SBS builds the core banking, lending and digital banking platforms that more than 1,500 financial institutions run on, across 80 countries. Rather than sell AI as a separate product for those institutions to integrate, it has put it inside the systems they already use, drawing on data that is already governed and secured within each institution’s own environment.
Eric Bierry, chief executive of SBS, on why bank AI pilots succeed where rollouts fail, and what has to change before the full rollout in 2027.
The Fintech Times put seven written questions to Eric Bierry, chief executive of SBS, on why so few banks have taken AI past a pilot, what a realistic path to enterprise rollout looks like inside a regulated institution, where the technology is still being oversold, and what happens between now and the full platform rollout planned for early 2027.
1. Only around a third of banks have managed to scale AI for even a single core process, despite heavy investment. What is actually stopping them?
The technology isn’t really the problem, and I think that surprises people when I say it. What stops banks the most is their data. These institutions are sitting on enormous amounts of data spread across systems that were never designed to talk to each other. On top of this, banks are working with compliance requirements that make moving data around complicated, and nobody has had the time or mandate to sort it out properly.
When you try to run AI across an institution that doesn’t have organised data, you hit a wall almost immediately. The pilot looks great because someone curated what went into it, but then the large-scale rollout fails because the underlying data is a mess. That's the pattern we see again and again, and it’s why we spent years getting the data foundation right before we launched the AI foundation.
2. SBS has built AI directly into the core banking, lending and digital banking platforms institutions already run. Why take that route rather than offering AI as a separate product to integrate?
We’ve watched banks buy AI products, and we know that it tends to end poorly. Typically, when adding a new product, there are eighteen months of integration work and somewhere down the line, in months to years, someone is maintaining an expensive product that nobody quite understands anymore. We didn’t want our solution to have the same fate.
Our clients have been running on SBS systems for a long time. They trust those systems, their people know them, and critically, their data is already inside them. So we thought, what if we just made those systems smarter? It sounds almost too straightforward when you say it, but it changes everything about adoption. You’re not asking a compliance officer to learn a new platform. You’re just making the one they already use give them better information.
3. How does drawing on data that is already governed and secured within each institution’s own environment change the regulatory conversation around AI?
When a bank sits down with a regulator, and the regulator asks where the bank’s data is stored and who controls it, the answer matters. If data is going to a third-party system, even temporarily, you are opening a whole set of questions about ownership and control.
When it stays within the institution's own environment, you can show them exactly what the model looked at, why it produced the output it did, and what the audit trail looks like. That’s what makes deployment possible for banks under strict compliance requirements. Historically, a lot of banks have been stuck because they couldn’t get comfortable with that answer.
4. What does a realistic path from AI pilot to enterprise rollout look like inside a regulated institution, and how long should it take?
If you’re starting from zero on your data foundation, it takes years. There’s no clever shortcut to getting fragmented institutional data unified, governed and trustworthy enough to support AI decisions. What we’ve done is take that problem off the table for our clients, because the foundation is already built within SBS infrastructure.
Even with that infrastructure, though, banks still need to validate the use cases carefully. For example, nobody should be putting AI-generated insights in front of a lending decision or a regulatory filing without being very confident about being able to explain them. But this work shifts from taking years to taking months, which is a significant difference for clients who have been waiting a long time to demonstrate return on their AI investments.
5. Where is AI delivering measurable value in core banking today, and where is it still overpromised?
The places where it works best are driving efficiency and better decision-making. For example, an executive who used to wait five days for a data report is getting that answer in minutes now. Another example is a compliance team cutting down the two-day process to assemble evidence into a couple of hours instead. These are not the flashy use cases that make for exciting conference panels, but they add up to something significant over time.
I think the industry is still getting ahead of itself around autonomous decision-making. Credit decisions, complex client relationships, anything where judgment and accountability matter. AI is not replacing that anytime soon in banking.
6. What should a bank’s technology leadership be doing in the next twelve months to be ready?
Stop agonising over which model to use because the model is almost never the constraint. What I see holding banks back, consistently, is their data. It’s ungoverned, it’s fragmented, it’s sitting in systems nobody has touched in fifteen years. Technology leaders who spend the next twelve months getting serious about that, actually doing the work, are going to be in a completely different position when AI capabilities mature further. The ones who don’t are going to keep watching their pilots stall and wondering why.
And it is not about 100 per cent of their data, rather it is about the data that is used by AI and impacts the outcome of AI. This is a very different way of thinking from traditional approaches to data within a financial services organisation.
7. What comes next for SBS AI across the 1,500 plus institutions on your platforms?
The full rollout is planned for early 2027, and what happens between now and then matters a lot. We have clients already using this, and we’re learning from them constantly. We are paying attention to what’s landing, what isn’t, and where the assumptions we made in development turned out to be wrong.
Those lessons are shaping what the platform looks like when it reaches everyone. One exciting aspect of our platform is that it’s going to look slightly different for every client using it. Even though a UK building society and a global auto finance company are both underwriting risk, their workflows look different, and the ways that they interact with AI will likely look different as well. And both of those will look very different from how a multinational bank uses it.
Getting that specificity right, rather than offering a generic capability and hoping it fits, is where I think we can do something that nobody else in this market is positioned to do.
The post SBS Chief: bank AI Pilots Stall on Data, not on the model appeared first on The Fintech Times.
