Voice-over: This is Leaders in Finance, a podcast where we find out more about the people behind a successful career. We speak with the leaders of today and tomorrow to discuss their motivations, their organizations, and their personal lives. Why? Because the financial sector could use a little more honest conversation. Your host is Kees de Wit.
Kees: Welcome, listeners, to an extra episode of the Leaders in Finance podcast. My name is Kees de Wit, Managing Director at Leaders in Finance Events, and I’m currently at Pakhuis de Zwijger in Amsterdam, right after our third Leaders in Finance AI event. Today, I am joined by four individuals who not only attended the event but also played a key role in making it happen.
I want to reflect with them on today’s event, especially for listeners who did not attend, to give them an idea of what has been discussed. But first, I will briefly introduce my four guests. Menno Bonninga, Partner at EY Consulting and EY Netherlands AI Lead. Bira Thanabalasingam, Managing Director at Yellowtail. Kyriakos Fistos, AI & Analytics Advisor at SAS. And Yorick Naeff, Chief Innovation Officer at ABN AMRO.
Welcome, everyone. Let’s kick things off. Menno, would you like to add anything to your introduction? Just briefly tell us a bit about yourself.
Menno: No, absolutely, Kees. I’m Menno Bonninga indeed, and I’ve been spending quite a lot of time in financial services, about 20 years or so. On a day-to-day basis I have three responsibilities.
Firstly, driving the adoption of AI across our own organisation of around 5,000 people in the Netherlands. Secondly, building AI capabilities into the services that we bring to our clients. And thirdly, supporting all the AI-related work that we deliver to our clients in the Netherlands.
So I’m really enjoying it and I’m very close to the topic every day.
Kees: Thank you very much. Bira, maybe you could take over the microphone and tell us something about yourself as well.
Bira: I didn’t prepare it quite as well as Menno did, but I’ll do my best.
I’m Bira Thanabalasingam. What I find really interesting is not just the technology side of things, but how we can implement it in such a way that humanity still has a role to play in everything we do. Yes, we’re a company that focuses on technology and consultancy services, but how can we help our customers, and people in general, adopt and harness the power of technology without losing the human aspect?
That was also part of the talk I gave today. I found it very refreshing that the past two years have mainly focused on everything AI can do, whereas this year it was much more about stability. Within large banks, for example, you need to have certain standards in place before you can use AI. There was also much more emphasis on the human side. How do you harness AI, and how do you guide it in such a way that it benefits society as well?
So that’s a bit about me—perhaps a bit too much.
Kees: No, it’s never too much. Thank you very much. We’ll dive into your use case, which you referred to, later on as well.
Yorick, can I hand it over to you? It’s not your first time on the podcast, but could you briefly introduce yourself?
Yorick: Thank you. I’m very happy to be here because this is a topic that’s very close to my heart.
I’m Yorick Naeff. I’m the Chief Innovation Officer at ABN AMRO, but I was also one of the co-founders of BUX, the retail investment platform that we sold to ABN AMRO. Today I’m here representing ABN AMRO, and as Chief Innovation Officer, AI naturally plays a very central role in what we do.
Kees: Thank you very much. You started today with your keynote, so the first question will be for you. So hold that thought.
Kyriakos: That’s great.
Kees: Thank you very much. Kyriakos, can you tell us a bit more about yourself?
Kyriakos: Sure. My name is Kyriakos Fistos. I work for SAS, one of the leading companies in AI and analytics.
I’m part of the global Financial Services team, specifically within the AI and Data Science practice. My role is to advise our financial services clients on the development and implementation of AI and machine learning techniques. So I’m definitely very happy to be here today.
Kees: Thank you very much. We’ve talked a bit throughout the day, and I know you were very happy. Many people were happy.
So I think, for the people who are listening, you really missed something today.
Yorick, as promised, the first question is for you. I referred to your keynote, The Future of Banking: Where We Are & What’s Next. It was a very refreshing keynote. You started, for example, with an insight about the often misunderstood importance of having good data before beginning with AI.
Can you tell us a bit more about the keynote you presented today? Could you give us a brief summary?
Yorick: Well, a brief summary. I think we have about 15 minutes, and my keynote itself was around 15 minutes, so I’ll try to condense it into a very short story.
I think there were a few key messages that I wanted to convey. The first is the one you just mentioned. If we talk about AI and its implementation, what is often underestimated is the importance of data and the foundation that’s needed, especially when we’re talking about agentic AI and how we can really bring that into practice.
Data is often underestimated. Of course, all these tools, models and applications are fantastic. I’m not taking anything away from what’s already available in the market or what is currently being developed at great speed.
At the same time, you need to make sure those systems have the right context, especially if you want to use them in customer-facing situations or to support employees. Providing that context means making sure your data foundations are in order. I think that’s something that’s very much underestimated, particularly within larger organisations, because many of them have decades of legacy systems and what I would call a spaghetti infrastructure. That simply makes things much more difficult.
That’s one element.
The second element is that organisations need to make sure they’re connected to the broader ecosystem. If agentic AI really is the future—and many people believe it is—then those agents need to be able to find you and interact with you. That means creating the necessary APIs and perhaps implementing MCPs or other protocols to ensure you can play a role within that ecosystem.
Last but certainly not least, especially when it comes to banking, I think it’s worth asking what the role of a bank will actually be in the future. Will banks essentially become technology companies? Will they change the way they operate, the way they’re organised and the way they serve customers in order to become future-proof?
I think we’re standing on the edge of a very significant transformation.
Kees: I think I can agree with that. I see, Bira, that you want to respond.
Bira: Yes, I wanted to ask Yorick a question.
You’ve been at the forefront with BUX, which for years was one of the frontrunners in making investing accessible and easy. And now you’ve moved to one of the major banks, with legacy systems and, as you described it, spaghetti infrastructure.
How has that transition been for you? And what’s the biggest thing that’s actually holding ABN AMRO back, given that you’ve experienced both worlds?
Yorick: For me it’s really two things.
The first is the way the organisation itself is structured. At a smaller, founder-led company, decision-making, processes, procedures, hierarchy and communication lines are completely different.
At BUX we were able to scale very quickly, make decisions rapidly and pivot whenever necessary. That allows you to test things, experiment and fail fast. Within a large bank that’s almost impossible. It’s simply the nature of the organisation. It’s a huge oil tanker. There’s fragmentation, decentralisation, and departments don’t always communicate effectively with one another.
On top of that, leadership is divided across many domains, each with its own mandate and accountability, with relatively limited interference from senior management or the Executive Board when it comes to day-to-day decisions.
The second aspect is the technology itself. As I mentioned before, ABN AMRO is a bank with a history of more than 200 years. It’s really a patchwork of organisations that have come together over time to provide the financial services we offer today. Under the bonnet, however, it’s highly complex. All those mergers and acquisitions have resulted in many different systems operating in the background.
That’s the major difference. You first have to solve that complexity before you can truly accelerate, become more agile and deliver new applications more quickly. It simply takes more time.
At the same time, that’s exactly why this is such a great opportunity for me. I’ve seen the other side, and hopefully I can bring some of those experiences into the organisation to help us work differently.
Kees: I hope you succeed.
Yorick: Thank you. Me too.
Kees: We all hope so, I think. Thank you very much, Yorick, and thank you, Bira.
In your keynote, Yorick, you also mentioned that we don’t know what the future holds. You elaborated on the role of the bank as a trusted safeguard, and that trust is not the same as loyalty. Banks really need to keep that in mind throughout this entire AI transition.
I think that’s a nice bridge to your keynote, Kyriakos, because when we talk about trust, AI governance becomes incredibly important. Your keynote was titled AI Governance: Safeguarding Trust While Accelerating Innovation, which is obviously a very relevant topic in the current AI era.
Could you elaborate a bit more on how organisations can accelerate innovation while at the same time safeguarding trust?
Kyriakos: It’s a complex question to answer.
A lot of people in financial services think that regulations and the governance frameworks around AI will slow down innovation and create barriers. Personally, though, I prefer to look at it differently.
We often say that AI governance must scale with AI. In financial services, AI is moving rapidly from experimentation towards enterprise-wide adoption. Governance therefore needs to evolve at the same pace in order to preserve trust, consistency and control.
Throughout today’s event we discussed how important trust is within financial services, and I really want to emphasise that point.
During my presentation, I also spoke quite extensively about trustworthy AI. Trustworthy AI requires oversight. Transparent processes, meaningful human oversight and continuous monitoring are what keep AI fair, explainable and aligned with regulatory expectations over time.
The third point I wanted to make is that innovation requires strong foundations. Sustainable AI innovation depends on clear governance frameworks, defined accountability and robust operational processes throughout the entire AI lifecycle.
Kees: Thank you very much. I think that was explained very clearly.
If you look back on today’s event as a whole, do you feel that organisations are sufficiently aware of the importance of AI governance, or are we not there yet?
Kyriakos: I think most organisations are definitely aware of its importance. However, based on my own experience, and also from conversations I had today, I think we’re still behind where we need to be.
During my presentation I shared some statistics from recent surveys. Many financial services organisations don’t yet feel fully prepared for the regulations that are coming. And now that we’re moving towards agentic AI, there’s also a great deal of uncertainty within financial institutions about how they will successfully meet all the regulatory expectations.
So the awareness is certainly there, but there’s still a lot of work to do to establish the governance frameworks that are required.
Kees: Thank you very much.
We’ve discussed that throughout the day as well, including during the keynote by Frans van Bruggen and in the session with Onur Can Koltukcu from De Nederlandsche Bank.
Menno, let me turn to you. EY was, of course, the main partner of today’s AI event, so thank you for that.
Looking at EY’s contribution today, Hans van den Heuvel gave the keynote Beyond Code: The True Organizational Impact of the Software Development Life Cycle with AI. He spoke about how software development is really at the core of the bank.
From EY’s perspective, what are some of the key findings you’ve seen throughout this AI transformation?
Menno: That’s a great question. I could probably spend 15 minutes answering it.
Overall, I don’t think creating output with AI is necessarily the difficult part. The real challenge is making sure you can actually trust the output that’s being generated.
That’s really what we’re working on. Building enterprise-wide, scalable AI solutions is, in my view, the next major step, and that’s where we’re focusing most of our efforts.
One thing I also really appreciated today, going back to the discussion on governance, was something Frans van Bruggen and Onur Can Koltukcu both mentioned. It’s very easy to hide behind governance. Of course we need to be careful, but governance should never become something that limits innovation.
With that mindset, you can continue pushing the boundaries while remaining fully aware of the potential risks. That’s essentially what we’re trying to do at scale.
Kees: And how do you actually scale something like that? As you said, it’s relatively easy for individuals to use AI, or to give everyone access to Copilot. But how do you scale AI across an entire organisation?
Menno: We basically look at it across three levels, or three horizons.
The first is the foundation. Within our own organisation—and we often use ourselves as “Client Zero,” just as many of our clients do—you first make sure that everyone starts working with AI. For us, that’s around 5,000 people.
One myth I see in the market is organisations saying they’ve invested enormous amounts of time in training people. In reality, many organisations actually underestimate how much training is required. It’s not just a couple of hours. You need to invest significant time to ensure people really learn how to work with AI and actually adopt it.
So the first layer is making sure everybody becomes comfortable using the core AI platforms.
The second horizon is embedding AI into business processes. Right now, many organisations simply sprinkle a little bit of AI on top of existing processes. You have a process with 55 steps, and maybe step 15 or step 47 contains some AI.
But that’s not how you should think about it. You need to ask yourself: what job are we actually trying to get done, and how could AI fundamentally change the way we accomplish it?
