Nordnet AB (publ) (SAVE) Earnings Call Transcript & Summary

Hello, and welcome to the Nordnet Tech briefing. So the purpose of toda really zoom in on the technology that underpins everything we do at Nordnet. So Nordnet is a tech company at its core. We think that tech is a real competitive advantage for Nordnet. Clearly, very important for our business and, of course, important for the investment case. But unlike our financials, the technology that actually drives those financials is not well understood. So our goal over the next couple of hours is to build a true understanding of how our tech has evolved, what our key strengths and capabilities are and in the end, why that's important to the customer and to the business and to the equity story. Now we know that tech can get pretty nerdy and esoteric and hard to understand. So we made sort of bake in plenty of time for you to ask questions throughout this date. So we'll have a couple of short Q&A breaks during the program and then a longer one at the end. So feel free to send in written questions throughout the day, whenever you think of them. And at the end, take the advantage of asking a live question. So let's have a brief look at the agenda of today. So we have a package underfull of fun and exciting stuff, and you're going to meet some of the key talent from our product and tech departments. So we're going to start off with some opening remarks from our CEO, Lars-Ake Norling. And then the actual program split into 3 parts. And in between each , we're going to have a Q&A session and then a longer Q&A at the end. So like I said, feel free to send in questions throughout the day. And at the end, we're going to take live questions, both from our live audience here in the studio and from our audience tuning in through Zoom. So if you're tuning in through Zoom, just used to raise hand button at the end to ask a question. So clearly, we have a lot of stuff to get through, so we might as well start. Here we go.

In the early days of the Internet, Nordnet broke new ground and online savings and investments. We took stock trading online and offered a digital fund supermarket before everyone else. We launched the first mobile app for stock trading. And with we took social investing to the next level, a milestone towards becoming the leading platform for savings and investments. Now at the heart of Nordnet, product and tech transforms ideas into new features and products, ensuring our platform meets the moment with resilience and reliability, ready for market demands with an uptime surpassing 99.9%. For us, security is paramount. Our Zero Trust architecture actively protects our customers, securing their investments. designed to be as intuitive as it is powerful. Our platform provides an award-winning experience across a comprehensive product range, making every customer interaction seamless. Our customer-centric development is supported by a data and insights capability that allows us to understand our customers' behavior and shape our services in real time. We constantly move forward integrating machine learning and AI to refine our offerings, personalize experiences and streamline processes, always striving to deliver the most inspiring savings experience. Our focus remains clear, whether a release is incremental or groundbreaking, our cloud-powered platform is geared towards serving the Nordic savers ever better, fulfilling our ambition of building the best platform for savings and investments every single day.

Hello. How are you doing?

Great. exciting event today.

Great. So you're CEO of Nordnet Bank AB. Do you feel more like a bank CEO or a tech CEO?

That's actually a good question. I think it's a little bit mix of both, but clearly, we are tech at core.

Yes. And of course, a bit of a leading question today when we're here at the Nordnet tech briefing. So hopefully, today, you feel a little bit more like a tech CEO. So do you want to tell us a bit about how tech plays into our overall strategy?

Yes. We can do that. So let's move to the next slide there. Great. So we have 4 strategic focus areas at Nordnet, starting at the top. We want to have the most satisfied customers. And we want to be a one-stop shop for same investments with a really good customer experience. And to get there, we are building the best platform for savings and investments. That's our main focus that we work hard with every day. And the fundamentals for this is our cloud part scalable tech platform. Then the second area is engaged employees. We know we can never have happy customers unless we have really talented and engaged employees. We want to, of course, see a high satisfaction among employees, but also that we can attract and retain top talent. But also here, Techpac.we are in the forefront of tech using the cloud technology, and that also helps us to attract really key talent and not least within engineering. The third area is sustainable business. We are a trust business. We need to earn that trust every day. And especially, we need to take care of our risk and handle risks in a good way, not least the compliance risks. But to do this in a scalable way, we need tech not least in the AML area where we need to work with a lot of customers and a lot of transactions. We also know one of the top risks we have in our company is cybersecurity. And it's, of course, really important that we have a modern tech platform that's built for security, and we're going to touch upon that later. Last area is profitable growth. We have a fantastic growth potential in the Nordics, and we want to capture that one, of course, by continuing to taking market share in the growing savings market. But to achieve this, it's important that we have a tech platform that can enable rapid release of new features, but also new products. And as you know, we have a very scalable business, and we want to continue to have that and a strong focus on cost control. But also here, TechDay a role is our cloud part scalable platform that enables this for us. But of course, then we work a lot with automation. But all in all, as you see here, tech enables us to deliver on all our strategic ambitions. That's the core, and that's the key, and that's a focus also for today. Thank you.

Thank you so much, Look. Okay. It's time to move into the actual tech part of this tech briefing. And I can think of no better person to kick this after our Chief Technology Officer, Elias Lindholm.

All right. So today, our technology really enables our strategic ambition, but that has not always been the case. Going back to 2017, when Nordnet was delisted, our technology was a clear competitive disadvantage. Our user experience in our channels was not up to par with what would be expected from a modern savings and investment platform. Our technology didn't support the fast innovation that we needed and our platform suffered from recurring outages. So we started during the delisting period to rebuild our online services as well as the back-end systems that power our platform in order to improve speed of innovation and the experience in our digital channels. And 2 years into this journey, we were struggling to make progress -- and from an outside perspective, more or less everything looked the same. Our user experience was still not what it needed to be. Our platform stability was still not what it needed to be, and innovation was still slow. And I'm going to show you a short movie from the Nordnet summer party in 2019. And I can tell you ahead that this was not the proudest movement during my engineering career. [Audio Gap] Today, that movie is a lovely piece of Nordnet history and culture. And thankfully, a lot has happened since 2019. Today, we are proud both about the stability of our platform as well as the user experience in our digital channels. And today, we also claim that we are the most innovative fintech in the entire Nordnet region. And what we mean by that is that we are renewing and improving our online services at a higher pace than any of our competitors. And what we are going to focus on today is both what we have done since 2019 to turn this situation around and also talk a little bit about what we are going to do going forward. And I'm going to start with enabling innovation at speed. In today's digital world, if you want to be innovative, you need to be really, really good at software delivery. And of course, the more dependent your company is on technology, more true this statement will hold. For Nordnet, for instance, if we want to improve user experience in our channels, deliver new products or reduce risk, for instance, improving platform stability or increase automation in our internal processes, software delivery will be at the core of all such efforts. And the same goes if we want to improve compliance with regulation, improve security or a wide range of other important aspects of our business, software delivery will be at the core. And I really like this quote from Sachanadela that every company is a software company, and you have to operate and think like a digital company in order to be successful. And the way we operate at Nordnet bears much more similar resemblance with what you would see in a modern technology company than a traditional bank. At the core, we have 29 empowered product teams who push our offering forward on a daily basis. And in order to do this, we need to have the right people with the right skills and the right mindsets who collaborate closely together and have clear objectives directly aligned with what we are trying to achieve as a business. But then our teams need to have the autonomy and freedom to start to pursue those objectives in the best way they see fit. And our teams are cross functional, which helps improve autonomy. We typically gather people from product, from design, from analytics, from engineering, working closely together to push our online offering forward. Last year, we made more than 30,000 production releases, and this is a number that we will get back to throughout this presentation because we really believe that production releases or production deploys is one of the best, if not the best productivity measure for a product team. And Rasmus, our CPO, was recently asked this question at a conference. How do you manage to approve 30,000 releases each year. And this question sort of completely misses the point we're trying to make here, approving releases is a measure of control, not generating speed. I've seen firsthand for all the wrong reasons, how security specialists or architects or designers or product people are going to gatekeep all releases and that drains speed. Our teams has all the power to push the production when they see best fit. And we typically say that one what we have in production is better than what we have -- but once what we have in our test environment that is better than what we have in production, then we ship it. And that's how we enable innovation and software livery at a high speed. And the benefits of operating this are substantial. Release frequency really drives business agility. For instance, when we release frequently, we gain earlier feedback on the software we are building, and that allows us to start iterating, refining and improving those services earlier. Further, when we lease frequently, we also gain increased flexibility. Put simply, we had more than 30,000 opportunities last year to course correct and start pursuing new objectives if we wanted to. Third, when we lease frequently, we also get faster time to value. Instead of having software waiting in some kind of testing environment, it gets immediate out in the hands of the users of the software and start to generate value. And last but not least, when we release frequently, that typically also means that releases are smaller, which means less risk. And we have this saying in the engineering community that we prefer to receive Walnut in our head every week or so compared to coconut in head every month or so. But then in order to operate a speed, you also need supporting technology. And going back to 2019, we faced 2 huge problems slowing down our software delivery. And the first one was technical debt. Technical debt put simply, is a necessary complexity in our software that adds extra lead time to all engineering activities. When technical debt is high, but 2-day engineering activity can easily turn into weeks and even months of work, and it drains people's energy along the way. Both the product team working hands-on on the technology as well as the stakeholders who are waiting for new software that never gets done. And technical debt was one of the primary reasons why we started rebuilding and replatforming our online platform and the back-end services back in 2017. But in 2019, another huge problem had also surfaced and that was infrastructure management. Our product teams who were supposed to rebuild our business services, we're basically spending most of their time building and maintaining infrastructure. For instance, it was this particular piece of key infrastructure that we needed that was estimated to take 6 months to deliver. And to make things worse, we weren't even sure whether this infrastructure would support all the needs that we had. So something clearly needed to change here. And we as engineers sort of all knew what the solution to this problem was, and that is cloud technology. But back in 2019, we had this preconception that as a bank, we are not allowed to use cloud technology. Well, in 2019, tech compliance and legal joint forces and ask ourselves, can a bank operate in the cloud. And as it turns out, there's nothing preventing banks from using cloud technology. The only thing I pay here is GDPR, and that goes regardless of industry. So we started building the necessary governance structure in order for us to be able to rebuild our entire platform, online services as well as internal application in the cloud. And in our case, it happened to be that we decided on Google Cloud. And since 2019, we've been rebuilding our platform on Google Cloud. And today, cloud technology really provides us with the necessary infrastructure we need to focus on our core business, for instance, those 6 months I were talking about earlier, they were brought down to just 2 hours once we started building our platform in the cloud. And as it turned out, we didn't use that initial P3 infrastructure that we thought we needed. Once we were in the cloud platform, there were even better alternatives available. Cloud technology also offers basically global scale. If we need to increase production capacity, we can improve our capacity in the matters of minutes compared to the month it takes us to order new hardware and install it on our own premises. And from a security perspective, we are literally standing on the shoulders of giants and all the investments that Google makes on a yearly basis in their platform and making sure that their cloud is very secure. And Google takes care on the ongoing maintenance for us so that we can focus on our core business and building an amazing platform for savings and investments. And last but not least, one of the core themes in the Nordnet strategy right now is to increase leverage from both data and artificial intelligence. And as you will learn more about later in this presentation, and cloud technology provides us with all the necessary infrastructure in order to really deliver on such emission. And today, our goal is crystal clear. By the year of 2030, our traditional data centers will be a thing of the Nordnet history. And as we continue to move applications to the cloud, so far, we have rebuilt roughly 40% of our applications. As we continue to rebuild more and more applications, we will see faster and faster innovation, improved security and improved scalability across all Nordnet's services. And with that, I'm going to hand it over to our director of engineering Michel Cupurdija, who is going to dig even deeper into software delivery and how we do that at speed.

