FUTURE OF FINANCE 2026

What it takes to build the AI-native modern bank

Building an AI-infused engine for innovation, simplification and modernisation will create sustained advantage

Summarise
    • The goal is a radically different model of banking, characterised by highly personalised customer engagement, hyper-responsive product management, and structurally transformed operations.
    • The goal is a radically different model of banking, characterised by highly personalised customer engagement, hyper-responsive product management, and structurally transformed operations. IMAGE: PIXABAY

    COMPETITIVE advantage and performance in banking have long hinged on how effectively institutions use technology and data. Artificial intelligence represents the next step-change, reshaping how banks develop products, manage risk and serve customers.

    The banks that configure their organisations to absorb and exploit each wave of advances in AI will compound their advantage.

    Recent discussions with more than 30 senior executives of technology-savvy global banks converged on a shared view: Building an AI-infused engine for innovation, simplification and modernisation, not simply deploying AI tools, will create sustained advantage.

    Moving beyond efficiency

    The goal is a radically different model of banking, characterised by highly personalised customer engagement, hyper-responsive product management and structurally transformed operations.

    The performance profile could be dramatically different from today’s banking model, aspiring to 10 times greater productivity, 100 times more experimentation throughput, 90 per cent shorter time to market, and a 10 percentage point improvement in the cost-to-income ratio.

    The strategic prize does not centre on efficiency. Rather, it involves velocity in time to market and experimentation. Faster trumps slimmer.

    BT Short Logo

    THE BUSINESS TIMES INSIGHTS

    Powering Asean's Future 2026

    Find out how enterprises are navigating power constraints and sustainability targets

    ACCESS THE INSIGHTS

    Most institutions can envision these performance goals but remain constrained by legacy estates, skills shortages, inadequate data foundations, capital allocation pressures, talent and operating model. Institutions that treat AI as a set of tools layered onto existing operations will underinvest in the changes that create enduring advantage.

    Banks that currently capture the most value from AI treat it as a lever for growth. An AI-native modern bank requires investment and enduring commitment across five elements.

    1. Agentic experiences, products

    The AI-native bank does not just improve existing customer journeys; it reimagines them from the customer’s intended outcome back to what the experience, product, and underlying process should look like when built from scratch. Hyper-personalised experiences will increasingly be orchestrated by intelligent agents rather than delivered through standardised products and predefined channels.

    Agentic customer acquisition and servicing models should be proactive, aware of context, and responsive to a customer’s major life events. They will resolve issues quickly at first contact, rather than routing through multi-step human workflows.

    To that end, several banks have formally abandoned process-mapping exercises, reframing the design question from “How does this work today?” to “What should it look like if we built it now?”.

    2. Real-time, autonomous workflows

    Autonomous workflows are emerging across functional domains ranging from marketing and lending to product operations and beyond. For any bank, a reasonable goal is 80 to 90 per cent autonomous process execution.

    3. Trust and security as a differentiator

    Trust and security in an AI-native bank stem from architectural choices made at the outset. Built well, they enhance a bank’s competitive position rather than serving as compliance requirements retrofitted afterwards.

    Risk and compliance should run continuously as self-executing and predictive features that catch issues in real time. Attaining this state requires treating observability, auditability and resilience as first-order design characteristics from the start.

    Two related organisational changes are emerging. Leading banks have more tightly embedded risk, legal and compliance within delivery teams rather than maintaining them as an end-of-pipeline gate.

    One bank has embedded audit checks directly into AI agents – an architectural feature of the workflow that ensures compliance. On agent governance, another bank automatically shuts down agents after a predefined time frame to force security discipline.

    4. Modern technology, data stack

    There is no viable path to being AI-native without a flexible, modern technology and data stack. Agentic AI cannot orchestrate effectively against fragmented legacy architecture; the strategy of wrapping legacy systems with digital veneers has run its course.

    Most banks face decisions about what belongs in the core, what moves outside it, and whether to push vendors to adapt for AI-native demands or keep the core narrow and build orchestration independently.

    One of the most important architectural decisions for AI-native banks is determining where deterministic systems end and probabilistic systems begin. The deterministic layer covers systems in which regulators require auditability; customers cannot tolerate variable outcomes; and the bank must reconstruct exactly what happened.

    This includes areas such as ledger integrity, payments execution, entitlements and regulatory controls. Agentic experiences, fraud intelligence and workflow orchestration sit in the probabilistic layer.

    Tools and application programming interfaces should behave deterministically; the orchestration layer deciding which to invoke can be probabilistic. Where this holds, the boundary is defensible. But where it blurs, regulatory exposure can accumulate.

    5. Talent, workforce, and operating model

    AI significantly weakens the traditional link between headcount and output. The main scarce resource shifts from capacity to judgment. Teams will be smaller and flatter, with product and engineering roles converging.

    History suggests that efficiency gains through technology often increase demand rather than reducing it, creating new requirements for governance, orchestration and oversight. There are structural implications for how spans and layers evolve, how accountability is redesigned, and how new roles emerge.

    Building an AI-native modern bank

    Defining a vision for the AI-native modern bank is the easy part. What is trickier – and therefore distinguishes leaders from laggards – is ensuring high-quality execution, realising value and building the engine for “always on” innovation, modernisation and simplification.

    While the five elements discussed define the destination, successful execution depends on sharpening capabilities in three areas:

    • High-functioning talent, culture, and operating model: These three components are in the early days of a generational shift with AI. Their importance will only grow in the AI-native era. Building the talent base, culture and operating model will be what sustains advantage over the long term.
    • World-class data assets: Many banks spend too much time and energy choosing models, copilots and the surrounding software, when the more important focus should really be clean, useful data. The AI-native modern bank will be “machine readable”, in which data is standardised, relationships are mapped, policies are encoded and processes are observable. This will require real-time enterprise data fabric, data lineage, high-quality metadata and knowledge graphs, machine-readable process data and a policy layer.
    • Modern, flexible technology platforms: Scaling up agentic AI will require new capabilities for agent orchestration, artificial intelligence for IT operations or AIOps, memory, observability and evaluation. The core deterministic engines that power banking products, ledgers, transactions and decisioning will need to be modern, flexible and interoperable with agent-executed experiences and processes.

    Across every major technology transition in banking, success has depended on sustained senior leadership commitment and sponsorship, and execution speed. AI appears no different.

    Banks building AI-native capabilities today are already beginning to open a lead over those still in planning stages.

    Waiting does not preserve optionality; it cedes it.

    The writers are Steven Breeden, partner; Vitor D’Agnoluzzo, partner (London); Joseph Edwin, partner (Middle East); and Daria Huang, practice senior manager, Bain & Company

    Decoding Asia newsletter: your guide to navigating Asia in a new global order. Sign up here to get Decoding Asia newsletter. Delivered to your inbox. Free.

    Share with us your feedback on BT's products and services