How AI is reshaping the bank–customer relationship in Europe

By Viktoria Danzer, Karen Neuhaus, Matthias Tewes and Nicola Breyer

Ask an AI what to do with your money and it will answer — without having to book an appointment, without bias on fees earned, and without judgement if you expose lack of knowledge or financial planning to date. Millions of people already do this. Each of those exchanges is a conversation a bank used to have, and increasingly does not. As customers turn to AI to interpret, plan and decide their finances, the relationship is moving from the bank to the interface that mediates it, and the economics of that relationship are moving with it. For Europe's banks this is a question of business model, not technology — and the period in which their advantages still convert into a defensible position is narrowing.

For most of the digital era, banks digitised distribution without changing its logic: the bank still decided what to offer, to whom and through which channel, and the customer still came to the bank to transact albeit in the app instead of in the branch.

It is tempting to read this as competition on products and distribution. That undersells what is actually changing. For most of retail banking's history, low switching activity was itself a form of stability — accounts stayed because leaving was inconvenient, comparison was hard or accounts were for free, and most customers had neither the time nor the tools to engage deeply with their finances. That stability is now being tested, not because banks stopped competing, but because the cost of engagement just collapsed. Banks competed at the margin on products and rates, but the franchise rested on a level of customer attention that AI is now making obsolete.

Artificial intelligence changes all of this. It makes engagement effortless and comparison instant, at no cost and with no sense of being judged. The question this poses is not whether a bank should adopt AI for efficiency. It is where financial decisions are now made, who is present when they are made, and who captures the value and trust attached to them. On the current trajectory, the answer is increasingly: not the bank.

Europe's banks are not starting from a weak position: they hold the trust, the regulatory standing and the licence to advise that no AI assistant has. The question this article explores is why that advantage is not yet translating into presence where financial decisions are actually made — and how narrow the window to change that has become.

Where financial decisions are now made

The behavioural shift is already advanced. Survey evidence shows that a substantial share of consumers already use general-purpose AI for financial guidance, and a majority of those who do say it lets them ask questions they would not put to a person. This sets an expectation the conventional bank interaction struggles to meet — organised around product KPIs, and requiring the customer to actively seek contact with an institution they are not used to engaging beyond technical issues.

The bigger adjustment for banks is organisational, not technological. Retail banks have historically been organised, and compensated, around products: loan volume, assets under administration, net interest income. The challengers run on different KPIs — the lifetime value of a customer set against the cost of winning them — which keeps the whole relationship, not the product, in view. Making one function accountable for the whole customer relationship — rather than for the next product sale — is the structural shift this moment calls for, and one several challengers were able to design in from the start.

That gap carried little risk while switching and seeking financial advice were cumbersome, and inertia did the work of retention. The risk increases significantly once an AI agent — able to compare, recommend and act in a single conversation — makes both effortless. Account switching was simplified by law years ago and customers stayed regardless — that persistence measured the cost of leaving, not the strength of the bond.

The point of inheritance illustrates the stakes clearly. As much as €400 billion passes between generations in Germany each year through inheritance and gifts, around €113 billion of it tax-visible in 2024, and this is also where an incumbent's advantages — an existing relationship with the family, knowledge of the estate, regulatory standing — matter most, if activated early enough. The heir generation rarely holds as strong a relationship with the parents' principal bank as the parents do, and is already exploring alternatives: younger customers in Germany consider a neobank for a financial product several times more readily than the boomer generation does. The assumption that children automatically bank as their parents did no longer holds, which makes proactive engagement at this moment a genuine opportunity, not just a defensive one. The demographic wave makes that moment unavoidable, as the population aged seventy and over rises by roughly a fifth over the coming decade.

There is no “one size fits all” answer, but key strategic choices exist

None of this dictates a single response; the right move depends on a bank's scale, capabilities and ambition. Several positions are legitimate, and deliberately chosen. A regional or cooperative institution can win by staying human-first — providing the trusted personal advice that grows scarcer as the market automates. A large institution can build a durable role further back, as the regulated infrastructure others depend on, even as customer-facing interfaces move to agents and platforms.

What is not viable is the full-service retail and wealth franchise that means to keep serving the customer's whole financial life while declining to be present where that life is now managed. The other positions are real businesses: serving a single segment well, or becoming the regulated infrastructure others build on, can be more profitable and more durable than a contested fight to be the customer's agent — and for parts of a large banking group, corporate and institutional banking above all, they are the franchise, not a narrowing of it. But for the retail and wealth business, whose economics rest on holding the customer across needs over a lifetime, none of those positions keeps that relationship intact. There, becoming the customer's trusted financial intelligence — the place they turn to understand and act on their money — is the only position that does. It is the harder path, and the one the franchise is built to take.

Whose model, and whose data

Becoming that trusted financial intelligence raises an obvious question: how does a bank build it? Here the common framing is the wrong one — that this is a decision about the AI model itself. Some banks will train sector-specific or proprietary-data models, and for control, compliance and data residency that can be a sound operational choice. It is a weak strategic one if mistaken for the source of advantage. No European bank will build a frontier model to rival the global technology firms, and a domain model, however good, sits in the layer commoditising fastest — where capabilities converge, each advantage is matched within a cycle, and the cost recurs. Sector-wide, AI spending has so far produced more activity than profit.

The advantage lies a layer up — though not in the way banks tend to assume. A bank does not, today, hold the whole of a customer's financial life: the salary flows out, the investing happens elsewhere, the picture is scattered across providers, as we have seen. But of all the players, the bank is the one best placed to be permitted to assemble it. A general-purpose assistant reasons from what the customer chooses to type into it, and answers with guidance that carries no licence and no liability. The bank can do what the assistant cannot: with the customer's consent and the access open banking already provides, it can normalise the scattered record into a verified, whole view — and give regulated advice on it, advice for which it is accountable. Here the weight of regulation, usually the incumbent's burden, becomes its moat. The model is a commodity input; the standing to assemble the customer's whole picture and the licence to advise on it are not.

The stakes rise further as these systems move from advice to action — initiating payments, moving balances, executing investments on the customer's behalf. The interface then becomes the point where transactions originate, not merely where questions are asked. A bank absent from it does not only lose the conversation; it watches its own balance sheet put to work through an interface it has no part in.

From distribution to relationship

The required shift is, in the end, a shift in business model. For two decades banks optimised the distribution of products. The task now is the opposite: to become the customer's trusted financial intelligence — to make a function within the bank accountable for the whole of a customer's financial life, and to compete on the quality of that understanding rather than the reach of distribution. The revenue logic follows the same path, from product margin towards the value of a sustained advisory relationship — including value that does not convert into a product sale within the same quarter. It is a less comfortable model than the one it replaces, and a more durable one.

The advantages European banks bring to this contest — trust, regulatory standing, and the licence to advise and to act — are real and, in combination, difficult to replicate. But they offer no protection unless they are used, and the period in which they can be converted into a defensible position is narrowing as customers settle into interfaces that are not the bank's. Subsequent articles in this series take each resulting question in turn: how banks segment and serve customers in an AI market; how an institution builds genuine presence inside AI systems; where the line falls between financial information and regulated advice; the data infrastructure any credible offering depends on; and how the same forces reshape corporate and SME banking.

The question facing each bank is no longer whether AI will reshape its market, but how deliberately it shapes its own response. Banks that move now — building on the trust and regulatory standing they already hold — are well placed to become the intelligence their customers turn to, rather than the infrastructure quietly running in the background.