Thank you, Elias. All right. I'm not to introduce myself. I just got an excellent one by Elias. I'm going to talk about software delivery and how we truly enable innovation at speed through software delivery. It might seem trivial, but it's not. Before I get going, I kind of want to get you started with this. I don't know if you ever saw the former CEO of Microsoft. He walked up on stage, and he was sweaty, super enthusiastic, and he said, developers, developers, developers, developers. And it looked quite silly because it was because -- but I think he had a quite a good intention. And what he was trying to say is that the center of innovation at a company maybe not stems from the developers but from the development teams or product teams, depending on what you want to call them. And our vision to enable speed is in that spirit. You can see it here. We want to maximize the time spent innovating for the customer and minimize the time we spend on toil. So what is toil? In the industry, we tend to refer to toil as tasks or things that a human doesn't necessarily have to do. It can be something that a machine can do better or maybe an automated test. To give you a couple of examples, what that means, number one, if I had to press a button each time we released our web to make sure it works, that is toil because that is something a robot could do or an automated test. The second thing is applying security patches or bumping the dependencies within our applications. Companies spend significant amounts of time doing this. We don't. We've automated the patching process, and we leverage technologies such as renovate to bump the dependencies within our applications. That unlocks so much time for our development teams to do more important and more valuable things. And at Nordnet, we continuously look for these areas of toil, identify them and remove them. And over time, we've become really, really smooth. And this is not something you can do overnight, but we've been doing it for many years now. All right. But to kind of get into it then. You can talk about software delivery speed in so many different ways. I've chosen to kind of break it down into a few categories that you will see now on the screen. The first one, leveraging off-the-shelf products. You see under there, I've written build versus buy. So whenever we build something or whenever we buy something in unit, we're very mindful of our size as a company. We're respectful of the complexity of the industry that we're in. So we don't necessarily build things just for the sake of building them. So an example of something that we could buy or procure is cloud infrastructure. That is not something we need or want to maintain. It can also be open source frameworks such as react and spring boot that I'm going to talk a bit about later, that enable us to focus on customer innovation and not fiddling with technology all the time. All right. So what do we build them? I'll take a pretty strong stance here and say things that we built are, for example, our trading platform. That is something that is super complex. We could technically buy elsewhere, but we choose to build that from scratch ourselves. And that is because we want to be able to tune that customer journey to our customers' needs. It's truly a core business at none. So that is something we choose to build. Moving on. Reusability. In many areas of the software delivery life cycle, you can get stuck into the trap of reinvention. We've all heard us saying reinventing the wheel, and it is actually the case. At Nordnet, we don't try to discourage but rather encourage a culture of knowledge sharing and reusability. And through that gain speed. A couple of examples to make sense of that for you. Number one, we have a design system. The design system is what allows us to have 10, 20 teams working simultaneously in our web -- our app, while still making sure it looks good. I'm going to touch a bit more on that soon. Another thing is software templating. I'm super proud that at Nordnet, we can go from having no code at all to having a web service in production with the database in a matter of minutes. Just imagine what you can do with that power. You're no longer wasting time, but Elias talked about those 6 months are no longer there. But of course, this is nothing without the third point, which is automation. Our mantra is no manual, nothing. Of course, at Nordnet, we apply principles that gets talked about a lot, continuous integration, continuous delivery, which are essentially fancy words for saying or rather taking the source code that the developers produce, packaging it and putting into production. Of course, a very oversimplification and involves a lot of automation, but regardless. And of course, automated testing. I'm going to be mentioning automated testing quite a bit tonight, but that's something we really, really do at Nordnet. We wouldn't be able to do 30,000 releases per year without and having the uptime that we have without having heavy test automation. All right. But last but not least, all of this culminates to the last point, the ability to release at will. At will does not mean next Thursday at wheel means that the software platform enables us to truly release when we will want to. Today, this moment, on Friday, on Saturday, or rather maybe 50% of the customer base or only employees to make sure it really works before we roll it out. The technology platform is not the limiting factor. We can release at any point of time with super high confidence. This is quite a lot to dig in, but I'm going to double-click on this even further and talk about how this manifests itself in our different development platforms. So let's have a look. We tend to talk about our web and our app and our back-end services here represented as a cloud because that's where we do the primary -- that's where we do the most development today in our repack and services. So starting with the web. At the top, we have React. It's an industry-leading JavaScript framework for building web applications. It's developer friendly and highly scalable. And it's highly scalable in 2 ways. Number one, it's scalable in the fact that it's performance. It just truly is able to deliver to our customers need when we need it to. The second point is, it has also allowed us to create a source code structure that truly allows us to have 20 simultaneous teams working in this code base at the same time. That is truly a marvel and allows that speed. The second point, editorial content. The post of the market is super important to note. We need to be able to get the content to the people using our platforms as soon as possible. But we also want to be able to do this without compromise. What we do here is we leverage content, headless CMS, allowing us to be in full control of the entire rendering life cycle. You might have visited websites where it flickers. Things change as the page is loading it might look completely different every time you visit it. That is not the ambition we haven't known it. We want you to have a smooth and streamlined experience with personalized content delivered to you. Moving on, the third point, API driven. The web platform does not contain any business logic, how much money you have, how much of that money are you allowed to trade? None of that logic exists in the front end, rather in the back end, which allows the web to be laser focused on providing or building a great UI and great UX. And last but not least, of course, like I mentioned before, the design system. The design system consists of many different things. You will hear more about it later from Mattias Schyberg, our Director of product. But it's truly what allows us to move fast while looking good. We have UI components that allow different teams to simultaneously design different web pages, but having them look the same. We don't have to constantly reinvent the way our button looks, for example. Moving on to the app. And here you might ask, "Well, wow, how do you manage to build an app with the staff that you have? And the funny thing is this, very little changes when we do development for the app. That is a superpower of Nordnet. I've highlighted the only change here really, and that is that we're using another framework that is very similar to the one we use on the web. It's called React native. It allows us to use JavaScript to write cross-platform apps for both iOS and Android. Again, it's developer friendly and highly scalable. But something that is a true power to Nordnet as a company right now is the fact that we've noticed that transferring between writing racks and recognative is actually quite easy. So we can very easily migrate our engineers and shift their focus very easily, allowing for nimbleness in the organization. Of course, we need a datorial content in the app as well. But the funny thing is we're leveraging the exact same system for the app and for the web. So the editors are working in the same tool all the time, able to push content to different platforms or both platforms at the same time. Naturally, it's of course, API-driven, and app has its own design system because the screen is smaller and went to communicate complex things with different means. Now Mattias Schyberg is going to show you what the app actually looks like in real life and kind of what that -- how would that manifest itself?

Thank you, Michel. I'm Mattias Schyberg, Director of Product. He has now at, and I'm going to demo some customer flows that we have in the app. So this is the first screen that you see when you log into Nordnet app. So it's -- you see how much money you have. You also see how you develop against an index. And you also further down, you can see how you developed against an index today. If you scroll further down, you see your different accounts and how that developed, and you can also set up a monthly savings plan, if you'd like. If you just go further down, you see how you're allocated between different asset classes and also different instrument types. And further down, you see our top movers in your portfolio. So you see the ones that are the most positive, but you also see the ones that are the most negative. But now I'm more interested in finding a new investment. So I'll go into watch list. So this is a new screen that we created together with our marketing department. So here, we really used our content platform, our CMS platform to really create curated list with different themes. And as you can see, what's trending at the moment is the dividend list. So you can see highest dividend yield in Sweden, monthly dividend stocks, dividend aristocrats and then you have the Warnbuffet list. And maybe I'm more interested in dividend aristocrats. I'll click on that list. And so here, you get a description of what the list is all about with a nice image. And as you can see, you also get those really lovely logos that we added a couple of months ago. And you can see all the different stocks that we have in this list that have our dividend aristocrats. Maybe I want to look further into Coca-Cola, so I'll click on that stock. So when I go into the stock screen, you can see that I get a similar view as I got on my portfolio review. So you get a graph with different time periods, you get the last price and how it developed today. It can choose different time periods, of course, and you can also bench it to whatever you like. If you scroll further down, you can see that you get some key dates to keep track of. So for example, you have the payout date here for Coca-Cola on July 1. So that's something that you might want to know of if you're going to invest in the stock, you'll get dividend payout at that specific date. You also get about section about the different specific stock and also some key data. If we go back up, you can see different tabs. We have overview news order book for them. Let's click on news. And here, you can see that we list the latest new verticals about the stock, and I can click on what's been written about Coca-Cola today or during the last month. So that was a brief demo of our app. I'm going to come back and demo some more later during the day, but now I'll give it back to Michel.

All right. Thank you, Mattias. Let's see if we can get the right screen up here. Anyway, I can start talking what the slide says is replatforming and what do we mean? Here we have... And I'm going to tie this back to what Elias was talking about, which is technical legacy. Technical legacy can be addressed in many different ways when it comes to addressing software. Just a couple of examples, if you can, of course, isolate the legacy, you can build it away gradually. But we've chosen to use replatforming. So what is that then? Some fancy words or No, it's not. It is what it is. It basically means rebuilding the platform from scratch. For the web and the app, you saw Elias showing examples of just that. The web used to be ugly and the app was just terrible and now they look excellent. But when you talk about the back end, it tends to be a bit more complicated because you have more connections between several old generations of platforms, which we had at Nordnet. So which approach did we then take? Well, I'm not going to leave this high in too long, but we decided to similar to the web and the app, we build the entire online platform on the cloud as well. So not a traditional lift and shift into the cloud that you see a lot of companies doing. I think we can get so much more out of the cloud by using this approach. All right. So now you have a bit of a context about what we did with the back end and why we're making such a significant effort moving into the cloud with our back-end services Getting back to a familiar slide, the back-end services, again, using a similar setup. At the top, you see spring boots. It's a web framework for writing Java, the programming languages, which allows us to really focus on customers and not again, fiddling with the technology that we don't need to fiddle with. The second point, infrastructure fit for purpose. On premise, we're bound to traditional monolithic SQL data betas as many companies are, and like Elias talked about, putting new infrastructure in place takes a long time. On the cloud, we use no limitation as limitation because we can fit -- use any infrastructure that we see fit. For example, we can use Google Pubsub for events. We can use memory store, and we really need to access data fast. We're cloud SQL, when you need a traditional SQL database. The trick is we also don't need to manage it when we're using it on Google Cloud. But none of this really matters without the third point, which is infrastructure as code. Infrastructure's code means we define everything as code, compliance, the way we set up databases, you name it. And what that really does is that it puts infrastructure at the fingertips of the developers. They no longer have to walk into an infrastructure team and say, "Hey, guys, you have time to set up a database for us. No, they can do it themselves, truly shifting left on things like infrastructure. And last but not least, again, I'm nagging about this, but heavy test automation, we can really release with confidence here in this area. And within the back-end services, and we do the vast majority of the 33,000 releases that we did last year. And we could not have done that without the test coverage that we have to support that type of risk. To kind of summarize this segment then. So software speed or delivering software at speed is a culmination of many different things. And we've worked on this for a long, long time. And we're still going to be working on this for a long time. It's a continuous effort. But we've made the investment, and we've come a long way. So jumping into the next section, which is this a fairly related segment, scalability and how scalability can provide availability and how cloud can really help us here. I'm going to talk about this in a bit of an abstract sense. Here's a graph. I'm going to ask a rhetorical question here. I don't expect you to answer, but what do you think this graph represents. These are the numbers of orders inserted into Nordnet's APIs on a daily basis. You can draw a lot of interest and conclusions from this graph, but I will kind of talk you through it first, so you can follow along with what I'm thinking. To my left here, rather to your left, you can see a peak. This is when the European stock market opens. I don't as a 9 and the peak there less or 5, 10 or 15 minutes. Then you can see the platform kind of going back into a snooze that I'd like to call it. You can see further away, traffic goes up again. That's when the U.S. stock market opens. So imagine you would buy hardware to match this graph. How would you do it? You would look at the graph at the top and say, "Hey, we need to match this top at least. But I've deliberately excluded the y-axis here. And that is because this peak varies quite a lot, 4, 5, even 10x in height, depending on market situations. For example, U.S. elections going to be a big driver of this peak. So not only do we have to buy harder to match this peak, but also some imaginary peak that we don't know about. And simultaneously, our customers expect the platform to work here the most. This is where they need us the most. And the great thing about cloud is that we don't need to buy hardware, like Elias talked about, we can scale it up as we see fit, depending on the load that we're detecting that day, we can scale it up to match customer demand. And the great thing, of course, from many perspective is that we can scale it down when we don't need it after that big peak, saving both money and the environment. But how does this then really manifest itself in something like a real use case that the customers interact with. They don't interact with this. This app view is a stock screener that we like to call it. Here you see the graph of Nordnet's and you take that graph for granted. It should be there, it should always work, especially when we're in this peak. However, getting this out to the hands of the customers into their app is a complex process. From the source of information, enriching it distributed simultaneously to so many clients is difficult. On premise, this was a problematic child. And I'm going to be frank to say there were certain days where we would cross our fingers and say, hope the graphs work. That's not where we are today. This is not where we are today. Today, this is no longer an issue for us at all, and it just works. So to just give you some perspective of what that really can do for us. So moving on to the lines that Elias was talking about in the next slide. Here. So what does this mean? What do these ones actually mean? To your right, you can see 233 minutes of downtime in 2023 that represent roughly 4 hours of downtime. But if we jump back just a couple of years, 2019, 99.5, how much is that really? 1,200 minutes of downtime. That's 20 hours of downtime. And we've been able to do this decrease or rather improvement in our stability while also handling a 4x increase in traffic during the same period. I think this is a true testament to what the cloud and how cloud Committee accelerate what Nordnet is doing. And the effort that we put in is really paying off. The more systems like the graph you just saw, we migrate to the cloud, the more stable our platform is going to be. Now I'm going to hand it over to Elias, who's going to talk about security.

