Summary

AI Event 2026

Looking to be part of next year’s event?
Join our waiting list to stay informed—we’ll only reach out twice a year with key updates.

Leaders from across the Dutch financial ecosystem gathered for the 2026 edition of the Leaders in Finance AI Event, bringing together roughly 150 attendees from more than 50 organizations for a morning of keynotes, interviews, panels and cases on AI in financial services. The day moved between excitement and anxiety: last year’s conversation was about the possibilities of AI, this year’s was about the much harder work of implementation, trust, and what remains distinctly human once agents start doing the routine work.

This document summarizes the speeches, interviews, panels and cases at the event. It is not a verbatim transcript, but a paraphrased synopsis of the key points made. It has been prepared and published by Leaders in Finance. Please note that this summary was created with the help of AI tools. While care has been taken to ensure accuracy, the content may contain errors or omissions. For full clarity or specific details, please feel free to contact us.

Key takeaways

The hype phase is over, and implementation is harder than expected. Across almost every conversation, the same shift came up: last year was about AI’s possibilities, this year is about getting it into production. Many organizations still lack the data foundations, governance and oversight to do that responsibly, and speakers estimated that 80–90% of AI initiatives still struggle to reach production.
Trust, not technology, is the real bottleneck. Whether the topic was regulation, onboarding or legacy code replacement, the recurring message was that codifying rules, building explainability and keeping humans in control at the right points is what allows AI to actually go live, especially in a sector where mistakes are expensive and reputational damage is severe.
Agentic AI is starting to sit between banks and their customers. Several speakers warned that as customers increasingly let agents choose where to bank, invest or insure, financial institutions risk being pushed into the background as an interchangeable “pipeline.” Being present in the agent connectivity layer, through open APIs, is becoming a competitive necessity, not a nice-to-have.

Identity, fraud and compliance are being reshaped faster than defenses can keep up. With AI-generated documents, deepfakes and synthetic identities, fraud is compounding fast, while some institutions still run multi-year procurement processes to replace basic verification tools. Compliance and identity checks need to move from box-ticking to genuine, continuously updated judgment.
The human differentiator is shifting from creating to evaluating, from routine to relational. As AI takes over codifiable, verifiable tasks, several speakers argued that lasting value will come from judgment, relationships, ethics and creativity, the skills that are hardest to codify. The advice throughout the day was to use AI like a Formula One car that sharpens your thinking, not a self-driving car that dulls it.

Opening – Irene (Moderator)

Irene, moderator of the day, opened by sharing themes she had picked up in conversations with attendees ahead of the event. The clearest shift compared to last year was the move from possibility to implementation, with several people noting that despite constant new model releases, 80–90% of AI initiatives still struggle to reach production, creating what one attendee called an “anxiety gap.” Agentic AI came up repeatedly, with warnings that banks not present in the agent connectivity layer risk losing business, or being reduced to a “pipeline.” She also referenced the term FOBO (fear of becoming obsolete), the ABN AMRO CEO’s remark that AI is “humanity’s fourth narcissistic wound,” and concerns about Europe’s exclusion from cybersecurity initiatives such as the Anthropic Methos model. She closed by polling the room on AI tool usage, agents, and whether organizations are experimenting with, cutting costs with, or generating new revenue through AI, finding that far fewer hands stayed up for genuine new revenue generation.

Keynote – Yorick Naeff

Yorick Naeff, Head of Innovation, ABN AMRO, opened the day by arguing that banks need to evolve from traditional financial services companies into tech companies. He described most of a bank’s traditional layer, payments, basic transactions, as increasingly commoditized, and therefore a natural target for agents to handle on a customer’s behalf. He outlined four things banks should focus on in an agent-mediated world: recognizing that trust does not equal loyalty (customers may still switch if an agent finds better terms elsewhere), owning high-quality data, securing a place in the connectivity ecosystem via APIs, and competing on pricing and transparency. On the pace of change, he predicted a hybrid future rather than a sudden takeover by agents, comparing today’s AI enthusiasm to overly optimistic predictions made about digital disruption 15 years ago.

