Rise of AI Conference 2026 · Berlin · Meta Stage

Scaling Sovereign AI in Banking

Julia Koch · Finanz Informatik

At Rise of AI 2026 in Berlin, Julia Koch, managing director at Finanz Informatik, the central IT partner of the German Savings Bank Group (Sparkasse), explains how her organization is scaling sovereign AI from assistance tools to agentic systems across more than 350 independent legal entities and about 200,000 employees. Finanz Informatik runs its own data centers, built before ChatGPT existed, and operates an on-premise AI platform using open-source models where possible to avoid vendor lock-in, developed with engineering support from Nvidia. Its flagship tool, the SKI-Pilot (SAI-Pilot), is used by 200,000 employees, generated 2.8 million prompts in a recent ten-day period, and its meeting-preparation agent cut advisor prep time from 30-45 minutes to seconds, with over 175,000 meeting preparations completed in April. Koch describes deep integration with core banking systems and Outlook as key adoption drivers, and previews a widget-based, role-specific dashboard planned for deployment in June to support further agentic capabilities in a highly regulated environment.

Key takeaways

  • Finanz Informatik built its own data centers over 10-20 years, letting it run AI in-house years before ChatGPT existed, which Julia Koch frames as the basis of its sovereign AI approach.
  • The SKI-Pilot agentic assistant is used by 200,000 employees and generated 2.8 million prompts in a recent ten-day period.
  • The meeting-preparation agent cut advisor prep time from 30-45 minutes to seconds and supported over 175,000 meeting preparations in April.
  • Finanz Informatik avoids vendor lock-in by switching between open-source and other language models depending on the use case, developed with Nvidia's engineering support.
  • A new widget-based, role-specific employee dashboard is planned for deployment in June as the foundation for broader agentic capabilities.

Detailed notes

Finanz Informatik's role and scale

Julia Koch introduces Finanz Informatik as the central IT partner behind the German Savings Bank Group, serving more than 15 million common-foundation customers, over 35 million online banking customers and more than 20 million users of the Sparkasse app, who together generate more than 22 million digital banking interactions per month, up from about 7 million per month when she joined in January 2022. She frames trust as the most valuable asset of banks and stresses the responsibility that comes with deploying AI across more than 350 independent legal entities, each subject to ECB and local regulatory requirements, comparing the scale to running '350 times Deutsche Bank.'

Why sovereignty rests on owning the data center

Koch says Finanz Informatik has operated its own data centers for 10 to 20 years and is one of the largest data center operators in Germany, which let it work with AI years before ChatGPT existed. This in-house infrastructure lets the platform connect directly to bank data, processes and institutional knowledge, including partners like Deka and LBS. The platform runs on open-source models where possible, switches between small and large language models depending on the use case, and avoids vendor lock-in. Koch credits Nvidia's engineering support for helping the team solve enterprise-scale problems specific to operating across roughly 400 legal entities, something neither side had faced before.

From assistance to agentic AI: the SKI-Pilot

Finanz Informatik's agentic tool, the SKI-Pilot, integrates directly into employees' Outlook environment, which Koch calls the adoption 'icebreaker.' It started with text generation and summarization and has since added agents that prepare emails, advisory meetings, requirement processes, financial statements and controls. In the ten days before the talk it logged 2.8 million prompts. Its meeting-preparation agent, given a customer number, pulls together customer data, transaction analysis and advisory suggestions in seconds, replacing a manual process that previously took advisors 30-45 minutes; in April alone it supported more than 175,000 meeting preparations. Koch attributes the tool's effectiveness to deep integration with the Sparkasse core banking systems, via a routing layer she refers to with the MCP icon shown on her architecture slide, connecting agents directly to bank processes.

Outlook, organizational change and closing principles

Koch previews a next-generation, widget-based dashboard tailored to each employee's role and regulatory needs, due for deployment in June, which she says is the foundation for deeper agentic capabilities across sales and service processes in a highly regulated environment. She frames the technology shift as equally an organizational one: business, design and technology experts must work in shared 'collaboration rooms' to find real use cases, many of which are discarded when they do not prove helpful. She notes Germany's retiring baby-boomer workforce makes AI-driven efficiency essential since replacement hires will not be available in sufficient numbers. She closes by summarizing sovereign AI at Finanz Informatik as strategic principles built on their own AI platform: customer-centric, compliant, effective and explainable to supervisors.

