For most of the mobile-banking era, the strategic assumption was simple: if a bank could persuade customers to open its app frequently, it could deepen the relationship. The app became the branch in a pocket - the place where a customer checked balances, moved money, applied for credit, bought investments, changed a card PIN and increasingly received financial guidance.
That assumption is beginning to weaken. A growing share of financial activity can now be initiated from outside the bank-owned interface. Open-banking services can aggregate accounts and initiate payments. Digital wallets can become the point of payment choice. Merchants can embed finance into checkout. And a new generation of AI assistants may eventually act as financial proxies: comparing offers, moving cash, paying bills or initiating transactions without requiring the customer to open a bank app at all.
The shift is no longer purely theoretical. In the UK, Open Banking Limited reported in July 2026 that open banking supported more than 19 million active user connections and over 40 million payments each month. Its API statistics also show that the infrastructure is now operating at billions of calls per month. Those figures measure connections rather than unique individuals, but they still illustrate how quickly bank data and payment functions are moving into third-party journeys.
The strategic question is therefore not whether banking apps disappear. They almost certainly will not. It is whether the app remains the primary gateway through which customers experience their financial lives. If that gateway weakens, banks may still hold the regulated account and the deposits while losing part of the customer attention, product discovery and decision-making that once came with them.
The banking app is becoming one interface among many
The mobile app was powerful because it bundled identity, data, payments and product distribution in one controlled environment. A bank knew who the customer was, could see what they were doing, could surface a relevant product and could complete a transaction without handing the customer to another provider. That combination made the app commercially valuable even when individual digital transactions were cheap to serve.
Open banking breaks that bundle apart. With customer permission, regulated third parties can access account information or initiate payments through standardised interfaces. In practice, that means a consumer can manage multiple accounts from a budgeting service, make an account-to-account payment from a merchant journey or use financial data in a lending application without first navigating through each bank separately.
The UK data show the scale of this disaggregation. Open Banking Limited reported for June 2026 around 2.81 billion successful API calls, 40.16 million payments and 18.81 million user connections. The organisation cautions that connections are not the same as unique people because one individual can connect through more than one brand. Even with that qualification, the direction is clear: core banking capabilities increasingly travel to where the customer is, rather than requiring the customer to come to the bank.
This matters because digital distribution is not merely a service channel. It is also an information advantage. The firm controlling the interface can see what the customer is trying to achieve, what alternatives are being considered and when a financial need arises. If that context sits increasingly with a wallet, marketplace, AI assistant or specialist fintech, a bank can become excellent at manufacturing regulated products while becoming less visible at the moment those products are chosen.
Agentic AI raises the stakes
Open banking allows data and payments to leave the app. Agentic AI could go further by shifting financial decision-making itself away from the app. The distinction is important. A conventional chatbot helps a customer understand an option. An autonomous or semi-autonomous agent can be authorised to act within boundaries: monitor balances, compare savings rates, optimise bill payments, search for insurance, adjust cash allocations or initiate a payment after predefined checks.
The UK Financial Conduct Authority made this possibility explicit in its July 2026 Mills Review. The regulator said AI could transform consumer journeys and reported research suggesting that one fifth of surveyed UK adults were likely to use AI that can act autonomously within pre-set goals. The FCA framed the longer-term development as a movement from assistance toward delegation, while stressing explainability, fairness, resilience and accountability.
If delegation becomes normal, the customer may interact primarily with an AI layer that sits above several financial institutions. The customer might ask for an outcome - keep £3,000 liquid, maximise interest subject to deposit protection, pay recurring bills from the cheapest source of funds, or find a better mortgage - and allow the system to orchestrate the underlying actions. The bank app becomes one execution endpoint among several.
The infrastructure is already being tested in narrower settings. The FCA said in July 2026 that its second Supercharged Sandbox cohort would explore safer agent-led payments and commerce, alongside AI governance, fraud prevention and financial inclusion. Separately, BIS researchers have shown in simulation that a general-purpose AI agent can perform several intraday liquidity-management tasks, while emphasising the need for safeguards, human oversight and further research.
The real contest is for the customer relationship
Banks often describe the current account as the anchor relationship because it captures salary inflows, recurring payments and day-to-day behaviour. In the app era, that relationship also generated attention. A customer who checked a balance several times a week repeatedly entered the bank’s digital storefront.
An intermediary layer changes the economics. The deposit account may remain at Bank A, a credit card at Bank B, investments at Platform C and insurance at Provider D, while a separate interface becomes the place the customer actually visits. That interface can decide which balance is shown first, which product is compared, which provider receives the next payment instruction and potentially which institution gets the next deposit.
