For ABN AMRO, the breakthrough came early. As early as March 2023, the bank gained access to a private version of a large language model through Microsoft. Within six weeks, the first concrete application went live in the contact center: automatic summaries of customer conversations. That pace made clear right away that this wasn't an incremental technology, and it brought AI straight into the boardroom.
The moment AI became "serious"
Both the executive board and the supervisory board were involved early on in the strategic implications. The conclusion was clear: this called not for cautious experimentation, but for deliberate acceleration, within clear boundaries.
Accelerating with guardrails
That acceleration came with conditions. For a bank, trust is the foundation. That's why ABN AMRO chose from the outset to use a private AI environment that doesn't train on customer data. Alongside that, clear guardrails were put in place: mandatory training for employees, human review of AI output, and ongoing feedback loops to improve the models.
Even practical problems, such as errors in recognizing gender in conversation data, were solved together with employees. This approach made AI not only safe, but also usable right away.
From optimization to radical redesign
According to Annerie Vreugdenhil, the real value of AI lies not in improving existing processes, but in rethinking them from the ground up. The core question shifts from "how do I make this process more efficient?" to "how would I build this process if I started from scratch today?" That means letting go of existing assumptions.
"How would I build this process if I started from scratch today?"
The customer of the future: human and agent
One of the biggest shifts is the rise of AI agents acting on behalf of the customer. Where customers once came into branches in person and later switched to video calls, a new scenario is now emerging: a digital agent that acts on the customer's behalf. That raises fundamental questions: who is the customer, who holds the mandate, and how do you maintain trust once interactions become machine to machine?
For banks, this means trust itself needs to become machine-readable, a new and still largely uncharted challenge.
Criminals use AI too
At the same time, security has to be safeguarded. AI doesn't just make banks smarter, it makes criminals smarter too. Document fraud, identity forgery and synthetic data are all becoming more sophisticated. That shifts an important principle: traditional evidence can no longer be trusted by default. Banks therefore need to reinvent their data checks and verification methods, in what amounts to an ongoing arms race with fraudsters.
Employees as the driver of adoption
Notably, AI adoption within ABN AMRO is being carried largely by employees themselves. The contact center set the tone, and employees felt the benefits directly: less administrative work, more focus on the customer and more room for empathy. Around 85% of the organization now works with AI tools, ranging from simple prompts to complex applications.
"Change gets accepted when it makes the work more meaningful, not just more efficient."
AI is changing the work itself
The impact goes beyond productivity. Roles are shifting from doing the work to checking and judging it. Employees now need to interpret AI output, understand how systems arrive at their conclusions, and keep thinking critically. That calls for a new combination of expertise: subject-matter knowledge together with an understanding of AI.
The challenge of developing talent
That brings a new problem with it. Organizations traditionally build expertise by having junior staff gain experience on basic tasks, but that's exactly the work AI is now taking over. So the question becomes: how do we train new generations once that learning curve disappears? This problem extends beyond banking, into education, for instance, which struggles to keep pace with technological change.
From experiment to strategic direction
ABN AMRO's approach to AI is moving from bottom-up experimentation toward more top-down direction. In the early phase, the emphasis was on discovery and learning. Now the focus is shifting to prioritizing use cases, targeted investment, and actually realizing business value. The thinking, too, is becoming less incremental and more transformational.
Where AI already delivers visible value
In the short term, the bank expects concrete improvements for customers: better digital assistants as chatbots become smarter and more relevant, faster processes as mortgage applications that used to take hours speed up dramatically, and more efficient compliance as KYC processes become more targeted and less of a burden for customers. Speed, simplicity and relevance together make up the visible value of AI.
The role of governance: not a brake, but an accelerator
Notably, governance isn't a blocker here. On the contrary: because banks are used to working within strict rules and model governance, they can scale AI faster and in a controlled way. The challenge isn't governance itself, but differentiation: when does an AI application need heavy oversight, and when doesn't it?
The limit of autonomy
One choice stays firm: a human remains ultimately responsible. Fully autonomous, opaque systems don't fit within a regulated environment. Being understandable and explainable remain non-negotiable requirements. That, in the end, is also what sets the limit on AI applications in the financial sector.
Where ABN AMRO wants to win
The ambition is clear: ABN AMRO wants to remain the personal bank, in a world that keeps growing more digital and more autonomous. That means excellent service through both AI and people, a consistent customer experience regardless of channel, and lasting trust, including in machine-driven interactions. The combination of technology and humanity becomes the differentiator.
3 elements that make an AI transformation work
ABN AMRO shows that a successful AI transformation comes down to three elements: early, controlled adoption; strong involvement from both employees and leadership; and a focus on trust and customer value. The bank of the future won't just be smarter, it will be organized differently. And that change is already well underway.