Quotes

Our own data centers are our prerequisite to talk about everything in a secured and flexible environment.

Julia Koch

Sovereign AI means for us really the strategic principles on our AI platform. That is our common and broad foundation for everything.

Julia Koch

Questions answered in this talk

How is Finanz Informatik scaling AI across the German Sparkasse group?

According to Julia Koch at Rise of AI 2026, Finanz Informatik runs a sovereign AI platform on its own data centers, serving more than 350 legal entities and 200,000 employees, and has moved from text-generation assistance to agentic tools embedded in core banking and Outlook workflows.

How much has the SKI-Pilot AI assistant reduced meeting preparation time for bank advisors?

According to Julia Koch at Rise of AI 2026, the SKI-Pilot's meeting-preparation agent reduced advisor preparation time from 30-45 minutes to a few seconds, and supported over 175,000 meeting preparations in April alone.

Why does Finanz Informatik run AI on its own data centers instead of the public cloud?

According to Julia Koch at Rise of AI 2026, owning the data centers lets Finanz Informatik connect AI directly to bank data and processes securely, avoid vendor lock-in by switching between models, and meet strict ECB and local regulatory requirements across hundreds of legal entities.

Mentioned

sovereign AI, agentic AI, SKI-Pilot, MCP, vendor lock-in, open-source LLMs, core banking integration

Keep exploring

More from Rise of AI Conference 2026

More on Enterprise AI & Adoption

Transcript

Auto-generated transcript, not edited. · 3,828 words

Show the full transcript

Ladies and gentlemen, welcome one and all and most importantly, congratulations on 10 years of visionary perspective, 10 years of dedicated commitment to the rise of AI in Europe. When you had the first conference back in 2016, you already knew AI has the potential to change the world. Which it has long since started to do. For example, in medicine where it is revolutionizing diagnostics. It is opening up new pathways to new treatments. It often still takes a marathon run for a drug to be licensed. AI can turn that run into a sprint. In biochemistry, AI is solving problems that were stalled for decades. It is making materials research easier. Micro electronics, too. Think chip design in this context, which in turn benefits AI itself.

We all could provide a long list of examples and yet we are only just at the beginning of the rise of AI. There is still so much that we can shape and we must do so with firm resolve. For our sovereignty and because there's a huge market behind it. For me and the federal government, it boils down to this. To turn research strength into economic strength, to turn scientific breakthroughs into real products and services. This is why AI as a key technology is at the top of our high-tech agenda Germany. The agenda is providing a turbo boost for more transfer. Firstly, by developing computing capacity. The the super computer is already Europe's Europe's fast fastest and the most energy efficient in its class.

But but we are aiming for more of course. We want to establish at least one >> >> Okay, either you get the gist of it from where you can see the video message from your running machine and I switch to English. I forgot about lunch that we talked in English here right now. So, welcome back. I hope you had good discussions during during the lunch break. I certainly had. And I'm very thankful that you returned to us right now. So, we have an interesting afternoon before us and we want to kick it off if I may say with the managing director at Finanz Informatik, the digitalization partner of the German Savings Bank Finance Group as we know it as Sparkasse.

Yeah, she brings very practical perspective on how to deploy AI securely. Please welcome Julia Koch. >> >> So. Hello and good afternoon. It's a great pleasure to having me here and thanks for the translation. Sparkasse. So, that is more than less the background of that what I would like to explain to you in the next few minutes where we support and provide technology to the Sparkasse. We as a Finanz Informatik are the central IT partner and provider behind the German Savings Bank Groups. That means at least something happens. Oops. Yeah, great. That means at least that we support our savings banks where our savings banks are. Independent, if you are talking about the app Sparkasse or the branches or in call centers, we are there where our customers are.

And just to be transparent, we are talking about more than 15 million customers. That is our common foundation where we are and where we support with every kind of technology that you can imagine. That means for us, on the one hand, a huge responsibility for that what we are doing, because trust is the most value of banks. If we destroy the trust to our organization, to our banks, we would like to see a lot of trouble. Dedicated press news in the Morgenmagazin or in the Tagesschau, honestly, these kind of shitstorms are not our aim. Our aim is to provide our customers. We are talking about more than 35 million online banking customers and more than 20 million in the app Sparkasse.

