AI is changing banking. Is your architecture ready?

By Alexey Lapusta, Head of Solutions (FDE, DevEx & Agentic AI) at Plumery

Banking has undergone two major shifts in how customers interact with their financial institution. Internet banking reduced reliance on branches, while mobile banking put everyday digital services in pockets, raising expectations around speed, convenience and access.

Now AI is creating another shift. But meeting that demand is not simply a question of adding an AI assistant to a digital channel. As AI moves from answering questions to recommending products, initiating journeys and completing tasks, the underlying technology becomes increasingly important.

An AI assistant might explain how to open an account. But if systems cannot initiate and complete the journey, the experience stops when the customer expects action.

AI is putting the spotlight on the architecture underneath

The first wave of AI in banking has focused on chatbots, copilots and conversational assistants. These can help customers find information, compare products and navigate digital channels, but they are only part of the opportunity.

The next step is AI that can interact with banking systems and carry out tasks on a customer’s behalf. Opening an account, making a payment, applying for a product or setting a savings goal require AI to interact with customer data, business rules and operational systems. This is where legacy architecture becomes a constraint.

Furthermore, IBS Intelligence estimates that seven in 10 Tier 1 banks still operate legacy core systems. These systems may continue to perform their role as systems of record, but banks need architecture around them that supports faster innovation, integration and digital experiences.

This does not necessarily mean replacing the core. It means creating flexibility to expose capabilities within and around it to the channels, applications and intelligent systems customers increasingly expect to use.

The architecture behind the digital experience

For banks modernising their technology estates, several architectural principles are increasingly important.

• API-first and modular: Banking capabilities, data and business rules need to be accessible through standard interfaces. Web, mobile, open banking and AI agents should access the same underlying capabilities.
• Reusable banking journeys: Onboarding, payments, lending and servicing should be defined as reusable capabilities rather than rebuilt for every channel. Digital journeys should use the same underlying logic and data wherever they are accessed.
• Integration and orchestration: Modern digital experiences depend on multiple systems working together. Banks need architecture capable of connecting core platforms, customer data, third-party services and business processes without creating point-to-point integrations.
• Data and context: AI can only deliver useful experiences when it has access to accurate, timely and governed information. Deloitte found that more than nine in 10 users said the data they need is often unavailable or takes too long to retrieve, while eight in 10 identified data quality as a challenge.

These foundations matter as the cost of maintaining outdated architecture grows. Accenture estimates that around 70% of banks’ IT budgets are consumed by maintaining technical debt, while software costs have grown by around 8% annually since 2017, faster than banking revenues.

From digital channels to adaptable banking

Banks should not measure digital transformation by the number of new interfaces or AI features they launch. The more important question is whether their architecture allows new experiences to be introduced without rebuilding the foundations each time. The branch was once the primary interface between a customer and their bank. Internet and mobile banking changed that relationship. AI could change it again by making the interface less important than the customer’s intent. Customers could ask for an outcome and expect the bank’s technology to make it happen. Banks cannot predict what those experiences will look like. They can, however, build architecture flexible enough to support them. Modernisation is not solely about replacing legacy technology. It is about creating an adaptable layer between the systems that run the bank and customer experiences.

As AI accelerates change, adaptable architecture will allow new capabilities to be connected, reused and deployed without repeatedly rebuilding the digital experience.