AI knows your customers

Meanwhile, the people-heavy work in compliance is being transformed by multiple projects, already in early production or on the way. Customer due diligence –or KYC, as the regulation is called – is an area where ING is redoing workflows involving manual labour using AI.

“All the manual work we do in client due diligence, KYC, will largely be redone with a new model based on existing/available data, so we do not have to ask clients for all the data points,” said Van Stiphout.

He said that ING onboards clients largely digitally, unless they are complex: “Previously, when people wanted to get an account, they had to go to a bank branch to become a customer. They then got 100 questions and needed to give all kinds of data.

“What we are doing now is we’re saying to existing clients that need to be reviewed, and to new clients, that we’ve got so much data already from public sources and behavioural data. This includes how they bank with us and others. We can use it in the model, and if there’s 100 questions that generally make up a customer due diligence, we can probably use all the data we have and answer 70 or 80 of them.”

“We will also look at agentic AI for products like mortgages, redoing the way we work with clients on getting mortgages from us”

Marnix van Stiphout, ING Bank

He said this is more than enough to draw a first conclusion. As a precondition of this, the bank needs its data actualised so it can run in the model, added Van Stiphout: “But if that’s the case and you’ve got a good working model with good outputs, which we have, then basically you largely no longer need to do the manual work.

“We don’t need to call customers every time or send an email, asking for something extra. So, customers gain in terms of speed and not being annoyed by ongoing information requests.”

Customer due diligence now takes seconds versus, “either a day in a very positive situation, or weeks and weeks in a very negative one”, depending on the extra information required, said Van Stiphout, adding that this allows staff to focus on real risk analysis rather than data gathering.

Van Stiphout expects certain work in operations to be done by 25% fewer people when AI is implemented but said the 25% can be used for growth and more complex tasks.

Transaction monitoring

A good example of redeploying staff to more complex roles is seen in transaction monitoring. The bank is already using AI in this area, which Van Stiphout said gives humans more interesting roles and increases the number of investigations that can be carried out. This is already in production, but ING plans to work on more powerful models.

“We use analytics in our transaction monitoring models to help our investigation process, allowing us to close alerts and investigations more speedily if they’re quite [standard],” he said. “When you do the vanilla stuff right, the risky stuff stays open and that means that the people can do this work, so productivity goes up and we can investigate more.”

As explained to Computer Weekly recently by ING CTO Daniele Tonella, the bank is enabling development around AI in five areas: know your customer (KYC), call centres, in wholesale banking to improve customer due diligence, in retail for the hyper-personalisation of offerings, and inside tech for engineering.

He explained the bank’s “conservatively aggressive” approach to AI and why the COO oversees its development: “We brought in strict governance that focused all exploration in AI on five areas, and only under the control of the COO. This is important because AI has gained a lot of attention and traction. Without this governance, and due to the entrepreneurial nature of our bank, we might have seen bits of AI all over the place.”

Van Stiphout said that he and Tonella are in contact on an almost daily basis.

Separately, Van Stiphout was behind the bank’s move to a global delivery model for operations in around 2009, when it opened a captive centre in Slovakia’s capital Bratislava. It now also has hubs in Poland, Romania, Turkey, Spain and the Philippines. The hub in the Philippines capital Manilla is the largest, with more than 7,000 staff. 

“Almost anything in tech and operations can go to one of these hubs,” said Van Stiphout. “We like to combine the like-minded teams of technology and operations where possible.”