The third horizon is looking at the future of your organisation itself. What does your business model look like in an AI-driven world? What does your operating model become? And how do you transform towards that future?
That’s how we’re changing our own organisation, and it’s also how we work with our clients.
Yorick: Thank you. I’d like to follow up on something you just mentioned, Menno.
When you talk about your clients and AI adoption, are there any best practices you can share on how organisations can move from their current level of AI adoption towards becoming truly AI-native companies?
Menno: Absolutely.
One thing I see very often is that organisations are constantly chasing the next model. Everybody gets excited about the latest model, and suddenly the whole discussion becomes about technology.
Once they’ve selected the next model, they start looking at what it can do. But ultimately, that’s the wrong starting point.
The first question should be: what is the strategy of the organisation?
That’s where I see the biggest change today. More and more leaders are actively deciding whether they want to become AI-led organisations.
So first comes the mindset. Then comes the vision. After that you determine which capabilities you need to build. Only then should you start thinking about the technology.
If you approach it that way, people understand the direction. AI no longer becomes a side project where a handful of people build another proof of concept. Instead, it becomes part of the organisation’s strategic direction.
People have clear KPIs. They’re measured against them. Those are the things that really work.
Otherwise, organisations simply keep chasing the next model and building another proof of concept that ultimately goes nowhere.
So clear strategic direction and leadership that genuinely steps forward—that’s what organisations need right now.
Yorick: I completely agree. It really resonates with me.
One of the key things you mentioned is that technology and tooling should come later in the process. In the end, it starts with understanding what problem you’re actually trying to solve.
How do we move from where we are today towards a clear ambition, vision and strategy? Only after that should we decide which technology and tooling will help us achieve those goals.
I think many organisations still do it the other way around. There’s so much hype around the newest AI tools that they immediately start implementing them and encouraging everyone to use them, without first defining what they’re actually trying to accomplish.
Menno: Absolutely.
And ultimately—and this also came up today—we’re now moving from FOMO to FOBO: the fear of becoming obsolete.
Questions about the future of jobs and the future skills people need are becoming increasingly important.
I’m actually having a lot of conversations with universities at the moment about exactly that. What is their role in preparing the next generation of the workforce?
Kees: Thank you very much for elaborating on that topic, and thank you, Yorick, for the follow-up question.
You actually created the perfect bridge to the next topic. It was so perfect that I simply had to use it.
You could argue that organisations themselves need to evolve before AI can truly deliver results. And, by coincidence, that’s exactly what your presentation was about, Bira.
Could you tell us a bit more about your presentation and its key insights? Besides the fact that it was a really cool presentation, especially for the people who weren’t here, with the animation of you arriving from space—that was AI-generated, wasn’t it?
Bira: Well, yes, it was AI.
Kees: So that wasn’t actually me? It wasn’t real?
No? Well, that clears that up.
But could you tell us a bit more about how an organisation can actually deliver results with AI? And how does an organisation evolve alongside it?
Bira: I think what you mentioned, Menno, in response to Yorick’s question was spot on.
The hype cycle and the anxiety-driven environment we’re currently in don’t really give organisations the breathing space to think carefully about how they should approach AI. Everyone is chasing the next best model.
But what we’re also seeing is that AI models are becoming a bit like the iPhone. Who really cares whether the next version is called Opus 4.7 or 4.8? It’ll be twice as fast and take slightly better pictures, but that’s not fundamentally changing the conversation anymore.
Boards of large banks and financial institutions don’t necessarily experience that same sense of calm. There’s still a lot of anxiety, and we need to take that away.
The way to do that is exactly what you were discussing earlier. Our presentation focused on the fact that many things are happening at the technological frontier. Unlike media companies—which had to adapt extremely quickly—financial institutions still have what people today referred to as moats. Those moats still exist, but in the Netherlands at least, they sometimes become excuses not to move.
The real question is: how do we preserve those strengths while still adopting AI? Because AI is a general-purpose technology. We can’t afford to wait. We have to move, but at a pace that fits our responsibilities. Financial institutions don’t just have to innovate—they have to safeguard people’s money and the stability of the financial system.