20 million customers in the app Sparkasse are our foundation for the future. This is very important for all of us, because these both, online banking customers and app Sparkasse users, are the that one who use it 20 more than 20 t- two million times per months in the digital way for services, for sales and something like this. I started at Finanzinformatik more than less 4 years ago in January 20 22. There were nothing more than 7 million digital banking calls per months. Today, more than 22 per months. That means that our customer behavior changed significantly in the last 3 4 years. And that is the foundation for everything what we are talking about. Especially if we talking about AI in a sovereign and scalable environment.

And there are a lot of potential use cases for us. Just to setting the scene. We are talking about 50 million customers, more than 350 independent legal entities. That means in a simple word, 350 times Deutsche Bank. 350 times all the regulatory that the ECB and the local regulators expect from us. There are some dedicated questions how we can scale. For sure, we support and provide all channels independent if you are talking about branches, apps, Sparkasse, and so on. And for sure, we support and provide solutions for all banking products. Independent if you are talking about commercial banking or trade finance. And for sure, also the trade link platform for huge Sparkassen. Like a Hamburger Sparkasse, for example.

Or as we here in Berlin for a Berliner Sparkasse. That is the main challenge for us if you are talking about scaling AI in a sovereign environment. We have some dedicated and different questions to follow just to ensure that we really can be sovereign. If you are looking to a fintech or a startup, they are high privileged. They have in mostly one target group, one channel, one goal, one product. I'm so happy about it that I really can care about one topic and not about the whole world business because they could be faster than we. But and they help us to ensure to leverage. And so that is something where we can learn together how we can scale the next part.

Behind all these German saving bank groups are more than less 200,000 employees in Germany. 200,000. And you can imagine if we transform something with AI in their daily business, then we will see a lot of micro transformations and a huge change in the transformation of the workplace of the employees. This is very, very important for us because you know it in Germany, we saw we see an ongoing retirement process of baby boomers. That means we will lose a lot of these colleagues in the next few years. And honestly, we won't find some new employees in these huge amounts. So for us is it essential that we can change and leverage on it. And for sure, if we are looking to the future and honestly, it's not in future only for us, it's still existing here and there in their daily work, is that we are really have a domain-specific solution where we can combine it with our data in our own data centers and where we can combine it with our processes.

Only then it will be sh- secured and effective. Otherwise, we have a lot of tools somewhere in the area and honestly, we cannot really support our employees. Based on this, we have also an general idea about what does it mean working with agentic AI. In the last year, we are really developed our first agents. Agents in banking. Ooh, voodoo. Honestly, it's work. It's work pretty good, and I will present it later to you. And it's not just one agent. The core challenge is how we can ensure that we have really a multi-agent support for our clients, for our customers, for our employees. And that is a little challenge to the future. If you are looking to that, we could discuss about cloud, on prem, hybrid support, and something like this.

Honestly, that is not the matter of force. We have the really good situation that we ramped up and built up over the last few 10, 20 years our own data centers. We are as Finanz Informatik one of the largest data center operators in Germany. That is our foundation for everything what we are talking and working on. That means for us, we worked with AI couple of years before ChatGPT exist. Without these foundation, it wouldn't be possible to do so. So, our own data centers are our prerequisite to talk about everything in a secured and flexible environment, and for sure also on a cost efficiency environment in Germany. Based on this, we can build and we build on already yet sovereign AI platforms and for sure sovereign AI solutions.

If we are looking to that in a detail, does it work? Yeah. Nope. Okay. In a data. Um our own AI platform we are proceed and provide in our own data center. We are coming to this slide. Make it simple. It's a high-level AI architecture. I never expect that we discuss it in all the several stages. But let's start from the top. There you can see AI use cases for saving banks and their customers. Several different applications. At Sparkasse, as mentioned before, 20 million users. 200,000 employees. And so on and so on and so on. That is our challenge that we provide a platform that can really support and leverage on it. Otherwise, it wouldn't be possible to support our users.

These front ends, these several front ends that you can see here, that is just an extract of it. Our working and based on a AI platform that we provide in our own data centers. There are several models behind. We are on premise and use open source how much as we can. And that not as a paradigma. That is our common sense about it, how we can really leverage. What we observe over the last few years is a permanent model change. There are a lot of small huge large language models with different cases. And for us it's more important now what kind of AI use case I would like to support and based on this we decided what kind of LLM would help us.

A small one, a huge one, what is how competitive it is and so on. And the general support means we can switch and move often we want. And we are not in a vendor lock. That means that we only have to use one solution independent from whom. And that is one of our biggest advantage in the meanwhile. Now we can really do so. And one of these that enable us to do so is Nvidia. Thanks Marcus for all our great support because without the support from Nvidia it wouldn't be possible to proceed so. We learn together how does it could work. Make it open and we can be transparent in this little audience. We learned a lot.

Last summer we had a call together, guys it doesn't work. Last but not least with your help and your the colleagues from US we understood together what does it mean in an enterprise environment for 400 legal entities to provide this. It's one part to think about you have one company, one legal entity and this is another enterprise discussion about 400 legal entities. You have some questions that you never heard before and honestly you had never had an idea how you could fix it. The engineering support of all of them with a lot of other providers that are not mentioned in this slide but also very helpful are the foundation and the prerequisite that we can really leverage on it.

Because, and that is the part close to the bottom. Because everything is in our own environment, we can connect it directly to the data, to the processes, the knowledge of the banks, the knowledge of the saving bank group, something like Deka, LBS, however you like it, and ensure that we have more than purely knowledge in it. So, and ensure that we really can connect it to the processes. And there it's it is where agents coming in. You see a little button, a little icon like MCP. This connection the routes MCP to the processes are more than less the foundation for the scaling of agentic AI in a multi-agent way in Germany. And honestly, that is not a future, that's not the outlook.

We are still working on it. And we are sure that we can leverage on it, and we can really provide it here. And that is more than less in our data center. For sure, we also have a couple of GPUs with us from Nvidia, but that it's not so much as you think about it. The most technology is there where you're not in the hardware part. The most technology and the engineering, trust, and really the idea about how it could it work are in the levels be to develop. And that is what we still important for us that we have the knowledge, the employees who can engineer it, and for sure also the foundation in our own data center to ensure that we will tell and build a sovereign story here in Germany.

That is our foundation where we are. You can imagine a lot of AI use cases still up and running. What is still missing on the slide, especially on the left-hand side, left to the conception, a lot of ideas from my crazy colleagues, team members, and so on. They have a lot of ideas that would push out this tool. So, let's start from the left-hand side. We have a lot conceptions ongoing. For sure, proof of concepts for us are mandatory in the meanwhile, just to ensure that our idea is more than an idea. And if these proof of concepts are successful, then we are going to production. You see here a little extract of the current cases.

For sure, we can discuss about everything in it the whole afternoon. Unfortunately, the time is running out, so let's focus the next few minutes to the SAI pilot, or in German means SKI pilot. He's a little bit public and a little bit famous. So, we are a little bit proud on it because he support 200,000 employees based on that what I mentioned before in this architecture. And honestly, that is competitive to the world. And that is something what is still exist and happening here in Germany and here in Europe without any general discussions about. It's still ongoing. It's work. 200,000 colleagues use it. And what does it mean in currently? The SAI pilot revolutionize their workplace.

Their everyday work. For sure, we have started a couple of years ago for visit because text generation, summarize questions, and so on. In the meanwhile, the first agents working on it. And they help to prepare emails, help to prepare advisory meetings, help to prepare requirement processes, help to prepare the financial statements, help to prepare the controls. And a lot more. Because why does it work? Because we are deep integrated in the solution of the core banking of the Sparkasse. And the one part is how many uses. The other part is how often they work it on it with it. In this month, this number [snorts] is a little bit older. Older means last 10 days ago.

2.8 million prompts are in going. Can discuss with me is it's huge, is it small? It's equal. They use it day by day by day. And for sure, one of our main features that helped us to bring it in the workplace was the connecting to the Office Neo Outlook. That means nothing more that the iPilot is deep integrated in the Outlook system of the employee. That was a little icebreaker to work with. Because it's very comfortable and easy in that he helped us to answer then to create the initial answer in the early morning. Um it's better that he work without a coffee than me. Um but that was really the icebreaker. The success story in case of agents is the meeting preparation with the AI pilot.

