As digital services became the norm, banks shifted their focus from costly branch networks to mobile apps. Now, AI agents threaten to make those apps redundant by acting directly on behalf of customers, sparing them the trouble of navigating menus and tapping through screens. This invisible service goes beyond convenience— it promises a new era of seamless banking. AI’s real strength lies in its ability to learn and adapt continuously. It processes data from many sources, orchestrates transactions, and detects threats before they become fraud. Faced with this leap forward, banks face a stark choice. They can keep offering outdated interfaces, or develop secure, robust application programming interfaces (APIs) that customer AI agents can use for hands-free finance. Rethinking operations: Klarna and DBS A crucial first step is to use AI internally to remove inefficiencies. Sweden’s Klarna, for example, scrapped some 1,200 third-party software tools and built a new in-house platform on AI and graph databases. By consolidating data sources, it lowered costs and sped up the launch of new features – improvements impossible under fragmented systems. Meanwhile, Singapore’s DBS Group plans to eliminate 4,000 temporary roles over three years, even as it adds 1,000 AI positions. The bank already runs more than 800 AI models for credit underwriting, automated customer prompts and real-time fraud alerts. Its results underscore how AI can shrink costs and speed up delivery. Both examples show that embedding AI at the core of operations brings agility. Staff no longer waste time searching through siloed systems or supervising repetitive tasks. The bank can also stop relying on external software that often drains time and money. Preparing for agent-to-agent banking Greater upheaval lies ahead when customers begin using their own AI agents for everyday finance. A personal assistant might, for instance, be told to move money from a stagnant savings account into a higher-earning investment . Instead of opening an app, the assistant queries the bank, checks rates, schedules the transfer and confirms completion – all without human involvement. Banks must prepare for this model by exposing their systems through secure, well-designed APIs. Ensuring tight security and compliance remains essential, but institutions that succeed will be those whose internal systems can automatically validate and approve requests from external agents. If a bank’s back office cannot handle these agent-to-agent interactions, or if its data are inconsistent, customers will look elsewhere. Building the right ecosystem Shifting from apps to AI agents requires a single source of truth for data, governed properly. Fragmented or outdated records lead to poor AI recommendations and customer errors. A unified data model, combined with role-based access, allows a bank’s own AI tools to mesh with third-party fintech offerings or personal assistants. Many banks fear that their brand visibility will suffer if customers no longer open the apps directly. Yet a bank’s reputation will increasingly turn on reliability and minimal friction. If customers can open a new account or approve a transfer simply by instructing an AI assistant, they are more likely to stay with that provider. Readiness for AI transformation Before mobile apps, bank branches were central to finance – until technology made them optional. Now the same disruption looms for the apps themselves. Agile institutions will profit from lower overheads, faster transactions, and deeper customer loyalty. Banks that fail to adapt may find themselves excluded by advanced AI agents unable to “talk” to outdated systems. By deploying AI at scale for internal operations and offering clear, secure APIs to customers’ personal agents, banks can thrive in this new environment. In an era where services recede into the background, truly autonomous finance is fast becoming the next great leap forward. Dom Monhardt is the founder of one-fs.com, a leading fintech and digital banking newsletter in the Middle East and North Africa (MENA).