All right. Thank you, Michel. Being a technology company operating in a trust business, we naturally take security very seriously. And the threat landscape in 2024 is challenging. 10 years ago, the only thing or the primary thing at least we needed a concern about was cybercrime. But over the last years, we've also seen cyber warfare emerge with threat actors whose all intent is to either destroy or course disorder. That being said, the largest risk for Nordnet and for similar players are still are a ransomware attack. This is from a report that was published last year, where they predict that a company will be struck globally by a ransomware every 2 seconds in the year 2031. And in order for us to protect against ransomware attacks as well as other types of cyber threat, I would say that the single most important thing in our security strategy is our shift-left approach to security, which basically means that we engage our entire 250 people tech team in securing our platform -- and this is a pretest contract compared to the traditional way of security, where we have a central team who owns security that typically very quickly becomes a bottleneck. Both in your ordinary releases where they are going to gatekeep your releases to ensure security but also in all security work in the security work itself and improving security of the platform. What we do at Nordnet is what's called shift left by giving each product team full ownership of the security of their services. And that means that we build the security in from the start once we start developing new services. And this is a lot more proactive approach than trying to bolt on security afterwards. And when you have engineering teams in charge of security, you also see much, much higher levels of automation, which makes sure that Patris and vulnerabilities are removed and applied at a higher pace. But then if you would look at our organizational chart on a high level, it would look very similar to traditional setup because we also have central security functions, but their mission is significantly different. Our central security team, they enable our shift left approach. They drive our security strategy, which includes this shift-left approach, which I was referring to. And they also do continuous maturity assessments. They drive internal training, they do external threat monitoring, all to enable our entire tech department to take ownership of security. Then we also have a 24/7 secured operations center in partnership with Trustly. And I also like to stress that we have outsourced internal audit at Nordnet and that is a very good thing from a security perspective because we run audit each year in different areas related to security. And in those orders, we always have people with very deep skills in the specific area being audited who can provide us with very precise feedback on what we need to do in order to further improve our security posture. And I also want to point that an industry benchmark called UpGuardsecurity reading, UpGuard is a tool that scans our platform from the outside and looks for weaknesses. And in less scans, we received an 89% score in this benchmark. And you can compare that to the industry average of 78% for financial services. And it's worth pointing out that the financial services industry is among the top-performing industries in this benchmark. So if we compare Nordnet with players in other industries, we come out even further ahead. Going forward, I would say that the single most important initiative that we are driving right now in order to protect our platform is our move to the cloud. As I mentioned earlier, we are literally standing on the shoulders of giants and all the investments that Google makes each year into improving the security of their cloud. Our cloud platform offers Serotust, defense in depth, strong encryption, both addressed and in transit. And as Michel said, we treat all infrastructure as code. And that means that we can make a full disaster recovery in just 4 hours. And I'm not talking about a single application here. I'm talking about all the more than 200 applications that we are currently hosting on Google's cloud, we can recover all in just 4 hours. But then we also have a risk-based security road map where we each year make many tactical investments to remove the most pressing security risks on our old platform. And these are just a few examples of some initiatives that we've been implementing over the last years. I'm going to leave you with this one. In 2019, when we started rebuilding our platform in the cloud, there was someone who asked us how we, as a bank, there to build our platform in the cloud. Well, today, we ask the opposite question. How does someone try to -- how dare someone try to run the bank on their own infrastructure?

Welcome back. I'm Matthias Schyberg, and I'm going to talk about product development. So here at Nordnet we're all about working agile and -- for us, agile means being flexible, not only in how we develop products, but also how we work. And I think it's important to not take the agile manifesto and just apply it across your organization. I think that's what many of the larger corporations have done during the past years. And then they say that they work with the safe framework or that they work agile. But I think it's about taking the different things, bits and pieces from what's agile and really make it your own and make what's -- apply really to your organization. Michel and Elias talked a lot about speed and it's not only about technically enabling speed. It's also about the culture to be able to have the confidence to release early and often. The possibility to make mistakes and roll back, go forward, try iterate. And then capacity to iterate everything they do, both smaller features, larger features or when you work on a new customer journey. And like Michel said, or sorry, Elias, when something is better in test than it is in production, we ship it to production. At Nordnet we're organized into customer domains. And that means that all our product domains, they are based on different customer journeys. So we have foundation, discover, execution and pension and partner. And then we have the foundation that's layering on the platform domain. And all these product domains, they are based on different customer journeys. So we've grouped different customer journeys together and form these product domains. And so a product team can own different customer journeys. For example, we have one team owning become a customer and open an account, those customer journeys while another team owns deposit money and withdrawal money. And the product development team can look like this. It doesn't have to, but we try to place the skill set needed in order to solve problems, build products and essentially solve and provide customer value. So as he could be consist of a product owner, a back-end developer, a full-stack developer, a front-line developer, an analyst, a designer and an engineering manager. And these teams, they create their own goals based on the overall company strategy. And these teams, they maintain, but they also innovate within their customer journey. And that's important that they get that ownership and that freedom to do that innovation, both for speed but also to create the best possible customer value. When it comes to larger products or for example, we want to launch a new pension offering in Denmark. That's not a problem or a project that we provide to a product development team, but then we work together in the entire organization to really solve that problem. So then we joined forces a business area with the marketing department with 2 product development teams and even legal shipping. And these guys together, they solve the problem and build the product eventually. And based on where the product or project is tilted, it could be the product owner, the engineering manager or the business owner that's responsible for leading such an initiative. When it comes to design, we, of course, want to create the most inspiring savings and investment experience for our customers. And we try to do that with design. So we want to create a delightful design with visual appeal, engaging content and personalized experience with data. And how do we do that? We have a defined process and this process, you shouldn't see it as the truth that you have to follow, but rather a map and can guide you where you are in the process. And depending on problem or product that you want to solve or build, you can follow different steps, see checkpoints that's there to help you. Also, it's important that we have our clear brand guidelines, how do we look, how do we sound? How does our brand look? How can we use the logo in different situations? And how do we use our colors? And then Michel talked about this from a technical perspective, but it's almost more important from a design perspective, the design system, a designer should not recreate a button or a table or any feature function. You should just be there and you should be able to apply it to your customer journey and the customer need that you're trying to solve at the time. I know that we're all about developing products with our customers. And we do that through beta invites. So both inviting external customers early to a function. We can also invite employees, friends and family or different selected customer segments to start using a feature or a new product. We also have our user panel where we invite customers to come with feedback, both on existing products but also new products and customer journeys that we want to launch. And then, of course, all the customer feedback that we get through customer service, social media and, of course, the own channels that we created through giving feedback through our app and also through our web. Right now, we're seeing that the customer behavior is shifting quite a lot. And on this slide, you can see that -- the customers that joined us before 2020, they were a lot more spread out through what channel they preferred and even mainly tilted using the web, placing orders through the web, while looking at the customers that joined us 2024 and after, those really tend to use the app instead. So we're really trying to shift our mindset here when building and trying to capture that growth and really bring value to our customers. And if we bench the 2 platforms against each other, you can see that we have quite a lot higher daily average usage on our app compared to our web, while when it comes to number of trades, it's quite similar, actually. And this was not the case a few years ago. Web was much higher in terms of number of trades, but that is really catching up. Lastly, I want to talk about Shareville. I'm not sure everyone is familiar, but we have our own social investment platform. And so we have over 400,000 customers that have created a shareable profile with us. And that's quite remarkable to have that many users on the social investment platform. And we can see that these customers trade 1.5x more than the average Nordnet customer and also brings in 1.9x more commissioned than a normal Nordnet user. And Shareville is really all about to put your money where your mouth is. There's a lot of investment forms out there, but Shareville, you can go into someone's profile and see what have they invested in? Have they invested in the stock that they're out there talking about. And so I'm going to give you a little demo of what Shareville is all about. So come with me over here. And so here, I'm again on the overview for the web, and I get my portfolio overview with the graph and also how much money I have and also different winners and losers. But here is the serval section. And if I click into the shareholder section, I can go into my profile and look at what I've invested in. And as you can see, quite a scattered portfolio. Also, you get starts, depending on how good of an investor you are. And it's not about getting the most return, it's risk-weighted return. So it's really about how much risk you take as well. But let us go to Nordnet Rasmus, our Deputy CEO, his profile and see what he's invested in during the past years. So here, you can see that it's quite heavy in NVIDIA. You can also see that it has 3 stars. You can see that he has a lot of followers. You can also go into his activity and see what has Rasmus's been investing in during the past couple of weeks. And you can see that our emerging markets fund the other day. And all of this is anonymized. So you only see the percentage. You don't see the actual values in the share profiles. But let's see what people write about NVIDIA. So we click on the NVIDIA stock. And here, we're again on the stock screen. And so here, similar to the app, you get the performance today, but you also get a bunch of other data because we have more real estate to work with in the web. If I scroll down, you can see here that we have the Shareville section. And here, you have the different comments and what customers think about NVIDIA at the time at the moment. And you can like a comment, you can answer a comment. You can also click through more comments, if you like -- and that was it for me. I'll give the word back to Cecilia, who's going to talk about data and how we use data in our product development.

