EXCLUSIVE: "We’ll Take the AI Road" - Eric Bierry, SBS in 'The Paytech Magazine'

One Scottish building society’s bold bet on AI with SBS is demonstrating how the technology could turn a minnow into a maneate

Think of the AI front runners in financial services and a small mutually owned building society in the Scottish lowlands doesn’t readily spring to mind. Established in 1848, it’s the oldest building society in the world. And, with less than £2billion under management, it’s certainly not the largest.

And yet Scottish Building Society is the first institution to deploy SBS AI Foundation – a proprietary new offering from global financial technology provider SBS (formerly Sopra Banking Software), which provides a common layer across its products, so data from core lending, savings and digital engagement systems can be correlated natively. 

SBS entered the project with 30 preconfigured use cases for AI Foundation. Roughly 20 per cent were focussed on growth, another 20 per cent on risk and around half on improving back-office operations. Eric Bierry, CEO of SBS and Deputy CEO of 74Software, expected the building society to be cautious and begin by adopting risk modules.

He was wrong. “They took a decision to start with growth,” he says.

That says something important about the emerging AI divide in financial services. For smaller institutions, AI could provide access to analytical firepower previously associated with organisations possessing far greater technology budgets and workforces.

“It’s not a question of the size of the institution,” says Bierry. “It’s really a question of agility of the executive.”

Scottish Building Society has decided to use the technology for more sophisticated customer segmentation, correlating mortgages and savings with factors including age, geography and length of membership. That knowledge is particularly powerful where customers hold time-limited savings products or have mortgages approaching significant points in their lifecycle. SBS AI Foundation can bring those signals together: perhaps a customer is two months into a three-month promotional savings rate, has two years remaining on a mortgage, is under 40, lives in a particular region and prefers a particular comms channel. 

Generative AI can take those data points and use them to create a campaign around a personalised offer.

There is currently a human at the end of that process approving the final action. But, says Bierry, the Society is already ‘close’ to saying ‘I’m comfortable enough to let the action [happen]’, without that final intervention.

Such hyperpersonalisation could enhance customer attraction, experience, product holding and, ultimately, the bottom line – not just for Scottish Building Society, but also for any organisation previously too small to develop that level of insight.

For Scottish Building Society, that means connecting previously fragmented signals to understand what an individual member might need, when they might need it and how best to approach them – while allowing its own people to investigate far more opportunities than their finite working hours once permitted.

And if that translates into faster growth without proportionately higher operating costs, the implications extend well beyond one Scottish mutual.

From weeks to seconds

Of course, it’s possible to diligently research a customer and then carefully hone a campaign around them. But the advantage of AI is the speed at which it can do it. What required two full-time employees and potentially weeks of work can, says Bierry, be reduced to seconds. That does not simply mean doing the same work more cheaply. It changes what it is practical to find out.

 “Now they are able to select 10, 15 possible marketing initiatives,” Bierry says of Scottish Building Society.  “They run all of them the same morning, and in the afternoon they decide the one they should be executing first. This is really impactful.”

The legacy challenge

Legacy banking platforms were built to process data, not make it readily consumable by AI. Information held in separate systems may describe ostensibly identical concepts differently. Before AI can reason across it, institutions need a semantic layer establishing what data means, where it came from and when it was accurate. SBS has designed its AI Foundation around that problem, combining a data platform and semantic layer with preconfigured use cases employing large language models, unlocking fresh growth potential for the organisation without a corresponding increase in organisational weight. Building societies have traditionally faced a relatively linear relationship between portfolio growth and the number of people required to administer it. 

Not any more, says Bierry: “We are starting to give the proof that Scottish Building Society could be increasing their portfolio by 20 per cent without increasing the team.”

Of course, it’s possible to diligently research a customer and then carefully hone a campaign around them. But the advantage of AI is the speed at which it can do it. What required two full-time employees and potentially weeks of work can, says Bierry, be reduced to seconds. That does not simply mean doing the same work more cheaply. It changes what it is practical to find out.

“Now they are able to select 10, 15 possible marketing initiatives,” Bierry says of Scottish Building Society. “They run all of them the same morning, and in the afternoon they decide the one they should be executing first. This is really impactful.”

 The implementation for Scottish Building  Society only went live in July 2026, but if the thesis holds, the Society could grow substantially while remaining a relatively lean mutual – indeed, it’s already announced it is expanding its mortgage offering to the whole of England as well as Wales from this year. There is bravery in its strategy because the easier option was to wait for others to make the running. There are, after all, legitimate questions around the use of AI in regulated financial services such as the risk created by hallucination, explainability, data security and regulatory compliance. Governance on the platform is therefore critical. 

“Explainability is mandatory,” says Bierry. 

Every action and decision can be traced. Bierry says bank executives routinely tell SBS they want to exploit AI but do not want to find themselves ‘front page [of] a newspaper because of a hallucination or another problem’. And that, he believes, is encouraging the market to separate into ‘two worlds’: institutions accelerating into

AI and the ‘wait and sees’.

The latter are effectively betting that technology will become sufficiently mature that they can catch up quickly later. But Bierry questions that assumption because the changes taking place at Scottish Building Society go way beyond a technology installation programme.

“What we have seen there is that the learning curve on the ability to prompt on the system is also growing fast,” he says. “The need is not only a question of technology. It’s also about an organisation building its experience linked to using the technology as it evolves very rapidly.”

Employees are learning what to ask, what is possible and where AI creates genuine value.

Banking’s two AI worlds

Scottish Building Society is an intriguing test case, and it demonstrates that while large incumbents undoubtedly have greater resources, those advantages can reduce the urgency for change. As Bierry puts it, mortgage-heavy institutions can already see significant portions of their margins several years into the future: “So, they have to put themselves a bit in danger to move and drive the future growth.”

A smaller organisation that’s prepared to do that could therefore find it’s punching well above its weight. In fact, Bierry believes institutions able to make the transition over the next four, five or six years could dramatically increase pressure on those that cannot.

“The ones able to shift like this… they will put the others on the menu,” he says. “There will be a consolidation.” 

It’s far too early to know how much of the first-mover advantage Scottish Building Society’s experiment will reap. But what it’s already demonstrating is that AI leadership need not be determined by balance-sheet size. It may instead depend upon appetite, organisational agility and a willingness to learn while others wait.