By Jo Bevan-Taylor, Head of Business Strategy & Transformation –
Banking & Financial Services, NTT DATA UK
As banks give AI more freedom to act, customers need more than accurate recommendations. They need to understand, challenge and ultimately own the decisions shaping their financial lives.
A banking app that spots an opportunity to save you £200 a month sounds useful. But what if it gets there by recommending that you stop seeing friends, cancel a family holiday or give up a hobby? The numbers might add up, but whether the advice makes sense for your life is another question entirely.
That distinction matters as banks move from AI that explains financial options towards systems capable of taking action on a customer’s behalf. The industry is rightly excited by the potential to make financial services easier to navigate. Yet greater technical capability does not automatically give customers a better experience. Banks must decide what people should be able to delegate, what they need to understand and where the final choice must remain theirs.
There is evidence of that tension already. Experian research found that while 54% of consumers would be comfortable with an AI agent applying for credit on their behalf, only 5% would grant it full autonomy. Separately, the FCA Mills Review found that one in five adults are already open to AI making decisions for them, but concluded that wider adoption will depend on trust, control and access. The findings come from different studies and should not be treated as directly comparable, but both point to the importance of boundaries and control.
What a paper planner can teach a bank
I can’t think of a better example of that in the market now than Agende by Isobel Lorna, a physical planner. Think Filofax but for 2026. Despite a near £300 price tag, its initial stock sold out during pre-sale, generating a reported £250,000 in its first hour and £400,000 within six hours. Nearly 100,000 people had also signed up to its waiting list and just this month Isobel has announced that she has hit £1m in revenue in her first month of business
Its appeal is difficult to explain through functionality alone; cheaper notebooks and digital planning tools already exist. Agende’s real product is its creator, Isobel Lorna. She has built an audience of 2.1 million on a simple idea: save the pennies and the Porsche will buy itself. The brand describes her community as people who want more from life, are willing to work for it and want to enjoy beautiful things along the way. The planner itself is built around that. The paper inserts include a method for planning your finances, and the different insert types can be swapped in and out, allowing you to change your planner as your life changes. At a time when most of us run our lives through phones, she is telling her generation to put them down. Plan the way she does, manage your money the way she does, and you might just be as successful. The plan, and the ambition behind it, belong to the person holding the pen.
One premium product doesn’t tell us how banking customers will behave, but it offers a useful cultural parallel. Even when a tool can organise something for us, we may still value the experience of making the plan ourselves. Active participation is part of the value.
Financial services should take that seriously. People do not save, borrow or invest simply to maximise a number on a screen. They make choices about security, family, independence, enjoyment and the future they hope to build. An AI system can help calculate trade-offs, but it cannot decide which of those ambitions should matter most to an individual.
Design for decisions, not just recommendations
That calls for a more deliberate approach to the relationship between customer and AI. A bank might use AI to identify an unaffordable pattern of spending, explain the consequences of different savings rates or flag an upcoming payment. Each can increase a customer’s understanding and confidence. The problems begin when an apparently helpful suggestion becomes a decision the customer does not realise has been made for them by entrusting AI to take steps on their behalf.
Good design should make the distinction between human advice, AI-generated recommendations and automated actions unmistakable. Customers need to understand the information and assumptions behind recommendations, how these align with their personal goals, and how to question, adjust or reject them.
But the bigger risk may sit outside the bank altogether. A growing shadow advice market is emerging as consumers turn to general-purpose AI tools for financial guidance, often without realising it isn’t regulated advice. The Mills Review found that more than a quarter of UK consumers trust tools such as ChatGPT, Claude and Gemini for financial advice, while FCA research suggests 44% of younger investors mistakenly believe AI-generated financial information is regulated. Rather than relying on warnings alone, banks have an opportunity to make financial planning more personal, interactive and aspirational. Like Agende, their services should help customers work towards meaningful goals and evolve as their lives change. If banks can make that experience genuinely valuable, they give customers a reason to return rather than seek guidance elsewhere.
The competitive advantage is trust
For banks, the temptation will be to compete on how many tasks an AI agent can complete without intervention. That is an incomplete measure of progress and may risk a race to a colder, less personalised interaction with customers. The more useful questions are whether customers understand what is happening, feel confident enough to use the service and retain control when their circumstances or preferences change. This is as much a behavioural and service-design challenge as a technology one.
AI can give people better information and more room to act on it. It can help them understand the financial implications of the lives they want to lead. But it cannot live those lives for them. The banks that earn lasting trust will be those that give customers what pen and paper still offer: a sense of ownership over their plans. AI should strengthen their ability to make informed choices, not quietly make decisions on their behalf.