Thank you, Matthias. I would do like leased and take a step back, but focus upon our analytics platform and leverage our data. And not long after we were founded, we implemented our first business intelligence tool to support our need for data, reporting and insights. But 10 years ago, we understood that we were limited by the technical platform that we had. And as a part of the delisting, we also invested in migrating this to a new platform. We also added more data competence such as data scientists and data analysts. We invested in AI and had robo-advice through something called robosave and we also had an AI chatbot -- but despite our high ambitions, we couldn't turn this into value. A couple of years into the migration of the new analytics platform, we felt that we didn't get any progress at all actually. So we canceled that migration. The data scientists that we had, it was really hard for them to develop any machine learning model in the platform we have. And even if they managed to do that, it was impossible to get into production. And the AI chatbot we had was fired. So it was clear that we were lacking the technical foundation to releverage our data. The challenge we had was that we had a diverse platform for data analytics and experimentation. We suffered from port data quality, and we had long lead times in getting insights. We didn't have a great support for AB testing, and we definitely didn't have a technical platform for AI and machine learning. But we have our strategic capabilities and one of them are to be powered by data and AI. And what we mean with that is that we want to have a culture where data empowers everything we do. More in detail, that means that we first want to be a data-informed organization. We want everyone to have access to data they need to do their job in the best way. We want to drive a data-driven development product -- data-driven development and really drive a build measured learn cycle when improving our customer offering and really be fast in doing improvements and shipping improvements all the time to our customers. And the last part is that we want to harness AI and presentation when it makes sense to do that. But we're able to reach this vision, we still needed to migrate to a new analytics platform. And 2021, we started building the new analytics platform in the cloud. And the reasons for cloud is the same as Elias has talked about. This will enable speed, scalability and innovation in a completely different way than any other alternatives would do. But it was not enough with only the technical platform. We also needed a more data-focused organization. So what we did was to integrate data and insights with take to secure unified infrastructure and strengthen the collaboration and increase the efficiency. We created a data platform team to clarify the ownership of that and really that we needed to focus upon it. And since we have a complex business, we also need to decentralize the ownership of our data products to make sure that everyone understood how the data is used internally for all the analytical needs that we have. And this is the only way that we could ever scale the data capabilities in the way that we want to do. So what we have taken is our challenges within both the technical platform and our organization to put us in a situation where we will have a unified technical platform and a data-focused organization really enabling speed, high-quality data and experimentation. I will go into a little bit more about what this actually means in the daily work that we see at Nordnet will start with the foundation, which is just to be data from organization and have this easy access to high-quality relevant data to support you, whether it's regarding reports or insat generation or just giving you the right insights to make this difficult decision. One way to access this high-quality data is, of course, through dashboards, and we also have a self-service opportunity where you can analyze the data further, if you want to. Depending on who you are, different data would be relevant for you to monitor. And I will give you some example of dashboards that we have. The first one is our strategy dashboard, where we all can follow how we strive towards our strategic initiatives or strategic capabilities, such as power by data and AI and cloud-powered platform. The next one is where Matthias got his number from regarding the elective users, and we can see the similarities and differences between behavior and web and app and how these develop over time. And we are quite many that needs to understand what's happening in the market right now. So we have trade value dashboards. We have trade statchboard and positions dashboard to see that this is happening. These are the kind of customers that trade in an instrument or this has developed like this the last weeks or days or hours. And we can go all the way down to team-specific KPIs, where we, for example, I'm interested in understanding how the analytics platform is used both by the producers and the consumers. And this helped me and my team, of course, to make the right decisions and prioritize what we should or also like Matthias showed provided these teams that own a customer journey, really the right insights so they can focus upon creating even more value for the customers. And where we are right now is that 30% of our editors use the new platform on a monthly basis to get some kind of data or insights to support them in their daily job. We have doubled this over the last year. And of course, this will continue to increase as we add even more data to our platform. The next part we will talk about is data-driven development, and we really want that to be a part of our DNA and really drive the build measure learning cycle. And we have taken a situation where we had limited support for AB testing to a situation where we really have data-driven development. Experiments are now embedded in our development. And the majority of all the releases we do are supported by data experiments or leading metrics. I will give you a lot of examples on AB-test, but I would start with what it is and why we do it. And what we do is that we randomly divide our customers into 2 or more groups and give them different experiences in parallel. In this case, we replaced the sell button with the same monthly button. Why we do this is, of course, to be able to compare metrics between the 2 different versions. For example, conversion rate to first trade or commerce a new customer or the order flow. We can also isolate the change and rule out other factors. It's not something happening in the U.S. election, like Michel mentioned, it's more the change that we actually did that is causing this difference in commercial metrics as we see. We can always be much faster in iteration and can avoid spending time on development that isn't providing any value to our customers. And we can also be really safe, reduce the risk and test towards a smaller group of customers or a less critical flow because there are critical ones that we really need to take care of and be cautious about when we publish changes. But also coming back to what Michel mentioned, release at will. We released this and we go out with abet when we want to, and we can publish the changes where we feel ready and we have gotten the answer that we wanted from the data. So some examples of AB test, and Matthias already showed this earlier today. This is the watch list that is supposed to help our customers find new investments. And here, you have the old version. And what we did in the new one was to add Discovery bills. So it made it easier for the customers to find some area that they are interested in, like dividend or entertainment. And as Matthias earlier showed this, I guess it's no surprise that the result was positive. But through this, we had an increase of 15% of the engagement with the watch list. So of course, we help the customers finding their next investment a little bit better. The next example is that we also wanted to encourage our customers to save monthly. And what we did then was that for all the customers that visited a fund screen for a fund that they didn't own, we still showed the buy and the sell button, but they couldn't sell it. So instead, we replaced the sell button with a save monthly button. Through that, we saw an increase of 13% in number of customers starting a monthly savings journey. And when we are analyzing this further, looking at the conversion rate and the average amount in monthly savings and also the stickiness of multi savings, we can see that this actually adds NOK 52 million in net savings yearly through monthly savings. And of course, that's great, both for us, us and the customer. The next example is that we also want to encourage our customers to opening savings accounts since we have a profitable savings account with good interest rate. What we also wanted to understand was that if it was worth the well-being service to show the effective interest rate because we didn't have that in place. So we wanted to know if it was better to show the effective interest rate or could go with a generic copy. So that's why we had 2 versions. So if a customer visit account list, we added a small promotion with a open savings accounts with the effective interest rate and just a generic copy. In both these variations, we saw an increase of 100% for customers opening a savings account. And we didn't see any change between the 2 versions. So we didn't need to spend time on developing a service that would show the effective interest rate. But instead, we can focus upon something that we would provide more value to the customer. And once again, analyzing this further, we can see that during just 1 month, this small change added EUR 9 million of cash transfer into our platform compared to the one that didn't have this as an option. And as I said, we have a lot of critical flow. And with them, we focus a lot upon removing hurdles and making the user experience just as good as possible. The order flow is definitely one of those. And what we have done, for example, is having a preselected account in the order flow. This increased the conversion rate in order flow with 2.5 percentage points. Another example is that we added a number formatting for readability. So the customer actually saw what they were doing. And this decreased the number of validation phase, just errors that block the customer from placing an order when they wanted to from 32% to 1%. And this is one example of how strong it is to work with continuous improvement and the data-driven development that we do. The order flow, we can compare it to flows that we haven't improved as much and we work with us actively. And we can see that it's improved 3.8 percentage points. And this is for a customer journey that started 30 million times per year. So it has a huge effect for our customer and, of course, for us as a company. Now Matthias will demo the order flow.

Thank you, Cecilia. So here we are again, at the stock page, talk screen and see that I went into the ABB stock. And at this time, I'm interested in buying. So I'm going to click the buy button. And so here, you can see that I see 5 levels of the order debt on the bid side and the ASC side and also can change to latest trades if I want to see what's the latest trades been placed at at what price. And in the bottom, we have a configurable bottom tab where I can choose what account I want to place the order in. And also I can place what quantity do I want to place the order in and at what price. But I can also place the order in amount. So for example, here, I can put in SEK 10,000, and the price will -- the quantity will adjust according to the price that I set. And at this time, the price is set at the the the lowest ask price. And so I want to place it a little bit lower because I don't want this order to go through, so I'll click the minus line and a place order into the market. And here, I get a confirmation screen. What account did I place the order at what quantity price amount also if there's any transactional fees in this account? And then I'll click by. And now I get that my order is received and it's in the market. So I'll view the order, and I can see here is. And because we don't want it to go through, we will also delete the order for now. We can go back to Cecilia and how we work with data.

Thank you, Matthias. And if your order flow isn't looking like Matthias did, it's because we have a new bites live, which is trying to support our customers to buy when they know the amount they want to invest and not maybe the stock price A lot of success stories. And are we always right? No. And that's why it's so important to really test and learn what actually works and how the changes will do impact our customers. This is also a recent example when we wanted to encourage our customers to do more deposits or monthly savings. So what we did was adding shortcuts to the transaction screen where the customers can see their previous transactions. And the result was great. We saw an increase of 20% to 30% of customers go into the deposit and savings and withdrawal and transfer page. But when we're analyzing this further, we understood that the actual transactions didn't change at all. So we brought it back to the team, as Matthias talked about, with analysts, designer, a product owner and developers and started to think about why is this? Why are they not acting like we think they should. And when looking at the data and the customer feedback, we understood that the customers, they were in a different mindset. They were on this specific screen because they wanted to see something in a previous transaction. They wanted to filter on the deposits. We're not in the mindset of depositing more money because there was something they were searching for. So the actual result from this test, it was that we removed these shore cuts and will improve the filtering instead. And going back to the culture, where data empowers everything we do, which is what we say with powered by data and AI. The culture that we have is that it's fun to learn. And we have learned in every bit as I've shown, especially the last one, because we didn't see the result that we maybe were expecting. It's also fun to be proven wrong. We won't soon have 2 million customers. And of course, it's really hard to understand how they will act based on our changes, and we need to be humble about that and see what actually has the greatest impact. And that is the last point that it's really fun to see how our daily efforts really impact our customers, and this really motivates us all to just keep working on this and keep continuing improving our customer journeys and provide them with the best platform for savings and investments. And then going back to the time line that I left you with -- that we were canceling the migration of the new analytics platform. We stopped hiring data scientists and our AI chatbot was fired. The last year has been quite different, of course. We started with the cloud migration. And as Michel has talked about the replatforming has been so important because we are connecting the work with a new analytics platform also into that. So 2 years into the cloud journey, we started building a new analytics platform in cloud. And really, the last years, we can see how the leverage of data and our tools really has increased and that we are creating a culture now when data can power everything that we do, although we, oftcourse, have a long way to go still, which we also do in these kind of cases. But I'm also happy to say that we could, once again, start hiring data scientists and really provide an environment where they could create value. And that leads me into the last part of Powerboat and AI, which is harnessing AI and personalization, -- what we mean with that is that we want to boost our internal productivity. We want to solve customer solution, build customer solutions and solve business problems when with AI when it makes sense. And I will hand over to our data science Manager, Mathias Törnqvist, to tell you more about that.

Thank you, Cecilia. So harnessing AI. I will start to talk about the survey that I traded in 2020. They did a huge survey about the value they got from AI or comp cut from AI. And in this survey, only 10% of the companies reported that they gain significant financial benefits from AI. However, in this survey, they identified that the companies that use a learning approach, they name it organizational learning. They had a much higher chance of actually reporting a significant financial benefit from AI. So value from AI technologies. This is a projection done by Stanford Professor Andrew Ang. It is a projection about the value that different technologies create today and the value they will create in 3 years. And as you can see to the left most part, we have generative AI. It has very -- in a very short time, started to generate value. And in the upcoming 3 years, it will keep increasing in value generated. Next it, we have unsupervised learning, a more traditional machine learning approach, trying to find patents and relations in data -- and there, it's actually more value created today, and it will grow to a lesser extent than AI. But on the rightmost part, we have supervised learning, classic, hot dog, not too machine learning. And here we see a lot of value created today. And it will grow even more in the future. So with that, I would like to present Nordnet's strategy that is focused on preparing us for an AI-driven future. We are building upon our long-term commitment of integrating AI into our products to make them more personalized and engaging. We also want to utilize the anti AI. We've proven to increase operational efficiency. So how hard is it to do AI then? Not at all according to the CPTO at Spotify, which she delivered in a talk about making AI harder than it is. And I agree with Gustav, but you have to have some things in place. And one of this is technology and technological platform, something that we've been talking about a lot about today. Google Cloud is today giving us best practice methods and tools from their ecosystem. It's enabled us the usage of current technologies as well as future technologies. It minimize our technical overhead when exploring and executing on AI initiatives. And an example of this is there's a lot of AI models out there that people are talking about. And we've been experimenting with Gemini from Google, Lamar from Meta and mix roll from Mitral AI. And as we talked about, about buying or building, this we just used right out of the shelf to experiment with. So it was very easy and cheap to get started with. So Google Cloud is giving us a future-proof solution because at advanced AI and machine learning. It's integrated to our big data ecosystem and gives us a tool for collaboration and productivity. So what else do you need to make AI succeed? You need skills. And since 2023, we have started to hire world-class talent within data science and machine learning. And this we are organizing together with data analysts. So data scientists and data analysts working together with our business teams. They do that so we can align the usage of AI technology with their extensive knowledge about Nolet customers, products and business and the problem that we're trying to solve. So a couple of things that we do today. We talked about Shareville. And we think social investing should be driven by AI. So since we integrated Shareville to our platforms and that's enabled alone to participate in social investing. And much of the content that people are generating on Shareville is in the form of comments, ideally suited for traditional machine learning and generative I Today, we have developed internal tools for summarizing what people are talking about the last hour and also for flagging when comments are breaking our policy. Next, we have official we released in April, the FEED. It's a feed of news posts. It might be Sherrelpost, company events, et cetera. There's a lot of posts that are generated on our platform on a daily basis. So we built an algorithm to make the content as good as it should be for you. And this is done by modeling on how recent something is published on our platform, how relevant is it for you as a customer and your investment and how engaging is this. So other nodes might comment like or visit a new vertical or stock. And this is then together calculated to give you the best fit for you. So we also encourage the note in several parts of Nordnet, and one of these are marketing. Since last year, we have been heavily committed to utilize these tools when generating campaigns. These 2 campaigns was fully generated by mid journey. One is about our new net film network to the left and the other one is about our pension offering. And this gave us very like flexible speed when it comes to iterating on the campaigns as well as broadening the creative freedom when executing on this. And I would absolutely like to see how I will look in 25 years, but also how my investments will look then. So we are also upskilling our organization. We're continuously working on to enhance our organizational skills in generative AI. So we're trying to enable and encourage every part of the organization to engage with Generate. So to summarize, AI at Nordnet, we are leveraging our strategic partnership with Google. We're building upon our capability to utilize and evaluate AI in every capacity. And lastly, learning together with AI. And we're hiring world-class talent to lead us on the way on this journey. So with that, I would like to leave over to Elias to talk about industry benchmark software delivery performance.

Thank you, Matthias. All right. So far, we have talked about how we enable innovation at speed. Next, we are going to look at an industry benchmark on software delivery performance. And I want to start by just setting the scene. Software delivery performance refers to an organization or team's capability to deliver software with speed and with quality. And I mentioned at the beginning of this presentation that in order to be successful today and be innovative, you need to be really good at software delivery. Well, you just don't have to take my word for it because there is extensive research that shows that there's a very strong correlation between software delivery performance and an organization how successful that would be. And that goes regardless of the industry and regardless of whether we look at profit organization or nonprofit organization. And you can look at a wide range of success factors, such as profitability, market share, customer satisfaction and many others. Software delivery performance will be a strong leading indicator of how likely you are to succeed. And the benchmark we are going to look at is called DORA, it's published early in a report called State of DevOps. And what we really like about this benchmark is that it focuses on a few key outcomes rather on whether you're following specific processes or principles or using specific practices. And in order to score high in this benchmark, you need both to have the right types of people with the right skills, the right mindset, the right way to working as well as the enabling technology. If you remove one of those pieces, you won't score high in this benchmark. So for instance, if we take the best performing team in the world and give them poor technology, they won't be a high performer in this benchmark. And taking the other way around, you can have all the best technology in the world, but if you don't have people with the right mindset skills and ways of work in, you won't either score high in this benchmark. And DORA measures software live performance using 4 metrics. First one is what we have talked a lot about already today, which is deployment frequency, which basically tracks how often does the product team push new code to production. The second metric is lead time for change, which basically starts a clock. One team is done developing some new code, and the clock is stopped once that code is available in the production environment for the users of the software. And then the third one is something called change failure rate, and that tracks how often does the product team cause an incident as part of making a production release. And then the fourth and final metric is fail deploy recovery time, and that tracks once the team or Ifedine has introduced or caused an incident as part of deployment, how long does it take for the team to restore to normal operation again. That's the fall-eploye recovery time. And then each team gets a performance level that will be either low, medium, high or lead depending on how you score on those 4 metrics. And in order to reach the elite criteria, you need to reach elite level on each of those 4 metrics. So for instance, if you have one team that is elite on 3 metrics, but just high on the fourth metric that will be a high-performing team, not a lead performing team according to this benchmark. And I'm super happy that as of Q1 this year, 86% of our Nordnet teams are either high-performing or elite performing teams according to this benchmark. And you can compare that to the industry average, which in last year's report, which was 49%. And I'm also very happy that we have 0% low-performing teams at Nordnet. And of the medium-performing teams that we have more or less everyone has been forced to work in our oldest part of our code base where there is higher risk of causing incidence as part of a production release or you're working slower due to Tecta or lack of availability to cloud technology. So I'm confident as we continue to rebuild our platform on Google Cloud, we will see improved score on those benchmarks going forward. And just to give you some perspective of what software delivery performance means in practice. Well, for instance, we are releasing our customer-facing web every 1.6 hours on average. And we're making more than 2 releases each week of our native app. And all in all, as you've heard us mention many times today, we made more than 30,000 production leases last year, all improving different aspects of our online offering as well as the internal processes and internal applications that we use at Nordnet. Going forward, we will introduce 2 new metrics in our quarterly replatforming. First of all, and we're going to include this benchmark that I just presented at you on software delivery performance. And we are also going to include numbers on our cloud migration because we believe that the best leading indicator for how good we will be at software liver performance at Nordnet is how large share of our applications. We have migrated to Google Cloud. And as I told you earlier, software delivery performance, in turn, is a strong leading indicator for how successful we will be in the future. And as before, we will also continue to report on our platform availability.

Thank you, Markus, and thank everybody who's here today. We've been teasing this day, Marcus and I for a while now. So it is really good to finally be here to be getting to explain and boast to you all about what a wonderful technical platform we've built, but also what it means for us as a company for our customers and ultimately for you as shareholders. So let me start with what it means for us as a platform. And let me show a slide that I've shown many times before, but try to put it in a bit of a different context given the topic of today. So going ultrawide. Michel showed you what the API calls looks like on our platform. We see this peak where it spins up to 23,000 requests on European market open. It slows down, and then there's another peak when the U.S. market is open. Now this presents, like Michelle said, a unique technical challenge. There are a lot of other platforms, say, Spotify, that have a higher API rate, but it's constant. People listen to music as the sun moves around the earth, right? So we have this problem where it spins up, peaks and then slows down. So we have to be able to scale dynamically. Venata showed you how we design and how we build engagement in our app and web to the point where we have over 360,000 unique users in the app every single day and another 100,000 on the web. Now let me add some commercial aspects because as these people log in, they start putting trades through. And on a typical day in the first quarter of 2024, they're putting 216,000 trades to the platform, 63,000 of which are cross-border for a total traded value of $6.2 billion. Then on any single gate, they're also bringing up 1,000 new people joining us as customers, and they and existing customers are putting $300 million of fresh money onto the platform. We pay out some $63 million loans and to do $240 million of net fund purchases. And what's all this in service for? Well, to me is easy. It's building the best platform for savings and investments. This is our brand promise. This is what we promise our customers and everything we do is always in service of delivering on this promise. Now I get nerdy not when it comes to tech, but when it comes to language. So I love the ING form of this word. We're building the best platform. It's not to build because to build would signify that you can be done, I would instead say, we're never done. If we ever think we're done, we need to move the goalposts. So what you've been seeing here today, everything that my colleagues have been talking about is really putting the technology, the data, the design and service of this brand promise. So what is the key selling proposition of the Nordnet Platform? Again, I would argue that it is the combination of fantastic UX in the channels as Mattias has been patiently demoing throughout the day. but also crucially combined with a wide product range. Now if you allow me, I think there are a lot of fintech start-ups to have great QX. They have a beautiful app. Once you tap your way through that app, there is nothing there. You end up in 1 of 3 fund of funds. I used to work in an incumbent bank. So incumbent banks, they have all the products in the world, but they can offer the life of them, get them out through the channels in a meaningful way. So this is really where we shine. Add that we're also then delivering these products through smooth and automated customer journeys, all spinning on a scalable platform in the cloud where we work with what we call real-world agile. I think to me, that is the secret sauce. But let me take just a few minutes to do a slide each on the top 3 skills here because we spent a lot of time on the platform already. So starting with leading UX in the platform. To me, this is really about elevating the investment journey through intuitive design. And how do you do that? In many different ways, as my colleagues have talked about, but I want to just briefly mention usacentric approach. And Mattias talked about how we codevelop with our customers. And this is key to us running closed betas, running focus groups, inviting in panels of customers. And this is how we, for example, have been building travel over the last few years on to the Nonetplatform. Then we have my favorite word, delightfulness. We try to bring every day delightfulness to the platform in terms of liter interaction designs, little micro animations, whether that be sound designs or just how you feel this platform is interacting with me as I'm working on it. Combined with innovative features such as the curity watch lists that have been demoed a couple of times here, personalization, for example, the feed that Mathias showed. And then finally, the power of social through Cherryl I network. All of these things together really work together in order to deliver the superior user experience in both app and web. Again, if that was all there was, it wouldn't be much of a platform. So we marry this with a one-stop shop for savings and investments. Now this hasn't been the topic of today, nor is it now, but let me just briefly mention that once you log in once you tap through and become a one customer, you need to find what you want to invest them. That's what you're here for after all. So coming out of a heritage of securities brokerage, we have over 100,000 exchange chicken instruments. We built a fund supermarket with over 2,000 UCAS funds, along with all the tools and sorting tools and guidance flows to help you find what you're looking for. We made those available through pension accounts, whether that be endowment or occupational pension. Now given that we work with excess cash on the platform for customer counts, we also complement that with select lending products, chief among them being marginal lending, which is highly complementary to securities brokerage. And now following the announcement of the sale of our personal loans business, that retail lending line is really only a mortgage for our PB customers in Sweden and Norway to attract and retain held-away assets. Mattias Vibes showed you a version of this slide when he talked about how we organize ourselves. I'm showing it here as a way of how we automate and improve customer journeys. Now he talked about each domain owning some of these macro customer journeys. And as you may notice, they're all expressed in the first person, again, another example of customer centricity. I want to become a customer, inspire me to save and invest. I want to buy yourself security or who my pension, what have you. The highlighted ones here are the ones that we're working on this year, and we keep working through these customer journeys and automate, take out little friction steps, take out cost, take out operational risk all the time in order to improve the platform. But ultimately, we heard everybody from Elias on down talk about speed and speed of innovation. And I love this quote from Tim -- he at Gaping -- we don't have to be the best. We just have to be the fastest. And what I love about it is by being the fastest by being able to ship it will, like Michelle talked about by being able to AB test in real time, like Cecilia talked about, you pretty damn quickly would be the best. So that's why I keep harping on about speed and speed as a competitive advantage. We love showing this slide. It's been shown twice before today. So let me give you my spin on it. Elias just went through it, but pushing new features and functions every 1.6 hours just gives me as a Chief Product Officer and my entire product department, such a fantastic experience of working on this platform. And we are so indebted to our colleagues on the tech side for building and enabling this kind of speed. So we work so closely together. And without having this, I think it'd be much more frustrating experience for people like me who have very short patients. Again, just mentioning the app, every 3 days, a new version of our native app is out, that's counting iOS and Android as 1 release, not 2, and that includes the time that we need to go to Cupertino or Mountain View for approval before we release. What it also enables is for me to keep talking about always be shipping. Now this started as a joke originally a kind of a tongue-in-cheek internal joke. It's become a mantra. It's become a mantra, not just for me, but for product teams, our future teams, our marketers, and we are able to keep a higher rate of market communication around new features, functions to the market shown here on X, but also, of course, through other social media channels, including our own Sharebuildplatform. And I think it really is a testament to how we're always trying to improve this platform. We're always building the best platform for savings and investments. So if that's what it means for the platform and ultimately, our customers, what does it mean for us as a company and as a stock. Well, the kind of growth that we've been printing over the last 5 years would not have been possible without having a scalable and efficient platform. And that's what everything we've been talking about today has been about. So pushing a 90% top line CAGR in terms of customers, another 25% top line CAGR for savings capital is totally dependent on having a scalable and efficient platform. It becomes even more apparent if you take it to the P&L, with the top line revenue of 30% CAGR, keeping costs at only 5% through all the scalability, the efficiency, working through those processes, being better at finding cost efficiencies, meaning that most of this drops to the bottom line with a CAGR of 66% on the profit line. So let me end where Lars-Ake started. All of this technology really enables our strategic ambition, all of it. It's the platform, it's the cloud, it's the data and insights platform. It's the AB testing. It is the design. It is the way we organize our product and feature teams in order to really build that engagement with the customers that is our strategic goal. And Lars-Ake said in the beginning, this is a fantastically engaging platform for all of our employees to work on as well. And I've already shown you now in these charts what true profitable growth looks like and how that is enabled by technology. And finally, if you could hardback to what Elias was talking about when it comes to security, which is a key cornerstone of course, for any bank, but especially a digital bank, it will enable us to have a sustainable business. So with that, I want to say thank you. There are swag markers for everybody in the room in the back, don't forget to pick them up, and let me hand back to Marcus.

Okay. It's time for our first Q&A session. So we've got a bunch of questions written from our digital audience. [Operator Instructions] The first question here is about uptime.

So you said that you had 230 or so minutes downtime in 2023. Is that a couple of really big blackouts or a lot of very small ones. What does that look like? And then a follow-up, what's the most common cost of downtime? And is it possible to get to 100% uptime.

Yes. I mean, I guess I can start with the last question. I mean, 100%, we're not really aiming for that. The more names you add, the diminishing returns are so significant. It costs so much money and it takes a lot of time away from, like I said, innovating for the customer. We believe that 39 as it is right now, really provides a happy balance between risk and be able to innovate at speed. And to your other question regarding the minutes. No, generally speaking, like Elias said, I really love the walnuts in coconut analogy. We released so many times. That also means that we're really, really fast and rolling back. So if something goes wrong, we roll it back. In other words, those 233 minutes tend to be spread over the year. So that's generally how it looks.

Okay. I can also add that there was a question regarding the most common cause for incidents. And by far, it is third parties. And last week, for instance, we had an outage in Bank affected in tire Sweden. And when we track availability, we do that from a customer perspective. So regardless of whether we were causing the problem or in third party, we hit our downtime or our availability, I should say. So the single most common source is external partners. And the way we can protect, of course, is to have redundancy. But in the bank ID case, we have free ID as a redundant alternative. But unfortunately, so far, there aren't that many suites who use that. But at least for those who do, it is an alternative to log in on our platform in case of traveled BankID.

Another question here is about the ongoing cloud migration. So what does it mean that 41% is in the cloud. So what kind of functionality is in the cloud and what's not? And then how do you prioritize? And then finally, are you getting to 100% in the code will something stay on-prem?

Our long-term goal is for sure to make sure that everything is in the cloud. Those 41%, we basically have 500 or so applications that powers different parts of our offering. Of those, 41% we currently host on GCP and the other applications we still host on our own premises. And the way we have driven the cloud migration so far is to focus on enabling innovation at speed has been the primary decision-maker for what to migrate and what not to migrate. So what is holding us back when it comes to innovation pace, that's what we have replatformed and move to the cloud first.

This one is about sort of buy versus build, which you touched upon a bit Michel. So what's your criteria for deciding what to buy and what to build are there areas that you need to have in-house? And is it more of a question of cost, quality or is it a strategic or proprietary question?

I guess I can start maybe you can fill in Elias as well. But I think it's a bit to what I talked about. If it's something that's really core business for none such as the trading platform, we will often choose to build it ourselves because we can kind of tune it to our needs. But if it's, for example, a banking ledger system, that is not our core business. We would buy that. And it is a very strategic thing. There is no flow chart that exactly determines when we pick one over the other. It's a bit of a gut feeling as well as you're doing it.

I can add, but maybe just adding some flavor to what you just said, we want to have full control of all software that is differentiating from a customer perspective, so that we can be really fast at innovating and iterating at those areas. And Michel mentioned back office systems, but another example is the cloud. Instead of building infrastructure on our own, we buy that from Google so that we can focus on what we do best, which is built an amazing platform for savings and investments. And we want to have full control of all the software that is necessary in order for us to deliver that experience through our digital channels, both the web and the app.

So what would you say the lead time is for a company to start moving to the cloud today from deciding to move the cloud to actually starting to implement in terms of any hurdles like compliance or regulatory and so on. Is it -- I guess, is it months or years or faster than that?

I mean once you make the decision, that's the beauty of cloud technology, it's super fast, both start experimenting and start implementing. But I mean in order to also get full leverage from cloud technology, you need to adopt and replatform as we do your applications. So I mean just if you're just going to track how fast can we start building something in the cloud, that will be very fast. But I mean, getting the full set of innovation capabilities, scalability across all your platform, security and a lot of other aspects, then you, of course, need to make sure that you also are rebuilding and adapting your applications to be able to run on a cloud platform.

I guess, it kind of depends on what you start with. I mean, the context is we're at 41% functionality after a number of years.

So... 5 years into the journey. I mean I can add the fact that in the cloud journey, there is a bit of a hurdle that you need to get over. I think that kind of applies to any company in a regulated business, such as known it is, you need to get your compliance department up to speed on what cloud means security department, your legal department, all these different departments need to start understanding how does cloud encryption work, how do these contracts look? Can we shape them to our will in the same way we can do with a Swedish small company. All those questions and the organizational change, I think inherently takes time. And I'm really happy we're over that hump.

Great. Okay. I think we'll move on from there. Thank you, guys. So now it's time for a short break. After the break, we go into Part 2, are we going to talk about product development, data analytics and AI. So stay tuned, and we'll see you back here at 3:10. Okay. It's time for another Q&A session. So we've got a bunch of written questions sent to us, so we'll try to get through a couple of them now. And then, of course, we have a longer Q&A session at the end. So keep sending questions. And if you're tuning in on Zoom, I want to ask a question live at the end, just like the raise hand button. You can do that already now. So you are in the queue. Okay. First question.

This is about dependence on Google. I got a bunch of variations on this question. So clearly, it's something that's important. So it's being beholden to Google's cloud infrastructure, a long-term cost risk view, how easy is it to switch cloud providers if cost becomes untenable?

Yes, I can give it a try. It's a very good question. I mean any form of meaningful partnership will mean some kind of winner lock-in that's unavoidable. And our partnership with Google is no exception from that. The way we have sort of make sure that we are too affected by price adjustments for Google is that we make really long-term agreements. Our current agreement ends by the year of -- in the end of 2027. So that's one way. And we know that we are sort of a poster boy for Google in the financial ecosystems in the Nordics, and we're not that large a player in total. So I know that for Google, it's important that we continue to be successful. And that also is good from a negotiation standpoint because there's other benefits than just the bills we pay for the use of the cloud services for Google.

So what's the main driver of product ideas? Do you start from a business need or customer demand? Or do you take ideas from competitors.

Yes, I can start. So I mean, first of all, we, of course, have a price strategy that we want to go towards and that aligns with the overall company strategy. So all the teams have kind of a set road map an idea of how they want to build up on their customer journey. But like I said, being an agile and flexible company, it's important to be able to take in here is a competitor doing this or customers are now saying that or here's a new regulation. So it can basically come from anywhere. But I think we're set up in a way that we can really tackle all of these different inflows of ideas. But it can come from the ones you mentioned, but of course, regulation as well.

In which area is see the greatest potential to utilize AI? Is it mainly a tool for cutting or voting cost or can actually drive growth? And are there any areas where it wouldn't make a difference or where you don't want to use AI for compliance or regulatory reasons. If you want to start material?

Of course, compliance and risks are very like you talked about when it comes to generate AI. So I think when it comes to cost reduction and those kind of tools, it's better to buy those kind of products from companies actually develop them and test them so we can get good products to use out of the box. But when it comes to growth products like the feed example that we showed, then it's very important to build our own solutions for doing that, like fitting the technology to our problems that we solve. So I think when it comes to Nordnet and the things we build, it will be more on growth side and then when we buy and use proven technologies, it will be on cost reduction maybe.

One about the app versus the web. So does the mobile app exactly mirror the web? Or are there any differences in terms of functionality or any other differences.

No, not at all. And that's not the strategy either. I mean we see a different use case, using the app. It's more on the go, you trade value-wise, a little bit less, and it's a different type of customer segment who's still using the web and they have a different type of need, a little bit more data heavy, a little bit more advanced, for example. So no, not at all.

This one is about AB testing. So how do you decide what and when AB test do you do it for every product? And does that not slow down the release rate?

No. And I would say we do it when it makes sense. If it's a critical product that we haven't even launched, then we will launch it and then improve from there with iterative process once again. It needs to be enough traffic in the flow to also make sense to actually get some kind of answer. So we want to learn. It can be a technical learning or of course, how we can improve further. But that's how we decide upon if we want to do it, test or not.

I would like to add on that. And I mean I think we did AB test 10 years ago, but it was so cumbersome to do that. But today, it's like a few extra minutes or hours to add at Adest. So that's why we can do it much more frequently today. And we have 10, 20 Sous running as we speak.

Exactly. And that is the unified platform for analytics and experimentation. As you said, this is the third AB, I think we are right now. But it took time. You did the nav test, you were running it and then you were waiting for insights, and that is not how you get it to really be a part of the culture because you want to be there, you want to feel it and see the change and then it's fun, and you will continue to do it. And that's where we are right now.

So this one is about Shareville. So what do you think is the main value of Shareville. Is it a retention tool or customer recruitment? Or does it boost activity mainly?

I think activity mainly and really catering for -- that we want to be the most inspiring savings and investment platform. You can really use the social part as engagement everywhere. I mean we have 400,000 customers helping us out with content and inspiration that others can use to take better and more informative investment decisions.

Thank you. Okay. It's time to move on. So we have one final section left before we go to a longer Q&A and this section is presented by Rasmus. [Operator Instructions] So maybe I'll start off with a question that was sent in and then hopefully, we'll get some questions from the room.

So this one is about products and flagship products. So after the launch of the Danish Levantension. -- end of this year, you'll have delivered on all your flagship product launches. Are there any other major capabilities or products after this? And if not, what do you use those resources for they're no longer needed for major products?

I mean we have a super exciting road map ahead of us, and I think Levante is the next big thing. And there are other things that we haven't disclosed yet. And of course, like Mattias was talking about in the last Q&A, we see and find innovation everywhere. So there will be other fantastic things to do. One that we have announced is we are looking at whether we should release a mortgage for our private banking or higher worth customers in Finland. But there are so many things to do now. We're about to close down the old Several platform where we've integrated all of the social functionality in the main Athene web. And like Mattias Törnqvist was showing, I think we're just at the crawling stage of what we can do with that data and how we can use jen AI to create fantastic customer experiences.