Speech – Frans van Bruggen

Frans van Bruggen, Lead Values, Standards, and Regulations AIC4NL, Future Bridges BV, formerly senior policy officer for fintech and AI at De Nederlandsche Bank, argued that Europe over-regulates rather than under-innovates. He raised concerns about Europe’s dependency on US and Chinese frontier models, the exclusion of European players from initiatives like Anthropic’s Methos cybersecurity project, and the scale of AI infrastructure investment happening in the US and China compared to Europe. He warned against “token maxing” (deploying agents that burn budget without adding value) and shared research suggesting significant disruption to business, finance and administrative roles. On regulation, he stressed that the EU AI Act’s impact on financial services is narrower than assumed, and encouraged institutions to proactively engage regulators rather than over-interpret rules defensively. He closed with a case for human value in creativity, decision-making, relational work and skilled trades, noting research that philosophy graduates currently have some of the best job prospects in the AI era because of their critical thinking and reasoning skills.

Boardroom Conversation – Onur Can Koltukcu &a Cyprian Smits

In a moderated conversation, Onur Can Koltukcu, De Nederlandsche Bank, and Cyprian Smits, Head Data Science, Rabobank, debated whether regulation slows down innovation. Both agreed the answer is “yes and no”: regulation sets boundaries within which institutions can innovate, but uncertainty around definitions like bias and explainability can genuinely slow decision-making, and banks’ own compliance departments are sometimes stricter than regulators require. They discussed the idea of “codifying” regulatory requirements into scalable, automatable rules, with Onur cautioning against freezing today’s limited understanding of concepts like discrimination into rigid thresholds. Using the Bunq court case as an example, they illustrated how regulatory rulings can be successfully challenged when the underlying standards are unclear. On European competitiveness, both stressed that regulation is only part of the challenge, alongside legacy systems, data quality, fragmented legal jurisdictions and a lack of European investment in European AI companies. Their closing advice: institutions should invest in foundational tech capability to stay in control (“you can scale as fast as you can control your agents”), and Europe needs sustained demand for European alternatives, not just subsidies for innovation.

Speech – Duco van Lanschot

Duco van Lanschot, founder, Duna, an AI-native platform for business onboarding and compliance, argued that financial institutions should be far more “paranoid” about AI-driven fraud than they currently are. He described how AI has made identity fraud dramatically easier and cheaper to produce at scale, through AI-generated documents, deepfake video and synthetic company identities, while online fraud is estimated to be growing around 30% year on year. He contrasted the multi-year procurement cycles some legacy banks still run to replace basic verification tools with the “unlimited time and zero marginal cost” available to attackers using AI agents. He argued that most compliance today is 99% box-checking and 1% judgment, and that this ratio needs to flip, with identity treated as a living, continuously updated system rather than a one-time check. He also highlighted the business case for better onboarding: AI-native flows reduced submission time by over 60% and follow-up questions by 51% in his company’s data, while poor onboarding experiences are driving customers toward challengers like Revolut.

Interview – Sheila Gemin

Sheila Gemin, formerly Global Head of Corporate Technology at ING, currently on sabbatical, argued that technology is the least interesting part of AI transformation. Having scaled technology platforms across 44 countries, she said the hardest part was never the architecture itself but organizational resistance, particularly when change is not “invented here.” She stressed that AI strategy should start with the question “for what,” warning that too many organizations focus on models and infrastructure without clarity on the business outcome they are pursuing, especially new revenue rather than just efficiency gains. She distinguished between three types of AI in use at ING: end-user tools like co-pilots, AI embedded in business processes, and AI used in engineering and code generation, and flagged third-party AI risk (referencing the Salesforce/Odido incident) as a growing governance challenge. Her closing message was that organizations underestimate the “middle layer,” the product owners, process owners and engineers who need to be brought along in the transformation, not just the innovation teams at the top.

Speech – Kyriakos Fistos

Kyriakos Fistos, Senior Data Scientist and AI & Analytics Advisor, SAS Software, focused on AI governance as a business imperative rather than a nice-to-have. He cited research showing that while nearly nine out of ten financial institutions already use AI in production, only a minority feel fully prepared for regulation, a clear gap between adoption and governance maturity. He described governance as requiring both strategy (who is accountable, what the rules are) and technology (enforcing and measuring those rules), organized across four dimensions: culture, operations, compliance and oversight. He outlined a maturity path from foundational (fragmented policies) to responsive (reactive compliance) to proactive (governance built ahead of regulation), and gave four practical starting steps: gain visibility into all AI use across the organization, assign clear accountability (including RACI matrices), review models continuously before and after deployment, and treat monitoring as an ongoing process rather than a one-off exercise.

