UK banks spend around £3.3 billion every year to keep legacy systems running – around a quarter of their total IT budget. About half agree that these platforms struggle to keep up with daily demand and important priorities.
For decades, these costs have been accepted as the price of stability. Critical systems process millions of transactions every day, support regulatory compliance, protect customer trust, and keep cash flowing. Changing them long ago seemed too dangerous, too expensive, or too inconvenient.
The article continues below
SVP, Solutions and Data Tribe Head at Mphasis.
AI accelerates the pace of software change. Tools that accelerate coding, automate testing, and support documentation compress development cycles. Business stakeholders increasingly expect faster delivery, personalized services, and faster decision-making.
Yet while AI capabilities are advancing rapidly, many banking application delivery models have not kept pace. Teams are often tied to manual, disparate processes built for slow time. This gap between what AI makes possible and what legacy environments can truly support becomes impossible to ignore.
Hidden weaknesses of delivery models
Legacy plans are only part of the challenge. The deep problem lies in how to bring change around them.
In most banks, the requirements are not consolidated or defined consistently. Business insights are embedded in different systems, repeated across silos, or interpreted differently by multiple teams. Institutional knowledge often resides with a small group of long-standing experts rather than being drawn from shared, structured assets.
Over time, this fragmentation creates layers of technical and operational debt that are difficult to identify and even more difficult to manage. Introducing AI tools into this environment does not automatically solve these problems. In fact, without clear requirements and harmonized rules, AI can increase it.
The results generated by AI are only as reliable as the input provided. If business logic is unclear or inconsistently defined, AI simply reproduces that inconsistency at scale. Teams may appear to be moving quickly, but releases are always fragile, unpredictable, and highly risky because there is no stable foundation beneath the surface.
The result is a familiar paradox: banks are investing in AI to speed up service delivery, yet the transformation still feels slow, expensive, and fraught with compliance risks. AI has broken delivery; it has revealed how fragile the culture was.
Why estate costs keep rising
The £3.3bn annual maintenance bill is not just a function of outdated technology. It is a symbol of accumulated complexity.
Over the past decades, banks have built dense webs of integrations, custom workarounds, and point solutions around their core platforms. Each regulatory update, product launch, or merger adds new layers.
Documentation quickly becomes obsolete as logic is rewritten to meet urgent needs, while deep system knowledge resides with a shrinking pool of professionals approaching retirement.
In this case, even small changes can create cumulative effects. Test cycles are stretched because teams fear breaking systems they no longer fully understand. Release windows are strictly controlled and surrounded by risk mitigation procedures. Innovation is held back by the need to maintain stability.
AI illuminates this complexity. When leadership teams see what modern, AI-enabled advancements can achieve in less constrained areas, the differences and their plans become apparent. The opportunity cost of maintaining the status quo becomes difficult to justify.
From automation to intelligence led delivery
If AI exposes the limitations of legacy systems and outdated delivery models, the answer isn’t just more automation. What is needed is the delivery of intelligence led.
Intelligence-led delivery embeds business logic and decision context directly into the software development cycle. Instead of letting rules move across code bases or depending on individual interpretations, organizations centralize and compile them into structured, reusable formats.
Requirements are measured and tracked from business objectives to implementation and testing. The logic of the decision becomes transparent, version controlled, and readable.
With this foundation in place, AI operates on a very strong platform. Code generation complies with approved business rules. Testing automatically maps to defined decision paths. Impact analysis becomes more predictable because dependencies are visible and understood.
This reduces reactivity and risk while reducing critical time through precision and control rather than speed alone.
Measuring AI responsibly
For financial institutions, speed cannot come at the expense of robustness, security, or compliance. Intelligence-led delivery puts the decision perspective at the heart of the lifecycle making change manageable, visible, and explainable.
As the adoption of AI increases, regulators are paying more attention to how automated decisions are designed, implemented and governed. Organizations that cannot clearly define how business rules are applied across systems face scrutiny.
Making logic transparent and traceable helps address these concerns while enabling innovation.
This approach also supports talent flexibility. By reducing reliance on third-party knowledge and manual interpretation, teams can focus on higher-value work: refining business logic, improving customer experience and exploring AI-enabled services.
A practical way forward
The transformation does not require extensive changes to critical systems. Few banks can afford that level of disruption. Progress can start with the delivery models themselves.
By identifying areas of high change, codifying business rules, and aligning AI tools with systematic decision management, banks can gradually reduce their reliance on lean processes. Over time, maintenance overhead decreases, release confidence improves, and a realistic path to improved development emerges.
The billions spent on legacy maintenance represent both a burden and an opportunity. Redirecting even a fraction of that investment to intelligence-led processes enables a shift from defensive adjustments to active innovation.
AI is not the enemy of legacy systems or a silver bullet. It is a catalyst, revealing the weaknesses that build up over years of incremental change. Banks that rethink how they deliver software will be well positioned to scale AI safely and sustainably.
Those who don’t may find that the biggest risk lies not in quickly adopting AI, but in sticking to delivery models that can no longer support the pace of modern banking.
We have installed the best AI chatbot for business.
This article was produced as part of TechRadarPro’s Expert Insights channel where we feature the best and brightest minds in the tech industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you would like to contribute find out more here:
