By Rav Hayer, Managing Director for UK & Ireland and Head of European Financial Services Practice at Thoughtworks
Let’s imagine a scenario: a customer opens their banking app. Their AI assistant has already paid three bills, moved £500 into savings based on this month’s surplus, and rebalanced their investment portfolio to account for overnight market shifts. No notifications. No approvals requested. It simply happened.
This is not a far-off concept. It is beginning to happen now in pilot programmes across major financial institutions, like JP Morgan and Barclays. We are entering what I call the “Do It For Me” era, where AI does not just advise or inform; it is beginning to act.
Previously, AI in banking meant chatbots that could check your balance or virtual assistants that could schedule payments. Useful, certainly, but limited. These systems waited for instructions. They responded but rarely initiated.
The question facing financial institutions today is not whether to adopt AI, but something more fundamental. Are they simply using AI, or are they AI-native? The distinction will determine which banks thrive over the next decade and which struggle to keep pace with customer expectations.
When AI is able to start acting
Agentic AI represents a shift in what we expect technology to do. These systems do not wait for commands. They make decisions and complete tasks autonomously on behalf of customers or internal teams.
Take Mastercard’s Agent Pay as an example. This is not a chatbot helping you find a payment option. It is an AI-driven payment system that can autonomously process transactions and complete purchases without human intervention at each step. The friction that once existed, including the mental overhead of managing payments, simply disappears.
In finance, similar developments are already underway. Early pilots are testing AI agents that rebalance portfolios automatically based on market conditions and client preferences, with no approval needed. The system understands risk tolerance, tracks market movements in real time, and then acts when opportunities arise.
What makes this shift significant is not the technology itself, but what it enables. Customers increasingly expect financial services to work like everything else in their digital lives. They do not want to manage their finances. They want their finances managed well.
This expectation is not limited to retail banking. Corporate financial leaders are asking why their systems cannot automatically optimise cash positions across currencies and accounts. Insurance providers are exploring autonomous claims processing that goes beyond assessment to actual settlement. The trend is rising to reduce friction, increase speed, and act on behalf of the customer.
It is important to note this is not just solely about convenience. Autonomous agents reduce manual intervention, which leads to fewer errors and faster processes. They handle repetitive tasks at scale, which frees human teams for work that requires judgement and relationship building. The operational case is as compelling as the customer experience case.
The problem with bolting AI onto legacy systems
This is where many institutions are getting it wrong. The temptation is to treat AI as an add-on. Integrate a chatbot here, deploy a predictive model there, and add some automation to back-office processes. It feels like progress, and it is, incrementally. But the truth is that legacy systems, siloed data and fragmented governance keep AI stuck at the pilot stage, while locking up nearly 80% of IT budgets just to maintain the past.
Incremental add-ons will not sustain the “Do It For Me” era, as you can’t build autonomous agents on systems designed for human-driven processes. AI-native systems are different. They are built from the ground up to support autonomous decision-making. This requires composable architecture that allows rapid deployment of new capabilities and real-time data pipelines rather than batch processing from overnight runs. It also means embedding security and compliance into the core design, not retrofitting them later.
Banks that treat AI as a feature rather than a foundation will face mounting problems. Integration costs will rise as they try to connect incompatible systems. Innovation cycles will slow because changes require coordination across fragmented architecture. It is about having technology that can actually do what the business needs it to do. The next step is not just using AI. It is rethinking the entire operating model so AI can drive outcomes autonomously.
What the future of finance is looking like
Autonomous AI agents will increasingly handle routine banking tasks, from payments to compliance checks and reconciliation, reshaping how banks operate.
I predict this shift will coincide with the rise of tokenised money, particularly stablecoins and tokenised deposits, which provide instant, programmable, 24/7 settlement, perfectly aligned with AI-driven financial decision-making. Customers will experience assistants that anticipate needs, optimise liquidity, and execute transactions in real time across traditional and tokenised rails.
