Internet banking ended the need for branch visits. Mobile puts banking in customers’ pockets. Both times, rigid architecture meant costly workarounds and years spent catching up. With AI advancing faster than previous technology shifts, are banks about to make the same mistake again?
This isn’t the first time banking has fundamentally changed how customers interact with their bank. Internet banking meant customers no longer had to visit a branch to manage their money. Mobile banking puts everyday banking in customers’ pockets, raising expectations around speed, convenience and always-on access.
Each shift rewarded banks whose technology could adapt to changing customer behaviour. Those with more rigid architectures often found themselves relying on expensive workarounds, fragmented customer experiences and lengthy transformation programmes just to catch up. Today, I believe AI represents the next major shift in how customers interact with financial services. According to Visa, 86% of banking leaders believe AI will reshape retail banking by 2030, and many institutions are accelerating investment into new AI-powered customer experiences. AI has the potential to transform banking interactions from simple conversations into actions that deliver tangible outcomes for customers.
I’m already seeing this seismic shift. McKinsey found that 57% of customers would consider using a third-party generative AI financial agent if their bank did not offer one, highlighting how AI is increasingly reshaping customer relationships. As expectations shift beyond conversations towards AI-powered actions, I believe banks will need the architecture to support these new experiences. But if AI is just layered on top of existing digital experiences, those expectations will collide with legacy architectural limits. The breakdown in integration would result in incomplete customer experiences, such as an AI agent that can explain how to open an account but not actually open it, or recommend a product but is unable to initiate the journey to access the product.
For banks, the challenge, as I see it, is that simply adding AI to an existing digital experience won’t necessarily meet those evolving expectations. If the underlying architecture can’t support AI carrying out real banking journeys, customers are left with assistants that can answer questions but can’t actually get anything done.
Here is why adding AI isn't the same as becoming AI-native
The first wave of AI in banking has largely focused on chatbots, copilots and conversational assistants that make it easier for customers to find information, compare products and navigate digital channels. These are valuable improvements, but I see them as only the first stage of AI adoption.
The bigger opportunity lies with introducing AI agents that can carry out banking tasks on the customers’ behalf, whether that’s opening an account, making a payment or setting up a savings goal. Delivering these experiences requires AI to interact with banking journeys, customer data and operational rules, which is a key step further than simply generating responses.
This is where I see many banks reach the limits of their current platforms because they are missing the key ingredients to building their AI-native architecture.
Deloitte found that only one-quarter of banking respondents considered their data management platforms highly prepared for generative AI adoption. Without the right foundations, AI remains limited to answering questions rather than completing them.
The right ingredients for AI are in the architecture
While many banking platforms now promote AI capabilities, the real differentiator lies beneath the surface. The architectural foundations that enable banks to deliver seamless experiences across web, mobile and open banking channels are the same ones that will determine whether AI can move beyond answering questions to completing meaningful banking tasks.
Headless and API-first: Every journey, piece of data and business rule is available through an interface. I don’t see channel as a separate product, so it should be seen as another way of accessing the same foundation. Web, mobile, open banking and AI agents all become front ends over the same set of capabilities.
Great DevEx and AgentEx: As agents increasingly write and orchestrate software, platforms need to be as understandable to an agent as they are to a developer. The same journeys, data and rules that developers build against are the ones agents can discover, reason over and extend. Developer experience and agent experience are becoming the same investment.
Journeys as reusable capabilities: Banking journeys such as onboarding, servicing, lending and payments should be defined once as reusable capabilities, rather than rebuilt for every channel. The journey a customer completes in a mobile app is the same one an agent can call, with the same logic and data behind it.
Ownership: SaaS and white-label solutions can help banks launch quickly, but they often mean relying on someone else’s platform and roadmap. To be AI-native, banks need control over their code, journeys and underlying platform. As agents increasingly build software, ownership becomes the difference between building a long-term asset and renting one.
Rethinking what AI success looks like
Banks shouldn’t measure AI progress by how many features they launch or how many AI tools they add. The bigger question is whether their technology is ready to help AI deliver real value for customers.
The next wave of AI will finally move beyond conversational assistants towards systems that can understand context, make recommendations and take action on a customer’s behalf.
The challenge we’re seeing is that banks cannot predict exactly what these experiences will look like or which technologies will shape the future. Instead of building around one specific AI use case, they need foundations that allow them to adapt as new capabilities emerge without repeatedly rebuilding their digital platforms.
AI capabilities are advancing faster than any previous technology shift, which makes the underlying architecture more important than ever. Banks that win in this environment will be the ones whose architecture lets them deploy new AI capabilities quickly, efficiently and securely, creating better customer experiences as expectations continue to evolve.
Alexey Lapusta
Alexey Lapusta is Head of Solutions Engineering, DevEx & Agentic AI at Plumery. With 15+ years of experience in fintech and digital banking, Alexey is a seasoned leader in digital transformation, cloud-native banking, and AI. At Mambu, he helped drive the company’s growth to unicorn status, while at Backbase he worked with leading institutions including HSBC and Citi. Now, as Head of Solutions Engineering, DevEx & Agentic AI at Plumery, he helps banks build and launch next-generation digital banking experiences.