So let's see Nicolas Vaysselier from BNP Paribas Exane.

I just have 2 questions, refocusing a bit on the business. The first one, we've seen in the asset management world companies like Amundi or BlackRock, commercializing their technology to some of our competitors. And it is a way for them to generate revenues and leverage or CapEx. So I wonder if this is something you have considered doing in a country where you are not competing. And if it's possible at all what would be needed? And second question is a bit coming back on your replatform in journey. Like I'm wondering how difficult or not it would be for some of the incumbents or other online players to go through the same journey you've been.

Sure. I mean I think we get a lot of inbounds regarding that. I think it's a very different proposition. Of course, if we were to become a SaaS company, delivering software as a service to a third-party. I'm sure you've noticed today, we are extremely proud of the platform we've built. We're not about to slap a white label on it and ship it to somebody else. I think that is a respectable business, but it's not our business.

I can give a shot. I mean if you start -- look at the bigger players to be the big banks, I mean, we know that they've been for many years, trying to get away from their mainframe systems, and that has been challenging. I would expect it will be even harder for them to do a full cloud migration in the way that we do it. That being said, I also know that most of the big banks are starting to leverage cloud technology just as we do. But I think to get the really the maximum benefit you want to build a cloud-native platform where everything is in the cloud. And that would be very challenging is my guess for the big banks. Then we have the new players, some of them are born in the cloud, especially the new challenges that have emerged over just the last years. But then sort of the first generation of fintech, we have a in Sweden, they come from a similar situation as we do where they started building a modern digital platform on their own premises. And last time I heard about it, Avanza hadn't really started their cloud journey yet, but I'm sure they will.

I can just add, Marcus, because having been an incumbent bank for 10 years and around digital for the last 4, it's exactly what the leases. I mean, how many failed big bank migration projects have we heard about how many quarterly earnings calls have we set through were eventually they hire the tech guy away and then us disappears. They have been unable to get out of the mainframe. I mean I ran digital, I had 4 release windows a year. So imagine versus 1.6 hours having once a quarter that is real release anxiety, if you don't get your feature out through that window and forget about rolling forward drilling back or AB testing. So they need to call before they can run as well and they're nowhere near running.

Yes. And like you said in your presentation, you need the UX and the product. And of course, BBank said the product but don't have the EWAC in the tech and the smaller start-ups, they might have a nice stack if they start from cloud, but they don't have that with -- they don't have that one-stop shop. Okay. I think we have a question from the room.

It's Enrico Bolzoni from JPMorgan. Just going back on the cloud, a couple of questions since it's so prominent. So the first one, it looks like you were on a journey in 2019 and a couple of years into that, you realize that things were not going the way you want it and you realize or we need to move to the cloud. So I was just wondering what has been -- when has been the epiphany of the realization or we actually need to go to the cloud? And why it hasn't happened earlier in your journey. So this is my first question. And then going back again partially to a question that has already been asked. But clearly, adoption of the cloud is becoming more commoditized. You have a lot of players in the financial industry than one way or the others are partnering with others. It took you just a few years to basically start from scratch, as you say, build the platform from scratch and be tremendously successful as you showed us. So I want to go back again to that question, what is preventing the incumbents to do the same? Is it the lack of talent? So if they don't have actually the developer to create the platform from scratch? Or if not, what is the reason why so far they have not succeeded. And then finally, just a question on cost. You say that you will have -- you aim to have a full migration to the cloud at some point over the next few years. Once that happens, is there a substantial amount of cost that would be able to completely switch off beatnight because you will be completely on the cloud?

All right. So I'm going to start backwards, I guess. On this one, I try to remember all the questions. A lot of great questions. Cost-wise, we pay quite a lot to operate our on-prem business. And as I said, we still have 60% of our applications are hosted on our own premises. And our goal is that by 2030, our data centers will be a thing of the Nordnet history. So after that, we will see a lot more or a lot less or none to be specific investments in new infrastructure. And we also have a lot of license costs related to running our own data centers. So there will be a clear decrease in cost once we turn off our traditional data centers. That being said, of course, as we move over more and more applications to Google Cloud, we will also see increased builds from Google. But I mean the net total, if we would take all our 60% remaining applications and move them over to the cloud, that would be a cost savings that I'm pretty confident about without giving specific numbers. And then let me see, what was the other end again. Yes, the incumbents again, it will be a similar answer to what I just said. I think the main reason why it is hard for them is, first of all, they have the mainframe systems. They are not built to be run on the cloud platform. So in order for them to do a full cloud migration as we do it, they first of all, need to get out of the mainframes. And then I also expect and this probably Rasmus has more intel because I don't -- I have never worked at a big bank, but I would expect that the culture and the sort of governance structure within those companies could probably prevent usage of very from technologies such as cloud technology. And I'm not also sure whether -- I mean, you know this and you support our cloud migration, but you also know that we need to spend a lot of time, both money and time of capacity that we could otherwise have spent on maybe immediately delivering new products. And I'm not sure that the big banks have the right structures to really make the investments and put in the effort needed to do a full cloud migration.

But a great point and it goes to culture and back to what Lars-Ake started but with your question mark is whether he was a tech CEO or a bank CEO because we are a tech company at heart, even how we organize and how we sit in our office, it's clear that the tech is at the core, whereas that's not the case at a lot of big banks, they will hide their IT departments, not even tech departments away in the suburb somewhere, and they will fax them requirement specs. So that's just a different way of doing business. I'd also say that, I mean, it is challenging. It was challenging for us to do a migration. We're not done yet on the back end, and we were taken private to launch a new app and new web and because it was easy to do out of the public eye and out of having to report every single quarter how it was going and how much it's costing. And so that is a big challenge, I would say, for the incumbent banks as well to be challenged with that. I think you need to remind us of your first question.

So the first one actually was just to a couple of years to realize that I think were not going as you wanted. So what has been the moment to make you realize we want to change? And what prevented you from having this realization.

So 2 things, put simply. I mean, what made us really realize that we need to do something -- it was this infrastructure management issue that we faced. Once we started the replatforming in 2017 and 2018, we spent most of the time building infrastructure rather than replatforming our business services. And why we didn't make the decision earlier because I think there was a lot -- there were a lot of engineers who knew that cloud technology would be a superior alternative for Nordnet but we had this preconception that as a bank, we are not allowed to operate on a public cloud platform, and that preconception was what made us wait to sort of make the decision and get the siad,well, we need to change and we need to go to the cloud. And that we did, fortunately, early 2019.

Let me just add to the previous question around -- I think you said other incumbent banks are actually looking at cloud. And that is true, and I want to play you in here, Michel. But when they do cloud, they put it in a separate like a sandbox or it's some sort of x environment where anything goes. They're not running their main business on that platform. They're playing around. And I think when you met with Google, that was the feedback you got, Michel.

Yes. And I mean our strategy when we went into it was we said we want to have the most critical data from the start. If the cloud platform goes down, the site goes down. That was kind of the outlet we had when we did the cloud migration. So we kind of got that out of the way really, really quickly. And like you're talking about like you're saying is a lot of the bigger players, they move some subsegment that is kind of irrelevant, which doesn't really force you to tackle those difficult compliance questions about data, about resilience, risk and so forth. We really front-loaded us there, and it was quite expensive for it and it was frustrating. But I think -- I mean, right now, looking back, it's just.

Alex Medhurst from Barclays.

Firstly, how do we ensure that the tech spend sort of lands with customers? I think clearly here, we've been picking up the point but I think some people struggle for example, where what is a very positive method we will get on the Pet side with maybe the performance in Sweden, for example, where the customer growth and NPS are below per that maybe doesn't give such a positive message. And the second question, how do you think about finding the right level of engagement for customers for a customer. So clearly, balancing increasing the revenue per customer engagement, retention, et cetera, but trying not to tip over that level where you achieve adverse outcomes from overtrading and those sorts of things.

Okay. The first one, maybe basically we do the customers care about the technological advancements we've done an example of how Sweden might not be growing as much as our main competitor who wants to have a Mattias, maybe on the customer.

Yes, the second one, regarding engagement, I think, I mean, we're not there at all when it comes to gamification. You can see what happened or engagement and gamification -- what happened in the U.S. with Robin Hold trying to like gamify options trading, that's nothing we would ever do, but we want to engage within the more noncomplex products such as funds and savings account and buying your first stock that there we really want to be engaging. But I mean we have MiFID II in Europe as well where it kind of prohibits us to even go that direction. So I have no concern at all. I mean engagement in itself, it's not bad, it's about teaching and growing with the customer and educating them, right?

Yes. I think to Mattias' point, we take almost the opposite view, right? So when we had the GameStop AMC craze a couple of years ago and it's starting up now, again, Gamestop, right. We actually went out, we had our savings economist write blog post about the dangers of buying these instruments that are quickly and volatilely changing their value. We had warning banners on all the instrument pages, and we also sent a message logged-in mode message to everybody who had owned or was looking at those instruments. So that's not where we want to make our money. And we want to build, like Mattias is saying, a long-term engagement, and this is what's beautiful with this business model. We win when our customers win. We're not a casino where we win, if they lose, it's a win-win. So we want them to do well. We want them to stay a customer. We have fantastic retention rate, 98% we wanted to be on the platform for a long while to grow their wealth level, and that's good for us and it's good for them. So we're very balanced on that. And I think I'm on the record around gamification, where it helps to encourage patly good behavior, like what he is saying, setting up a savings plan, diversifying your portfolio going through an appropriateness test that we are educating you and checking that you have the appropriate knowledge level to buy an instrument. That is all good things. There we can use gamification like devices. We don't want to use that to encourage options trading or buying sort of a 20x levered ETP on crypto underlying, definitely not. So I think -- and that's not who we are. So I think we have a very strong moral compass at Nordnet and what we want to promote and more we don't want to promote. The first question, if I heard you, it was a bit of a bad line there, Alex. The first question was, do customers care about the technology. It was kind of what I heard. Let me answer that and you tell me if that was the wrong question.

Maybe just to clarify sorry to jump in back and that's maybe a slight oversimplification. -- question. Now the question was more how do you ensure that the investments you're making in tech actually land with the customers and then get reflected in some of those KPIs that drive the financials, so customer growth and higher MBS stores and those sorts of things.

Okay. Got you. That's a much better question, Alex. So let me hand it over to Mattias because to us, it's about small incremental releases and it's about measuring how customers use them. Do you want to expand on that?

Yes. I mean, I can start. I mean, everything that we build enabling wise is based on either a requirement that we had in the past. So kind of when we have migrated everything to the cloud and where we invest in tech, that's either a very needed requirement that we've had to run this platform or it's a future requirement. For example, the pension -- new pension account that we launched in Finland that kind of prioritized up some of the technical enablements that we maybe would not prioritize if it wasn't for the Finnish wrapper that we launched last year.

And I think one good comparison is probably Nordnet before the replatforming. An example of where we can see the cost is actually really appreciate it. We have uptime now. We have the basic parts that the customers enjoy. And if you compare to where we were before the replatforming, we have had tremendous customer growth and to minus customer product holding value. So clearly, the customers I do appreciate it.

Yes. But simplifying it back to your question there, Alex, I think we stay very close to customers. We have our ear to the ground. We're very close to customers in all of our social media channels, including Shareville. So we get a lot of feedback. You can't build a feature for the squeeqees wheel. That's the wrong way to go, so we try to validate the customer needs through focus groups or through surveys or through other applications of data where we say, well, this is a feature we're missing. And to be honest, we also have expert overlay, I mean we all know how to build the platform, and we can make those decisions. But then we don't rely on that. So we rely on as Cecilia's data in the output. So we may get it wrong and that's okay. Well, we will AB-test that and we will go with the AB test if that's better. And then will ABT test that test and we'll keep iterating. And I think we've done several things wrong. We worked the order screen a couple of years back on the app and that converted worse. We cannot have an order screen that converts worse. That's our main business, right? So we took it back. We talked to customers and then we ship the new AB test and that converted better. So we ran with that. And we keep doing this. So we're validating it, and we're not sort of betting the bank unintended on any single feature. We're getting it out in the hands of customers and they're measuring what that does to conversion to onboarding to what have you. I think that's the right way to go.

