blah, blah, blah. Right.

Host: Yeah.

Jouk Pleiter: I think that's the one we're tackling here with, with banking OS. Now how to use it. I think we can. We designed the platform in such a way that you don't have. You can keep all your systems. That's the first good news. We're not introducing, you know, uh, we're not saying we have to replace or keep your core, keep your decisioning engine, keep your policy engines. It's very friendly to basically everything that is around there. What it does as a new control plane, the overlay, it starts to coordinate an end to end workflow. And this is probably the best answer. Start with one workflow. We have value consultants, they do an MRI scan with a bank. So let's say, let's go look at your retail business, retail banking segment and let's just look at all the customer intents and let's analyze what is the resolution time. There's a whole group of customer intents that you can resolve in a second in your mobile app. So that's not the issue, that's automated. But then you go into the long tail of the stuff that is like where you see a lot of friction. Your example where you have to touch five systems in onboarding or an ongoing fraud check or uh, payment disputes, notoriously you have to go back and forth all the way to Visa, MasterCard and then all the way to the front those, those journeys. You can do an MRI scan and you can immediately see the amount of value leakage or the amount of inefficiency or resolution times that are uh 20 minutes or 5 weeks or 3 months depending on what's going on. I would say you can just pick one, you can just literally start super small and you can basically for that particular journey you build a mini ontology, you're building the policies in the rules, you're building your first workflows, you're building your first agents. And what is really nice, I think it can be embedded in your existing mobile app. So it doesn't have to be a back based mobile app. You can just basically we are open with banking OS to any third party mobile UI or uh, web UI for that matter. So it's agnostic to the client ui, it's agnostic to the employee a ah, workspace. It basically starts to streamline and orchestrate the workflow, the agentic workflows, the policing and the shared truth and ontology for that particular work. So I think you can get, you can get real results in six to 12 months because you're basically solving one workflow and then from there you add another workflow and you evolve and the

Host: intelligence, the context increases as you add a uh, workflow. It's not one plus one equals two, it's one plus one equals three. In this case, which brings me to. We opened up by saying that fragmentation kills intelligence. And I think through our discussion we can safely say that coordination and customer centric context is where intelligence is created. And this is what your OS system is enabling, basically.

Jouk Pleiter: Yeah. What I personally really like most about it is that set, uh, aside from, let's say, and I fully acknowledge your point that banking OS can sound a little bit intimidated. Right. If you compare it with just building a mobile app. Yeah, it's absolutely a few steps up. Right. And but I think the good news is that providers like Backbase are taking the lead and are basically providing the white label capabilities. I think large banks can build on themselves, but I think the m. Vast majority banks would probably say why do I want to do the plumbing? I want to basically build my own ontology, I want to build my own policies, I want to execute customer value. And I think also that is ultimately the core of the discussion. It's not about AI or tech layers or blah blah, blah. I think it's about the next 10 years, the operating model to become superior in fulfilling the customer intent. That is the core of banking. That's the core of every business like customer center. How do you do the customer intent resolution cycle? And I think we made great progress with mobile banking, but in all fairness, it is basically the easy use cases which we were able to solve with deterministic logic just to wire money and to unblock your card. So the feature set in mobile banking is still, I would say, modest. I think the next frontier is basically all these very labor intensive, expensive fulfillment cycles that typically include a lot of people in the front office and in the back office. So I think that's the next frontier. I will say maybe something controversial, but if you really do a thought process and if you would take a bank and you would have a banking OS like Backbase and you will build the ultimate set of agents. I think you can run a bank with 50 people, 100 people, maybe 200 people, maybe max 500 people, and you can run a bank with 5 million customers. I think. And this is very aggressive and uh, maybe this is like an Elon Musk type of let's go to Mars type of statement. But I would like to make the statement because I think the person or the institution that is as aggressive in this ambitious technology wise, I think it's feasible. So that automatically means in a competitive lens, the more aggressive people or the more ambitious people will strive for maximum efficiency. Right. And that will change the dynamics of how competitive are you ultimately and so I think personally, if you really take this as the most extreme, the current fragmented operating model with hundreds of applications and thousands of people basically navigating between these systems. I think that that economic business model, operating model is not sustainable.

Host: Yeah. And with an ever increasing ratio, maybe

Jouk Pleiter: you're good in the next three, four years, but then there will be a tipping point. It will be the same like classic retail and E commerce. There are going to be, there will be these tipping points and you have to have that uh, basically that agentic efficiency built into your operating model because otherwise you're going to fall behind.