Let's have a little recap. Couple of months ago our advisers have to prepare a advisory session with a customer and have to check a couple of applications. More than less 30-45 minutes to invest as a pre-invest before this meeting take place. Today coming from the customer data and everything what we know about the customer the SAI pilot prepares these of German means Gesprächsvorbereitung and collect a lot of data information and develop initial idea what you can talk about with the customer. The last few weeks in April more than 175,000 175,000 meeting preparation took place. That means a direct support in the daily life of an adviser in a Sparkasse. And that is where we are start at and talking about how agents could help because only agents could collect all these informations.

And just to share a little bit brief overview on it I have a little presentation for you. Sorry, it's in German, but that depends a little bit from the language in Sparkasse. So, as you can see here it's a really simple entry. You insert the number of the customer then and the act the SAI pilot collect all the data, prepare the meeting. And a few seconds later he put all the information the board overview together starting with some information about the customer about some advices to the parameters could analyze the transactions behind and for sure can advise you in the consultant situation. These are only a few seconds. Couple of months ago you need a couple of minutes for it.

30 45 minutes. So there you can see a direct transformation in the day-by-day work of our colleagues. And for sure here are also the history of the contacts with the customer on a proper overview. And as it's running this dashboard this front end is more than less similar to that what we already have in our private use like ChatGPT, Gemini and so on. As next I would like to share with you a little outlook. What we expect in banking in the next 3 5 years. Honestly, I'm not sure if you're talking about the next 12 months at least, but let's see where we are. If we are talking about what's happened with the workplace currently there are several applications on it and for sure you have to be a little expert how you organize your day.

If you're looking to the future then we expect a new dashboard. >> [snorts] >> A new dashboard that is more than ChatGPT. A dashboard that based on widgets, based on the rules and functions of the employee where you can steer directly what is really important for you in your workplace. For sure, it help us to handle all the regulatory stuff behind. But also, our users can organize and edit their workplace as they have to use it. And that means we have an need-based entry point on it. And that is just a little outlook and that is not far away. We are still under that's still under development. We will deploy it in June. And that is also the foundation for the agentic capabilities because we need these enhancement of the front end just to ensure how we can combine all the several processes behind to in a really agentic world.

Because we expect if we have a agentic supported advisory meeting, then I expect also that all the processes for sales, for services behind will be supported by agents. And that in a highly regulated environment. That's a little bit tricky, honestly. But we have the foundation to do so. That is a technical part in it. If you are looking to the transformation in Sparkasse, in Finanzinformatik, then we see a lot of of challenges too. This transformation of technology means also a huge transformation of collaboration. Only if we are working and working in a several way with each other, it's possible to develop something like this in such a matter and such a speed last but not least.

We can scale. We scale in a highly economic way. That means for us that business guys, designer experience experts, technology experts working close together to understand how does it could work. And for sure, this collaborative corporation and collaboration with each other is the one of the key levers on it because we have to transform also the whole workflows in Sparkasse. The world change. The daily work changed repeatedly. And we have together to support our colleagues in Sparkasse, but also I have to support my team members that they can really work on it. And we can talk about technology the whole day long, honestly, but I'm a banker. I have only use cases in my mind.

That means coming from a use case is the most important part in it. It's very, very hard to define really use cases that helps the teams to scale. To scale in their day-by-day work. And so we connect all these lovely stakeholders with each other in little collaboration rooms and teams together just to ensure that we really figure out the real use case and how we can deploy it. And to be honest, we throw a lot away because it never work. It's not it wouldn't be helpful. For us, it's more important that we really figure out what help us continuously. As mentioned before, that means also a huge transformation at Finanz Informatik. We have a lot of initiatives and programs to enable our colleagues to think and learn in this world.

Because we are a traditional IT provider, that means that we have to change everything what we are doing. For sure, we need the engineers on its own, but the way how we figure out what is the solution and how we can really support our customers and our employees is another way like today. And that is a challenge what we have all together. Let me sum it up. Sovereign AI means for us really the strategic principles on our AI platform. That is our common and broad foundation for everything. With these, we can really scale sustainable. Last but not least, as mentioned before, we have a really pure customer-centric approach just to ensure that we really create something what shares value to the customer.

And as banks are already are, we have to be compliant, we have to be effective, and we have to be a clear about that what we are doing because we have to explain it to our supervisors. Thank you so much.

Rise of AI 2027 · Berlin · May 2027

Be in the room for the next one.

Ten years of talks started as conversations in one curated room in Berlin. The next edition brings 300 selected leaders together in May 2027.