IBM: How Banks can Accelerate Enterprise-Wide AI Adoption

As the excitement surrounding AI in banking begins to subside, the industry is shifting into a phase focused on execution.

Success now hinges on addressing the structural challenges that limit financial services organisations from scaling the technology effectively.

Jessica Ridella, Global Technology Managing Director at IBM, insists the moment has come for ambition to give way to action. Here, she delves into the strategies enabling banks to move AI into enterprise-wide production – and the ones holding them back. 

What is the biggest barrier to scaling AI in banking today?

It's interesting because the biggest barrier today isn’t really the technology itself; it’s about people and processes. Many banks have brilliant AI models, but they're stuck in the 'pilot trap'. The real hurdle is turning a successful experiment into an enterprise-wide capability.

This often comes down to culture and a resistance to change. You’re asking a highly regulated, traditionally cautious industry to embrace a technology that learns and adapts. The banks that are breaking through are the ones that recognise this is as much a change management challenge as it is a technology one. They are focusing on building AI literacy across the entire organisation, not just in the data science team.

Why are trust and governance now critical for AI adoption?

In banking, trust is everything. It’s the currency the entire industry runs on. As AI becomes more powerful and we start giving it more responsibility, we must be absolutely certain that it's operating safely, fairly and transparently. 

Strong governance isn’t just about ticking a compliance box; it’s about building the confidence needed to innovate. It provides the guardrails that allow a bank's leaders to sleep at night, knowing their AI systems are making fair decisions and that they can explain them. Because regulators are already asking, "Can you explain, step-by-step, why your model declined this person's mortgage?" If you don't have a good answer, you have a very big problem.

Without that solid foundation of trust, the most brilliant AI tool will and should, remain on the shelf.

How are banks moving AI from pilots into full production?

Over 90% of banks now see AI as more opportunity than threat – up from 80% in 2024 – and reported productivity gains from AI jumped to 59% from 32% the previous year. The most successful banks are treating it like building a modern factory, but for AI. Instead of building one-off custom projects that are hard to maintain, they are building a central, standardised platform, an ecosystem. 

This 'AI factory' approach provides common tools, pre-approved data sets and a clear process for developing, testing and deploying AI models. It makes the entire process faster, cheaper and safer. It also means that learnings from one part of the bank, say, a model that’s great at detecting fraud, can be quickly shared and reused elsewhere. It’s a shift from thinking in projects to thinking in platforms and it’s the key to achieving scale.

How can banks overcome data silos to deliver hyper-personalisation?

For years, this has been the holy grail of banking and we're finally seeing it become a reality. The prize for getting it right is enormous; according to McKinsey, unlocking personalisation at scale could generate up to $3 trillion in new value for the industry globally. But to capture that prize, banks must solve their oldest problem: getting customer's information from various systems to talk to each other.

The secret isn't to rip everything out and start over. Instead, we help clients build a 'data fabric', a smart layer that sits on top of their existing systems to finally connect the dots. This creates a single, complete picture of the customer without a massive, disruptive data migration.

And once you connect those dots, the relationship changes. Your bank stops being a place that just holds your money and starts being a partner that looks out for you. It means getting a helpful “nudge” to move cash into a higher-yield savings account, or receiving a pre-approved mortgage offer the moment you start searching for houses online.

Why is a hybrid build and buy model now the best approach?

That’s a critical question and it gets to the heart of a major conversation happening in the UK right now. Last year’s landmark Tech Prosperity Deal, which promises billions in US tech investment, represents a significant opportunity. But it has also prompted industry leaders to raise a valid concern: does this risk the UK becoming simply “users, not makers” of AI?

The hybrid model is the definitive answer to that challenge. It’s no longer practical for any single company to build everything from the ground up. The smart strategy is to 'buy' access to the best foundational platforms from global partners to leverage the massive investment and R&D that a deal like this brings.

But that is only the starting point. The real, lasting value is in what you 'build'. By using their own unique data and deep industry knowledge to create proprietary solutions on top of those platforms, UK firms can create something no one else can copy. It’s how we ensure the UK doesn’t just use AI but helps to invent its future.