Host: Yeah. And with ever increasing compliance and regulators

Jouk Pleiter: slow it down. So compliance in that sense is maybe a blessing. So that maybe compliance allows inefficiency to continue to go on for a long time. But I think we can now, as we discussed, I think we can now collaborate with compliance and basically take a glass half full approach. Have a control tower put in the rails and the rules to make it fully compliant.

Host: Yeah. Yoko, we could be talking for hours. These are exciting times, uh, around uh, the whole economy in our digital lives. How it's going to change. Let me close by asking you a personal question of beyond this launch. Beyond uh, AI operating system. What is energizing you? What is interesting for you? Whether it is some theme that interests you, I don't know, space, or how what kind of entertainment resonates with you, what books you're reading, anything personal that you would love to, to share with us.

Jouk Pleiter: I live in Amsterdam, downtown. It's a beautiful city. It's now spring. We just had our, let's say Liberation Day. Memorizing basically also the people that were fighting for our freedom. So it's always like we go to the central town square and then at 8:00 clock there's two minutes, uh, silence moment. Uh, the royal family is there. I was there last Monday and I really enjoy these moments where you think we have heritage. We have I think, uh, at least very personal. But people forget a lot of, let's say the privileges we have and you take them for granted. That is with personal health, that is with freedom, with all these concepts. And I was just happy to be memorized. In moments like that you realize, yeah, this is really important.

Host: Yeah, thank you.

Jouk Pleiter: There's one other thing and I think maybe that's important for every leader, both of us and others. There's this book from, called Innovators Dilemma from uh, Kristen Clayton. I think I would recommend everybody to

Host: read that book any reread it also

Jouk Pleiter: reread it reread it big time just to re. Because agency and a growth mindset and the willingness especially you. What I'm trying to say is that uh, as a leader of Backbase or as a leader of the bank or as an individual software engineer inside Backbase or any role, I think every white collar job is completely rewritten. The rules of how to run a software company, the rules are completely rewritten and the future is highly uncertain in SaaS businesses as you can see in the public opinion. Right. I think we're going to be very good. But okay, you need. This is the moments of highest level of clarity, like highest level of unclarity uncertainty. We're in the, we're in the fog of war within the software industry. I think the same thing is going to be applicable for the banking industry. The same thing is already happening for software engineers, lawyers, accountants. So what I'm trying to say is that what is the formula for an individual operator, an individual contributor, a business leader? I think like the willingness to go risk on super hungry to learn, force yourself to go there. I think like that was essential for us also to reinvent Backbase from a digital banking platform into a banking OS vision. Yes, yes, it is more intimidating. Yes, it is way more complex. Yes, I have to hire a very different breed of people to make and pull that off. But uh, I think it's part of an essential process to basically stay alive and to stay relevant. And I think it's not only applicable to companies and to banks, but for any professional that is facing AI.

Host: Yeah, yeah. What I'm hearing from you, Yoke, which relates to the unified platform vision and what you build is that you can't really just focus on, let's say servicing the customer. You really have to connect everything that goes also behind the scenes in the middle, in the back office as we call it. Uh, traditionally and clearly you have taken the bold step of creating that from being focused basically on the customer experience really now looking at the entire workflow and how things happen. Because only that way you can unlock intelligence both for the customer and for uh, the enterprise and allowing also for this agility going forward in a completely unknown, tragic trajectory of the whole economy.

Jouk Pleiter: The beautiful. I think you summarize it beautifully and I think uh, it's almost like we had to retrofit it the other day. Like we already have seen the pattern in the last eight years our most successful clients were thinking end to end. Our most successful clients, the chief digital officer and the chief operating officer got connected and they're starting to think holistically. And, uh, what is so funny? In all the uncertainty and all the new technologies and all the fog of war, one thing is still super true in first principle thinking. It starts with the customer, it ends with the customer. It's that simple. Right? And I think now if you can go back to that first principle thinking, okay, we are here to help these customers.

Host: Yes.

Jouk Pleiter: Okay, where do we have friction? And then now, with this new set of technologies, how can we basically, basically shrink the amount of friction and the value leakage? Uh, and how do we do that in a. Uh. And of course, then risk and compliance is basically another first principle thinking. Okay, we're doing this in a regulated industry. All right? Then there's only one path, and you have to come up with the right answer to solve that. Right?

Host: Yes, yes, yes. Joke. Thank you so much. Uh, it was, uh, an excellent discussion. Great pleasure to have you on board to share your insights. And let's hope that we can move the needle in the industry across board and help all the financial services businesses in this transformation, which is challenging.

Jouk Pleiter: We will. There's a lot of hope and possibility.

Host: Thank you.

Jouk Pleiter: Great. Thank you so much. I really enjoyed it. Thanks so much.