EXCLUSIVE: “Harvesting Trust” – Marnix van Stiphout, ING in ‘The Paytech Magazine’

ING’s blueprint for customer-centric AI-driven banking imagines a future where agents toil in digital farms, serving human overseers and, ultimately, customers. But it’s ploughing its furrow carefully

There is a version of the future where the world is run by systems that are autonomous from human beings, yet capable of knowing and characterising almost everything about their lives, including their finances. Every transaction parsed. Every behavioural signal scored. Every life milestone predicted before it is spoken aloud.

Credit pre-approved before it is requested; risk priced before it materialises. Invisible systems watch, learn and decide, quietly shaping the financial pathways available to each institution and the individuals it serves.

Omnipresent intelligence. Algorithmic authority. A financial panopticon, humming beneath daily life. For more cynical observers, the adoption of the technology driving this in banking edges uncomfortably close to the Orwellian. But as far as the man driving change at Dutch banking giant ING is concerned, that framing is wrong: ING’s technological architecture is being designed methodically, deliberately to empower customers, not have dominion over them.

Founded in 1991 through the merger of insurer Nationale-Nederlanden and NMB Postbank, ING has grown into one of Europe’s most prominent digital-first banking groups, serving tens of millions of retail, corporate and institutional clients across more than 100 countries. In recent years, the bank has paired balance-sheet strength with strategic transformation, reporting multi-billion-euro annual profits, returning capital to shareholders through buybacks, and investing heavily in digital infrastructure to support long-term growth.

Having built a reputation as something of a boundary-pusher, its financial resilience has given leadership the latitude to pursue large-scale technology modernisation, with AI positioned not as an experimental layer but as a core operating capability.

Operational infrastructure

Across operational workflows and decisioning environments, intelligent systems are set to permeate the bank’s core. But the strategic intent behind that roll-out matters, says COO Marnix van Stiphout. As he puts it: “I’m here to grow our businesses safely and securely, and I can only grow it when people want to bank with us.

“Anything to do with fulfilment is subject to AI-ification, if that’s a verb!” he adds. “Operating the bank means everything from business operations to finance, risk operations and HR, but also data, transformation and analytics.”

And in that context, AI is not a siloed innovation programme – it is operational infrastructure and, crucially, a growth lever. “The strategy of the bank is to grow across our markets,” says van Stiphout. “The task is to deepen those franchises safely, securely and at scale, making sure that we are so productive, that we can keep on adding customers, without necessarily having to add [resources] at the same rate.”

In this scenario, AI becomes the multiplier.

“It’s a fantastic opportunity and a big toolbox to build things at scale and do so safely and securely, giving us extra speed, extra content and also extra bandwidth,” he adds.

That doesn’t come without huge change. ING has been upfront about the number of human roles that could be affected by technology transformation at the bank. It has forecast that its total staff requirement could be 960 fewer by the end of this year. At the same time, new technologies will also bring better careers for people who re-skill and play a part in the transformation, says van Stiphout.

To be clear, customers are not crops to be cultivated, segmented and monetised, in a hyper-efficient, asymmetrical ecosystem that primarily benefits the bank. Rather, this AI production system is aimed at empowering and expanding the reach and capabilities of human employees across ING’s global business, by giving them virtual ‘staff’.

“If I am working today in wholesale lending operations, then tomorrow I could be overseeing business banking and other loans with the help of agentic solutions,” he says.

In this way, he is envisioning something closer to precision farming for maximum individual and communal benefit. It is financial wellbeing: helping customers avoid missed payments. Financial inclusion: access to appropriate credit. Financial protection: detecting fraud earlier and providing guidance tailored to customers’ circumstances. In this framing, the intelligence generated becomes protective and enabling. AI-driven self-service allows individuals to act instantly – refinancing loans, adjusting savings strategies, resolving disputes – without waiting in call queues.

The system anticipates needs, but does not remove choice. Human advice remains available, layered on top of automation rather than replaced by it.