Yes. And also what we did back then was to really spend like 6 months before we actually get it out and started testing and then it was not popular to show this negative results. And there was a lot of discussions, but now we are much faster and really trying the first version and improving from that. So we do it together, and we don't have like killer darlings. We don't end up in that situation anymore. We are much faster and much better.

I think -- I mean, certain technical investments are always going to be like a muscle higher cal needs things. I mean if the website isn't working, nothing really matters, right? So -- but those things we know we need to invest in to make sure they happened. The fact that we have good uptime now, it's not an accident, it's not luck. We made that luck and we need to keep making it.

This is Emil Johnson from DNB.

I'm wondering -- so imagine that I'm a software developer at Nordnet. And I'm interested in developing and deploying a new feature for the website, let's say. Now if I did not have the advantages of doing this in a Google Cloud environment, what are the -- could you give us some concrete examples of the sort of day-to-day things that I would need to do like the additional things that I would need to develop, the additional things that I would need to spend extra time on as a result of not having the benefits of Google Cloud in order to get that feature deployed.

Yes. I mean just a concrete example is, for example, testing environment, so you're developing a front-end feature. -- on-premise, we were limited to what we used to call cages, which were boxes of test environments. So we would have, however, many cages 14 15, and they were kind of bound by hardware because they were expensive to keep. So when you were developing something for the front end, you would take a ticket, you would get a cage, you would deploy your change to the cage, you would make sure it looks and feels right, and then the release cycle from that case was, of course, quite long. Whereas on GCP, we have an unlimited amount of cages. So now there's no ticketing system. So all the teams can have however many cages they like, and we've designed the deployment pipelines to also commit it for that. So that's like a concrete example of how you can really enable flow through the dynamic scalability. I don't know if that was.

I can add on. I mean, repeating what we talked about earlier, I mean, the experimentation cycles on a cloud platform, they are just insanely short compared to the experimentation cycles on an on-premise environment. So I mean, we explore new infrastructure and new technology on a daily basis basically. And when you can do that in a matter of minutes and hours, you speed up innovation, it's just in same compared to when, well, are we going to invest 2 weeks or a month to build this infrastructure and explore it. More often than not, you don't even try to do it because you're not certain whether that investment will be worth it or not. But when you can experiment in minutes, you experiment a lot more, which really helps up speed up innovation in all forms of ways.

I can add on that when it comes to AI and machine learning, which is heavily dependent on like BP machines and like a lot of stuff that maybe a company like Nordnet don't have the size to have much remaining infra, but we don't need that on the cloud. We can just experiment right off the box. So those cases could not even be possible on a on-premise solution. So I'm happy that the loan.

Great. Let's switch it up and do a written question. Somebody sent in a question about AI. So we'll go back to you, Mattias.

Is AI available in any form for customers to use for investment decisions, portfolio allocation, et cetera, -- and if not, will it be in the future, you think?

Good question. So no, we don't do financial advice through LLM today. And I don't see that we will do in a short term either, but I'm, of course, very hopeful that everyone can get good advice from these kind of models and technologies in the future. And we've seen a lot of like scary examples of companies promising stuff, and we would absolutely not want to risk that. So that might be a case where like it's fun for us to experiment and see how good advice is, but we might wait until we can purchase a service that is more tested and like built for the world. So trying to be like a lagging leader when it comes to global financial advice. When it comes to LLM, I think it's the same approach to as...

Yes, I agree. I mean the LLM are fantastic, and I was up late last time playing with GT40, -- but the issue is that they're still fabrillating, right? And the last thing you want as a regulated bank when the MiFID II license is an LLM fabrilating and certainly recommending things or advising things that are either not true or not appropriate to that customer's scale or experience level. So we're keeping everything one step behind, like Mattias is saying until we let them out into the wild. But I think the potential is there for sure. I mean, it's early days when it comes to these things and how you can control them and how you can improve them. So I think we'll see that, but we're not near that yet.

Yes. But also going back to what you talked about the value where it is, and it's still in machine learning and what we talk about is, for example, could we create a model that helps our customers to diversify their portfolios. And we won't say that they should invest in something, and it's going to be hard to convince a person who have invested in GameStop another to buy like banks or forest, but it might be a case that we could use and try around and find a balance where this actually can work, at least we are once again trying to enable higher returns for our customers using machine learning, and that could be an option to look into further.

Good afternoon, everyone. So Jacob from SEB here. At the end of the day, I mean, I'm a numbers guy and the more we can connect it into Excel, the better for us, at least. So the first question is basically, if you haven't -- or you didn't do the tech overhaul that you did back in 2019, how do you think your profitability would look like today? And then to better understand maybe your cost side, so using the cloud, do you pay 24-hour upkeep or you only pay Google when you actually use the different apps that you have on it?

Second one, we only pay for usage. So that's the beauty of cloud as Michel was referring to, if we need to scale up production capacity, we can do that and then we pay for it during that period. But then if we scale down again, then we don't have to pay for that capacity that we have removed. So it's pay-as-you-go.

Yes. I can comment on the first. I guess it will be hard to give you numbers there, Jacob. But we would have not been able to capitalize and some people say that we are a pandemic winner, and it's true. We are a pandemic winner, but we did the hard work in order to capitalize on that. If we had not done these cloud investments and not done the replatforming both on the back end but also app and web, this is no way we would have scaled to see the kind of print that we did those days. We had 1 day, I think, 60,000-plus trades in a day. We would have crashed such a long time ago, but you want to add on that.

No, you was saying exactly what I was intending to say. I mean, if we wouldn't have done those investments and this replatforming, we have more than twice the amount of customers now compared to in 2019 and the activities on the same level. Our platform wouldn't have managed it -- so -- and I mean you can imagine a savings and investment platform that isn't available, then you don't have that much of a business anymore.

Yes. Maybe we wouldn't even have been listed at that point. So it's a lot of what it's in that question, but.

And just a last question on generative AI. Where do you think we will see the first user cases? Are you going to first roll it out for your own usage on customer support side, et cetera? Or do you think in the not too far future, we're actually going to have it on a user interface as well where, for example, a user could rightly screen Nordnet stocks, which have 10%-plus EPS growth over the last 3 years and is projected to have it in the next 2 years, but is trading at a discount compared to the last 3 years or something?

Good question. So I can start with the generative AI part. Like you can copy all the information from like a stock page on our platform, just pace it into satiety and take whatever risk you want. But we don't want the risk of giving that to you with all that it comes. So now we're developing stuff for ourselves to get a feeling and understand and also like to see how we can work with our governance and compliance framework that we have. And I don't see -- we're not really building anything now now because of the status of this. And I don't think many of our competitors does that either maybe can up. They are a bit bigger than us when it comes to engineers and other stuff.

But the one thing you have in production, Mattias, which is super exciting, is this -- that you mentioned briefly the summary of your trading day, sentiment analysis, and that will be generative AI. So we're taking it to where it can be descriptive instead of prescriptive, if you will. So where we can say, your portfolio developed this way due to XYZ, which is basically a combination of an MO model and Gen AI putting that into sort of natural language in a way that is attractive and understandable by a retail user. But we did -- we ran a proof of concept during Q4. I think it was 2 developers and a data scientist, and they built an entire copilot for our Swedish customer service in that time. And when issues like vectorization come into play, you can't just take all the internal data and dump it into the context window. There's not enough the window isn't big enough. So you need to find a way of accurately sort of mapping, vectorizing if you will, the data pick up what you think may involve the answer to the question and then shift that into LM. And so we did that on using Google's tools, which meant that all the data is encrypted both and at rest within our -- in our domain, so we can layer Google's data tools and then LLM on top within our environment. And that was cool. And I think Bain has published a study where they gave their consultants 50 minutes into Chat GPT and then they saw a 17% uplift. But that's not interesting. What's interesting is when you bifurcated their consultants into new hires and senior and into lowly rated annual reviews and higher rated. So the higher rated and senior people didn't see much of an uplift at all on productivity. But the junior people, new hires and people with lower performance ratings, they saw a fantastic 40% plus increase. So when it comes to customer service, right, we have a lot of young people, we have a lot of students, there's a high staff turnover. So bringing those people on, giving them an in-house copilot that will bring them up the curve and be that much more efficient at their job early and productive in how they can help customers. But there -- again, there's a human. There's a CS agent between the LLM and the customer earlier the line. I think that's where we need to keep it for now until we just these things more.

And I mean one other thing that we built internally is the -- all the customer feedback. So we collect all customer feedback. And then for product people, part designers, developers, they can go in and ask this generative AI. What are customers saying about our app? What are our private banking customers saying about our mortgages, for example. And then it collects all that and gives you an answer of all the.

Rickard Strand from Nordea. So going back to Slide 12 in your slide deck, you showed an illustration of the time you spent on your sort of your legacy or technically that. Yes, exactly. Is it possible to give an illustration of what that looked 5 years ago and what do you foresee that being 5 years from now?

That's an excellent question. And unfortunately, we would love to have hard data on technical debt. But I mean the nature of tech that it is super hard to measure. Everyone who has worked hands-on in an engineering team knows tech-debt when you see it. And it really drains energy, and it adds a lot of lead time. And I'm not exaggerating when I'm saying that it can -- you can easily add 10 times or even 100x of work when tech debt high. But the problem is that we can't measure it in a good way. The way it is defined is if you have -- let's say, we are going to do a change, and we have a code with certain degrees of technical debt -- and let's say, it takes a month for us to implement that change in the current state of Victa. Tech debt is defined as if we remove all the necessary complexity, all the complexity that don't need to be in the code, but we still have the exact same offering, how long would it then take to implement the same thing. And when you do that, you might be able to implement the same change that took a month when Pete was high in just a day. But the problem is that we never get the chance to measure the difference because it would be ridiculous to implement the same thing twice. And every software delivery project is new. There's new things we are going to do and new parts of the code base. So it is very hard to quantify. We know it's there, but we can't measure it in a good way.

Giving you, I guess, a hint could be some of the outputs you showed Elias, both in the benchmarking and doing and the number of deploys. So those are 2 things that you can look at. And if there are -- those are higher or late, then you probably don't have a lot of pictures.

So we look at the outputs instead. So what are our productivity? What are the lead times for changes as we were referring to? And how often do we introduce failures because the source code was complex that we didn't understand it. So that's the numbers we look at, and we can't really measure the actual tech that that we know is there.

But a practical and recent example, I mean, Mattias Shevel used to be a separate platform, separate code base, and we lot teams that build that. We're no longer here. People were afraid to touch it. We couldn't upgrade it and.

Yes. And I mean, hopefully, in just a few weeks, we're actually going to kill the entire old shared platform, which means that we get rid of a lot of legacy and enable a lot more speed for that team because they don't no longer have to maintain that old platform. They can just focus on the new.

Last thing you want is a product whenever you come with an idea or an improvement or a bug reporters have the entire technical I don't want to look at that. And we don't get that a lot anymore. They go, yes, fine. When you need this Friday, to work for you. So that's where we are at. And that's another sort of totally subjective measure of tech debt.

Okay. Great. I think we'll wrap it up there. So thank you so much for participating. Thank you so much to the live audience, and thank you so much for our digital audience for participating today. We really hope that that's been lining in for you and that you've learned about our tech. If you have follow-up questions, feel free to reach out to me. Okay. Have a great rest of your day.