EXCLUSIVE: “AClearVision” – Allan Kissmeyer, SEB in ‘The Paytech Magazine’

SEB’s Allan Kissmeyer is a ‘big fan of clarity’. As more banks look to employ GenAI to help solve data-rich complexity in payment processes, he thinks there ought to be more of it from regulators

Universal adoption of ISO 20022, the international payment messaging standard, will deliver much richer information about the payment, its transaction journey and the payer and payee, and is considered to be the single biggest change in banking in decades.

It is costing the global industry billions of dollars to develop the necessary technological infrastructures to take full advantage of its merits – namely, faster, more efficient and secure payments – as well as to process the sheer volume of the accompanying data. And for that banks may well turn to large language model AI.

Transforming unstructured name and address data into an ISO 20022 structured format may not sound like the most exciting application of a technology that’s predicted to profoundly alter the way we live and work, but, for banks, this discrete application could be game-changing, particularly when it comes to cross-border payments.

Interpreting names and addresses in multiple languages, taking into account countries’ specific formats, and populating fields at a more granular level, as required by ISO 20022, is beyond much of the existing technology’s scope.

Allan Kissmeyer, head of cash management at Swedish business banking giant SEB, is excited about the improvements that harmonisation around ISO 20022 will bring to global payments – although a little more cautious about banks’ role in the development of generative AI (GenAI) applications here, as elsewhere.

“Cash management is really about standardisation,” he says. “It’s a marginal game, right? We process millions and millions of payments on a daily basis. So we are super, super dependent on a happy flow, the straight-through processing (STP), of every single payment, where small variations to that create quite a lot of hassle for us as a bank.

“That’s where AI can play an important role going forward, in interpreting those deviations. And that’s only one of the internal potential benefits I see with this technology. It brings a lot of opportunities, both behind the scenes for us as a bank, but also in terms of innovative solutions to support clients in their day-to-day business and create efficiency gains.

“But, personally, I think that banks should not be front runners within this [GenAI technology] area.”

Deciding the future

SEB is far from being an AI Luddite.

It was one of the first to introduce a chatbot in 2016. For more than two years, the bank has been using AI to analyse the foreign exchange market, giving it access to several hundred million data updates daily which provides traders with a powerful tool to analyse data and identify patterns, in turn allowing more effective decision-making processes.

SEB even has an AI fund, launched in 2023, that invests exclusively in companies that work with the technology. It enjoyed a strong first year, exceeding expectations with growth of more than 45 per cent. Many banks are already experimenting with large language models. An independent survey by financial technology researcher Celent earlier this year found that more than half of banks were evaluating or testing generative AI in some capacity, and more than a third of those polled believed the tech will have the greatest impact on the market in the next five years.

“When we regulate, we should leave no room for interpretation”

But only this year has the world seen the first legislation introduced to guide the development of the technology in the form of the European Union’s AI Act – and it still leaves question marks, according to Kissmeyer.

“I would like to see more clarity around regulation to avoid banks interpreting differently on what is allowed and what is not allowed,” he says.

The EU’s Artificial Intelligence Act lays down harmonised rules to provide AI developers and deployers with clear requirements and obligations regarding specific uses of AI. It also seeks to reduce the administration and financial burden for business, in particular small and medium-sized enterprises (SMEs). The overall aim of the new rules is to foster trustworthy AI within the EU and beyond by ensuring that AI systems respect fundamental rights, safety and ethical principles, and by addressing the risks of very powerful and impactful AI models.

However, previous experience has shown that standalone legislation can create contradictions and friction when it comes to implementation.

“I can give you a specific example and that is on the one hand you have AML5 that the local regulators are pushing hard on,” says Kissmeyer. “There are high expectations on banks in terms of making sure the underlying payment flow that we facilitate for our clients is compliant.

“You have sanction elements, you have fraud prevention, you have anti-money laundering, so to analyse that flow, banks must look into every single detail: who sends the money, what jurisdiction is the money going to, what are the industries involved, and so on.

“But on the other hand, you have PSD2, soon to be PSD3 regulation, that also pushes banks to open up their infrastructure for other players to utilise the rails. But how doyou let in others to use your rails if you are expected to know exactly what is done on those rails, down to every single little detail such as payment purpose?

“And what if you don’t have full trust in the player that wants to utilise your rails in their compliance function, their risk functions? How do you know what monitoring capabilities they have in place to avoid money laundering? And can you then exclude them from using your rails? Well, not according to PSD2 or PSD3.

“So it’s contradicting what we, as established banks, are supposed to do. I’m a big fan of clarity, so when we regulate, we should leave no room for interpretation.”

Although it’s not a regulation, that’s why Kissmeyer is such a big fan of the global industry adopting ISO 20022. By May of this year, he was confident that SEB was “pretty much over the hurdle, at least in Sweden, where we have most of our clients” of adopting the new format.

But he also stressed that the rails on which the payment system rests were being rebuilt, which are both time- and resource-consuming, meaning that other development initiatives had to take a back seat.

“In recent years, we have had limited capacity to drive other than mandatory changes. This will continue for the next two or three years,” he added. But the investment, in his view, is more than worth it.

“ISO is all about standardisation, right? It’s really about creating that solid foundation from which we can build and ensure that we have a robust and stable infrastructure. And that’s often what we tend to forget – that to get to that point where we together co-create with clients on using AI or blockchain or whatever other interesting technology out there, we need a solid foundation.

“We need a standardisation that provides the starting point, so we talk the same language, we use the same format.

This article was published in The Paytech Magazine Issue 15, Page 30-31