“The toolbox gets us a lot closer to our clients,” van Stiphout says. “When you are able to decide yourself to do something with a bank, but still have the opportunity to speak with people, what you produce as a banker is far closer to what the customer wants.”

It’s an inversion of the darker narrative: here, automation expanding autonomy rather than constraining it. Hyper-personalisation,after all, could edge towards manipulation.

Behavioural modelling, so powerful in anti-money-laundering detection, could theoretically be repurposed toward aggressive product steering. Predictive analytics could entrench bias if poorly governed. Which is why van Stiphout is precise in describing the degree of autonomy his envisaged AI agents have to operate.

“Autonomous sounds very negative,” he agrees. “But a lot of IT is already working kind of autonomously – applications that do things themselves because we programmed them to.”

The important limiter with artificial intelligence lies in the governance around it.

“When ING deploys it, we take a work instruction and give that to a digital agent, who becomes, in effect, a colleague, bound by the same rules. We tell it: ‘If you reach your defined boundary, you need to ask a human for an intervention’. Agents do not roam freely across the data landscape. They operate within fenced fields.”

Scaling such systems, however, can be a roadblock – one ING is addressing.

“The main challenge is not lack of ideas,” van Stiphout says. “The challenge is ‘can the organisation adopt it at scale?’.”

Skills, risk frameworks, operating models, all must evolve, and the question for the bank becomes ‘can we convert our teams to be able to work with all these things across the world?’.

Hence ING’s incremental deployment philosophy, which van Stiphout summarises as: “Let’s learn before we really start running with this thing.”

Small proofs precede scale. Agentic mortgage processes in the Netherlands and Germany, for instance, are being rolled out step by step with salary retrieval agents and document agents, each generating operational value and governance learning. Chatbots follow a similar discipline.

Hallucination risk – where AI generates responses containing false or misleading information – is mitigated through continuous testing.

“We are very, very prudent in how we go about these things,” adds van Stiphout.

This measured approach extends to regulation. But van Stiphout resists the notion that AI introduces wholly new operational risk categories.

“Why would it be different from any other IT that we apply?” he asks, framing AI as an evolution rather than an exception, albeit one requiring careful understanding before use is industrialised.

Don’t underplay the upsides

As fulfilment becomes automated, and employees shift toward oversight and reconciliation roles, van Stiphout concedes ‘that needs quite a rigorous retraining and repositioning of people and jobs’. Yet he sees mainly upsides here. Banking, long viewed by some technologists as legacy infrastructure with little to inspire a younger workforce, could become extremely attractive, he believes.

“We can actually get young, talented people in the bank to do that work. I think that gives us a real comparative advantage.”

Automation, in this sense, also reshapes employability. And always, the customer and their welfare, remains the defining principle.

“Yes, of course, we have conversations about social consequences,” van Stiphout says. “But those two [positive] things, in terms of employability and customer centricity, are true next to the debate about social consequence.”

The dystopian version of the future remains technically possible. A bank that can see everything must continually prove it will not use everything. Data minimisation, purpose limitation and ethical design become trust infrastructure as much as compliance obligation.

Van Stiphout is pragmatic about the unknowns, and the fact that new risks will undoubtedly surface. But ING is betting that governance, a strong guiding culture and regulatory oversight can bend the trajectory towards a more constructive outcome, where artificial intelligence, rather than command authority, operates as a financial co-pilot.

In that world, agents work the soil of data continuously, and their outputs are measured in customer resilience: fewer fraud losses, faster credit access, smarter savings pathways. It is still a system of immense informational power. It still watches, learns and decides at scale. But rather than an all-seeing command authority, ING would prefer that its customers experience AI as being something closer to an intelligent partner – one that’s present, analytical and, ultimately, in the service of their financial agency.

This article was published in The Paytech Magazine Issue #18, Page 06-07

The post EXCLUSIVE: “Harvesting Trust” – Marnix van Stiphout, ING in ‘The Paytech Magazine’ appeared first on FF News | Fintech Finance.