Speech – Hans van den Heuvel

Hans van den Heuvel, Partner, EY, formerly at ING for 25 years, focused on trust in AI-generated outcomes, particularly in legacy code replacement. He described the problem of decades-old core banking code, sometimes written in languages current AI models still struggle with, built for an era of paper transactions and monthly statements. He warned that automating single steps in a process without redesigning the whole flow creates “clogging”: more code gets written, but less of it reaches production reliably. His proposed fix is an agentic flow that breaks big tasks into small, checkable steps, wherever possible turning fuzzy rules into hard, deterministic checks that can be automated and audited, with a human only stepping in when repeated automated attempts fail. Every step generates a documented, approved trail for compliance. Asked what incumbent banks can learn from challengers like Revolut, he argued the advantage runs both ways: challengers respond faster to the market today, but will eventually face their own legacy and data clean-up challenges, as some tech companies like Spotify already have.

Speech – Bira Thanabalasingam

Bira Thanabalasingam, Managing Director, Yellowtail, framed AI as a general-purpose technology on the scale of electricity or the internet, meaning organizations have to adapt rather than treat it as a passing hype cycle. He introduced a model for deciding where to apply AI, based on how codified knowledge is and how verifiable the output is: highly tacit, hard-to-verify work (like client relationships or ethics) remains human core; codified but hard-to-verify work needs standardization; verifiable-but-tacit work benefits from AI-as-assistant with a human in the loop; and fully codified, verifiable work is where AI can fully substitute humans. He argued that as AI takes over routine, verifiable tasks, human value shifts up a “knowledge work pyramid” from creation to evaluation, critique and judgment. He demonstrated this with Yellowtail’s VisionAI product, used in roughly one in three Dutch mortgage applications, which extracts and checks mortgage documentation against acceptance rules, freeing staff to focus on edge cases and qualitative work rather than routine document review.

Interview – Frederike Grift

Frederike Grift, soon moving from NN Group to ING as a product manager for analytics and AI, shared her experience building an AI capability from scratch at NN’s Czech Republic business unit, where there were initially no data scientists and open skepticism from leadership. She started by training a small group of “business translators” from across departments to identify and build AI use cases themselves, one of which became claims straight-through-processing: using AI to read medical documentation and check it against policy conditions, work previously done manually by claims handlers. The initiative faced resistance from staff effectively helping automate their own roles, which she addressed through transparency and framing it as a chance to build future-relevant skills. The result: 15% of claims are now handled automatically, with a target of 40% this year, and resistance has notably decreased as colleagues, including the original skeptical sponsor, have seen it work. Her advice: focus on a small number of concrete use cases, connect business and technical people directly, and start small before scaling.

Closing keynote – Roland van der Vorst

Roland van der Vorst, Senior Advisor to the Executive Board, Rabobank, closed the day by framing AI as an “existential technology,” one that changes how we understand ourselves and our work, not just an instrumental tool. Drawing on neuroscience, he argued that AI excels at “left hemisphere” thinking: breaking the world into codifiable, verifiable, repeatable units, while humans bring an irreplaceable “right hemisphere” capacity for relationships, intuition, ethics and lived experience that resists codification. He warned that the current wave of AI adoption is mostly shifting work from the right to the left, and urged leaders to also protect and invest in the harder-to-measure right-hemisphere skills. On competitive dynamics, he argued that AI intensifies a long-running tension between the breadth of a customer base and the depth of the relationship with each customer, and that several of banks’ traditional competitive advantages, reputation, regulation, balance sheet strength, customer inertia, and expertise, are now under real threat from neobanks, agentic AI and even non-financial platforms entering financial services. Referencing companies like Comcast and Telefónica that survived earlier waves of digital disruption by investing early in adjacent capabilities, he urged banks not to repeat the mistake of waiting passively. His closing advice to individuals: use AI like a Formula One car that sharpens and challenges your thinking, not a self-driving car that quietly erodes it.

Closing remarks

The day closed with thanks to the event’s partners and organizing team, and an announcement that the next Leaders in Finance AI Event will take place on 3 June 2027.

Lunch & Networking

See you at the AI Event 2027!

Looking to be part of next year’s event? Join our waiting list to stay informed—we’ll only reach out twice a year with key updates.

Uniting the financial sector by discussing pressing topics and enhancing cooperation. That’s what we love to do at Leaders in Finance. By listening, learning, and connecting with others, we accelerate the sharing of ideas, thus powering (upcoming) leaders and organizations to shape the future of financial services.

Want to explore how we can benefit your organizational goals? We’re happy to meet and discuss opportunities. Each part of the Leaders in Finance Group has its unique approach. Want to explore how we can benefit your organizational goals? We’re happy to meet and discuss opportunities.