Our perspective was that, first of all, AI is a general-purpose technology. Every industry will be using it, so financial services should learn from developments in other sectors. This isn’t a challenge unique to banking.
Secondly, implementing AI inside financial institutions requires a layered approach. You can’t let technology move much faster than governance, but you also can’t spend years perfecting governance before showing any results, because then boards become anxious. You need early successes through relatively small use cases. Those successes help build confidence, secure further investment and create momentum.
The third point is perhaps the most fundamental one for society as a whole. If AI automates routine work, what are all of us going to do?
Today I heard some inspiring answers, but sometimes those also feel a bit socially desirable. Is everyone suddenly going to focus on strategy? No. Is everyone going to spend all their time building relationships? Also no.
Different roles will require different development paths, because some jobs are inevitably going to disappear. That’s one of the insights we wanted to bring: what does AI implementation actually look like when you deploy a product that has real impact? How does that affect the market, and how does it affect the organisation itself?
Kees: Thank you very much.
You also raised an important question: what is the end state? Where are we actually heading? I heard several speakers today say that we simply don’t know. We know where we are today, but it’s equally important to keep looking ahead.
That also connects to something Yorick discussed earlier about moving towards a world that’s perhaps 1% administration and 99% interaction.
Bira: I think that was actually Duco who said that.
Kees: You’re right, it was Duco van Lanschot. It was such a good point that I almost attributed it to you.
What I’m really curious about—and I think we should probably start wrapping up, although I could easily continue this conversation for another three hours—is where we’ll be one year from now.
We’ve talked today about the relationship between humans and AI, about the importance of keeping the human in the loop. We’ve discussed that generating output isn’t the difficult part anymore, but scaling AI is. We’re moving towards a stage where organisations are beginning to scale AI successfully. We’ve talked about trust, and about making sure we don’t confuse trust with loyalty.
If we look one year ahead, where do you think we’ll be? Will AI be fully scalable? Will the hype be over? Or will we be implementing AI in completely different ways than we imagine today?
Kyriakos, may I start with you?
Kyriakos: That’s quite a difficult question.
I don’t think that even a year from now we’ll be able to say that AI has been fully scaled across our organisations. Technology is evolving too quickly. New capabilities continue to emerge, especially with the rise of agentic AI over the past year.
So I don’t think we’ll be able to say we’ve completed the journey.
What I do believe is that we’ll have made meaningful progress. Regulations, governance frameworks, improved processes and greater knowledge-sharing—not only within financial services but also across industries—will help establish best practices for deploying AI more successfully.
We’ll also continue building the skills that organisations need for successful implementation.
So yes, I think we’ll definitely move forward over the next year, but there will still be plenty of work left to do.
Kees: So perhaps we’ll simply have taken a much stronger first step.
Thank you. Yorick, how about you?
Yorick: A year passes very quickly, so we shouldn’t expect the world to look completely different.
What I do expect is that the technology itself will continue evolving rapidly. That’s particularly interesting when you look at agentic AI. Today, more than 80% of agentic AI use cases are focused on software development—things like agentic coding—and that’s logical because that’s also where new applications are being created.
Over the next year I expect many more practical use cases to emerge. AI itself will increasingly develop technology that enables applications we simply don’t have today.
Think, for example, of refrigerators that automatically recognise their contents and order groceries whenever something is running low. Those kinds of technologies are coming.
The technology will arrive first, but adoption always follows at a very different pace. I think that’s something we all need to remember.
I often compare this with digital assets. Fifteen years ago blockchain technology emerged, and many people claimed banks would become obsolete because everything would move on-chain.
Fifteen years later, the technology is better than ever, yet adoption is still happening gradually. But it is happening.
I think exactly the same pattern will apply to AI.
Kees: So further implementation, perhaps at a different pace than people expect. Thank you very much.
Menno?
Menno: For me there are a couple of things.
Whether scalable AI arrives within a year remains to be seen. But there are two things I really hope we’ll spend much more time on.
The first is investing in upskilling the entire workforce. I think that’s absolutely essential.
Even today, many people in the audience weren’t yet active users of AI in whatever way suited their work. Hopefully next year everyone will immediately raise their hand proudly when asked whether they’re using AI.
The second thing relates to some research we’re currently conducting into agentic experiences.
In financial services we’ve traditionally said that during important life events—someone passing away, financial distress, those kinds of situations—you need a human being because only a person can show empathy and emotion.
But what we’re increasingly seeing is that people are actually perfectly comfortable interacting with an AI agent. They’re looking for empathy, but perhaps not necessarily for human emotion.
Hopefully next year we’ll not only be discussing what we’re doing inside banks, but also how customers themselves are experiencing these new AI interactions.
Kees: I love that.
So many good answers. Every answer leads to three new questions, but otherwise we’ll be sitting here all evening.
Last but certainly not least, Bira. The same question for you. One year from now—what do you expect?
Bira: It’s a difficult act to follow Menno.
I have both an expectation and a hope.
My expectation is similar to what Yorick mentioned. We’ve been talking about AI for three, four, maybe five years now. Yet in practice it’s still difficult to get things done.
What I do expect is that we’ll see many more AI-native companies emerge. Wherever inefficiency is part of your business model, there’s probably a Silicon Valley start-up already trying to disrupt it.
So I expect to see far more practical use cases next year—solutions from these companies that may genuinely change the world over the next five to ten years.
For the larger banks, I expect them to continue executing the transformation programmes they’re starting today.
And I agree with Menno that the first focus will be on improving internal processes, followed by creating value for customers.
My hope is that, alongside all those investments in AI, we’ll invest just as much in people. We need to develop the capabilities and skills of employees—and of society more broadly—so that our resilience in dealing with AI grows alongside the technology itself.
Only then will we find the right balance between humans and AI.
It’s not enough to invest only in AI. We should invest just as much in people.
Kees: I think we all share that hope.
Thank you, all four of you, for taking the time to join us today. I thought it was a fantastic event—although I’m admittedly a little biased—but I’ve also heard a lot of positive feedback from attendees.
Will I see all of you again at next year’s event?
All: Absolutely.
Bira: Is that an official commitment?
Kees: Yes, we’re recording it now.
Bira: Then absolutely.
Kees: Perfect. We have that on tape.
Thank you all again for joining this podcast. It has been a very insightful conversation.
One thing that particularly caught my attention today came up during one of the keynotes. Someone mentioned that Chinese investors are increasingly looking at European alternatives because of the governance frameworks we’re putting in place.
I think that’s a fascinating development, especially when you consider the different roles of the United States, China and Europe. My hope is that, when we’re sitting here again next year, we’ll have much more clarity on that as well.
Speaking of next year: we’ll be back with another Leaders in Finance AI event on 3 June 2027.
Thank you very much for listening.
I’d also like to thank the partners of today’s event: EY, SAS, Yellowtail Conclusion, the Nederlandse Vereniging van Banken, and Lepaya.
Finally, I’d like to once again thank Menno Bonninga, Partner at EY; Bira Thanabalasingam, Managing Director at Yellowtail; Kyriakos Fistos, AI & Analytics Advisor at SAS; and Yorick Naeff, Chief Innovation Officer at ABN AMRO.
Thank you very much. Enjoy the rest of your day, and I’ll see you next time.
Yorick: Thanks for having me.
Voice-over: You’ve been listening to Leaders in Finance. We hope you’ve enjoyed this episode and would love to hear from you.
What’s on your mind? Who would you like to hear next?
Let us know in a review, send us an email, or get in touch via our social channels. We’d greatly appreciate it.
Finally, we’d like to thank our partners for their ongoing support: EY, Mogelijk Real Estate Finance, Doen, and Lepaya.
Don’t forget to check out everything else we do at leadersinfinance.nl.
Thank you for listening.