ING’s Bahadir Yilmaz reflects on what the bank has learned after a year of testing AI. In this interview, he explains why ING is focusing on use cases that improve customer outcomes and how AI supports, rather than drives, the bank’s strategy.
Six months ago, Bahadir Yilmaz was cautiously optimistic. The Chief Analytics Officer of ING had just begun a set of experiments to test generative AI across the bank’s operations. His view at the time was pragmatic: there was promise, but also a long list of unknowns.
Today, the tone has changed. Speaking to The CFO at Money20/20 Europe, Yilmaz was clear. The bank has moved past experimentation. The focus now is on applying AI where it adds measurable value to the client.
“This is no longer about testing if AI works,” he said. “It’s about understanding where it works, and more importantly, where it matters.”
Yilmaz leads a 500-person analytics team inside ING, responsible for AI, machine learning, and the underlying platforms that support data-led decision-making. Their remit is wide, but not unfocused.
In the past year, they have applied AI to targeted domains such as KYC, anti-money laundering operations, contact centres, lending in wholesale banking, and content personalization in marketing. Each area was selected for a reason.
The goal was not to chase novelty, but to understand the boundaries of what AI could do in a regulated banking environment.
One of the first lessons was that not all gains are equal. Productivity improvements are possible in many areas, but only some of those generate meaningful value to clients.
In cases where AI contributed to speed, accuracy, or convenience, adoption was high. In cases where the benefit was primarily internal or aesthetic, adoption was limited.
“Clients respond well when the result is tangible,” Yilmaz said. “They care about whether the advice is accurate, whether the turnaround time is faster, and whether they can access the bank when they need to. The underlying method doesn’t matter as much.”
This approach has led ING to reframe its priorities for AI. Rather than investing in high-profile but low-impact features, the bank is now focused on use cases that align closely with core service delivery.
These include investment advisory services, mortgage operations, and always-on voice agents.
Each of these domains reflects a clear customer expectation. Clients want accurate financial advice. They want consistent service. And they want access at times that suit them, not just when the branch or call centre is open.
AI offers an opportunity to meet these demands without compromising cost, compliance, or quality.
“We are the largest mortgage provider in Europe,” Yilmaz said. “We have domain knowledge that can be transferred into digital workflows, supported by AI. This is where we see the most strategic benefit.”
The same thinking applies to advisory services. As more customers expect financial insights to be available on demand, AI can help ING scale the expertise of its staff without replacing it.
Yilmaz made clear that this is not about full automation. It is about enabling faster, more tailored responses in areas where trust and precision are essential.
Voice agents, meanwhile, offer a way to improve accessibility without increasing headcount. But as with all automation, the bank is careful to balance efficiency with user experience.
Yilmaz noted a significant shift in the way both colleagues and customers talk about AI. A year ago, many remained unconvinced. Today, the skepticism has faded.
“There has been an inflection point,” he said. “I hear from people who used AI tools in their personal life and were surprised by how well they worked. Now they’re asking if we can use the same technology to solve business problems. That conversation is happening across the board.”
Even so, ING is not rushing to apply AI across every function. Yilmaz sees clear limits to its usefulness in areas that require human nuance, creative engagement, or deep contextual understanding.
He also acknowledged that auto-generated outreach or synthetic marketing materials have produced little impact.
“People don’t want to read AI-generated emails,” he said. “They want communication that feels real and is relevant to them.”
This measured approach reflects ING’s broader view that technology must support strategy, not define it. AI is not a headline. It is a component.
This pragmatism was echoed in Yilmaz’s view of the broader industry.
Compared to last year’s Money20/20, where AI branding dominated the show floor, this year’s emphasis was on core functionality. Payment firms were focused on reliability and integration. Fintechs were talking less about AI and more about product maturity.
“The industry has come back to fundamentals,” he said. “Clients want products that work. Whether those products are powered by AI is not the first question they ask.”
For finance leaders, the message is clear. AI should be pursued where it improves service, reduces friction, or enables scale. But its value is not intrinsic. It must be earned through application.
“Everything we are doing with AI is based on what our clients expect from us,” Yilmaz said. “That could mean faster access. It could mean better advice. It could mean lowering the cost of delivery. But if it doesn’t support the client outcome, we don’t prioritize it.”