We will see personalised financial ecosystems emerge, where AI agents integrate spending habits and market conditions to manage finances automatically. For banks, this creates both opportunity and challenge. Stablecoins change the economics of deposits, while AI-native systems that can orchestrate across multiple payment rails will capture new sources of value.
Regulators are already asking hard questions about AI-driven decisions and tokenised money. Banks must demonstrate not only that their systems work but also how risks are managed and compliance enforced in real time. Autonomous AI combined with programmable money points to a self-executing future of finance, where institutions that act early can deliver the “Do It For Me” promise securely and at scale.
The challenges leaders are not talking about enough
Regulatory adaptation is necessary, but it will be painful. How do you audit an AI agent’s decision when it is made in milliseconds and based on thousands of data points? Current audit frameworks are not built for this. Financial institutions and regulators will need to develop new approaches together. That process will be slow and at times frustrating for all involved.
Legacy infrastructure presents an even bigger obstacle. Banks still run core operations on systems built decades ago, designed when same-day processing was considered fast. Modernising them is expensive and disruptive. Yet, without modernisation, institutions will struggle to compete with more agile challengers that do not carry this technical debt.
Then there is the talent challenge. Building and managing autonomous AI systems requires people who understand both finance and advanced technology, a rare combination. Institutions need engineers who grasp regulatory requirements, risk managers who can think in algorithms, and strategists who can envision what is possible with autonomous agents. These people exist, but there are not enough of them, and they are not cheap.
Hiring is not the sole solution; building multidisciplinary teams while investing in upskilling existing staff and forming partnerships with technology companies all help in modernisation. Banks that approach this purely as a recruitment problem will struggle.
Importantly, there is trust. Customers need confidence that AI agents will act in their interests, that their money is safe, and that they can intervene when necessary. A single high-profile failure, such as an AI agent making a catastrophic investment decision or processing a fraudulent transaction, could set the industry back years. Building and maintaining trust requires reliable technology and transparent communication about what AI is doing and why.
While these challenges are real and incredibly daunting. They are not reasons to delay. They are reasons to start now, while the stakes are still manageable.
The banks leading the charge in modernisation
Forward-thinking banks understand modernisation. They are investing heavily in it as an essential formation for AI-native operations.
Consider Xapo Bank‘s recent transformation. Facing the typical challenges of blurred system boundaries and congested delivery pipelines, they refactored their infrastructure to a decoupled, value-stream-aligned architecture. The results were striking: 50% reduction in time-to-market, 99% improvement in service restoration, and 98% reduction in change failures. When they needed to deploy a new interest payment feature, their modernised platform made it possible at remarkable speed.
This is what AI-ready infrastructure looks like. Clean data flows. Minimal dependencies. API-first architecture. Real-time processing capabilities. Systems that can scale instantly and integrate new services without friction.
These banks may not yet be running fully autonomous agents, but they are removing every barrier that would prevent them from doing so, building the roads upon which the “Do It For Me” future will ride.
What financial leaders need to do now
Financial institutions need to start prioritising AI-native transformation as a business goal for 2026. With leadership teams allocating resources, setting clear objectives, and embedding AI into the organisation’s DNA and culture.
Engaging in proactive discussions with regulators and customers is also important, as regulators need to understand what is being built and why, while customers need to see the value and understand the safeguards.
Banks that move quickly and correctly will unlock new revenue streams, reduce operational costs, and build these important resilient operations. They will deliver hyper-personalised experiences at scale that strengthen customer loyalty and enable growth that traditional approaches cannot match.
The “Do It For Me” era in finance has arrived and will reshape everything from payments to portfolio management. With the future being AI-native, financial leaders need to ask, are they simply reacting, or are they becoming AI-native?
Author Bio:
Rav Hayer is Managing Director of the UK and Ireland and Head of BFSI Europe at Thoughtworks. With more than 25 years of experience at the intersection of financial services, technology, and business transformation, he has had the opportunity to help organisations navigate significant change. At Thoughtworks, he works closely with clients to shape and deliver innovative technology strategies—whether that’s through digital transformation, AI, cloud, or data-driven innovation.