For incumbent banks, this creates a version of the old distribution-versus-manufacturing dilemma. Institutions with strong balance sheets, licences, risk systems and low-cost funding can remain essential even if the customer relationship migrates elsewhere. But margins can come under pressure if another platform controls product comparison and customer acquisition. A bank that once cross-sold inside its own app may have to win the same customer repeatedly in an external marketplace.
For fintechs, the opportunity is the reverse. They may not need to become full-service banks if they can own the orchestration layer. The most valuable position could be the service that understands the user’s financial intent and then calls regulated providers underneath. But that role also attracts responsibility: consent management, security, operational resilience, customer redress and the quality of any automated recommendation become central rather than peripheral.
Payments could move first
Payments are the most plausible starting point because customers already expect payment initiation to occur inside merchant, wallet and platform experiences. In many markets, the banking app is visible mainly as an authentication step. As open-banking payments mature and agent-led commerce develops, even that step may become less prominent, although strong authentication and clear authorisation will remain essential.
The Bank of England’s July 2026 Financial Stability Report notes that more autonomous AI systems are becoming an area of rapid innovation in payments. It says they could change how payment decisions are initiated, optimised and carried out, while raising questions about authorisation, traceability, fraud detection, liability, legal accountability, resilience and liquidity management. That list is a useful reminder that removing the visible banking interface does not remove banking risk.
The same principle applies to wholesale finance. BIS Project Agorá has demonstrated a prototype combining tokenised commercial bank deposits with tokenised central-bank reserves on a programmable platform for cross-border settlement. The project is not a retail AI service, but it illustrates a broader architectural direction: financial functions can increasingly be called through shared, programmable infrastructure rather than being tied to a single institution’s front end.
Banks still possess advantages that interfaces do not
A future with less app dependence should not be confused with a future without banks. Banks retain advantages that are difficult for an interface provider to replicate: regulated deposit-taking, access to payment infrastructure, credit underwriting, capital and liquidity management, compliance systems, customer protection obligations and, critically, trust in the safekeeping of money.
The European Central Bank’s work on a possible digital euro makes the point in another context. In 2026 the ECB stressed that supervised payment-service providers - including banks - would be central to distribution and customer relationships, with safeguards intended to prevent excessive disintermediation. The ECB’s design work explicitly places banks and other supervised PSPs at the core of the customer-facing model, rather than assuming that public digital money would eliminate private intermediaries.
The implication is that banks can remain foundational even when they are less visible. But being foundational and being commercially powerful are not the same thing. Telecom networks can be indispensable while consumer value accrues to services built on top of them. Banks will want to avoid an equivalent outcome in which they carry the balance-sheet, compliance and resilience burden while another layer captures the data, engagement and pricing power.
The hidden danger: commoditisation of deposits and products
A customer-controlled AI agent could make switching and comparison much more frequent. Today, many consumers leave cash in suboptimal accounts because changing provider requires time, attention and paperwork. If an agent can continuously compare rates, eligibility, fees and protections, inertia may fall. Deposits could become more price-sensitive and potentially more mobile.
That outcome is not guaranteed. Consumers may place substantial value on stability, brand familiarity and the simplicity of keeping money in one place. Deposit insurance limits, tax treatment, product conditions and fraud concerns also complicate automated movement. But even a modest reduction in inertia could matter for banks whose economics depend on sticky, low-cost deposits.
The same effect could extend to credit cards, personal loans, savings products and insurance. If an agent can evaluate offers across providers at the point of need, product quality and price become more transparent. Banks may benefit where they are genuinely competitive, but the cross-subsidies and behavioural frictions that support some retail models could become harder to sustain.
Trust may shift from the app to the permission layer
The most important design question is therefore not simply who owns the screen. It is who the customer trusts to act. In app-centric banking, trust is concentrated in the bank. In an agent-mediated system, trust is distributed among the bank, the AI provider, the data-sharing framework, the authentication mechanism and the rules governing consent and liability.
This makes permission architecture critical. A customer should be able to understand what an agent can access, what it can do, the limits on transaction size or product type, how long authority lasts and how that authority can be revoked. Banks will need machine-readable controls that are much more granular than a binary login. Regulators will need to determine where accountability sits when a delegated system makes a poor decision or executes an unwanted transaction.
The FCA’s Mills Review emphasises exactly these concerns, highlighting the need for reliability, consistency, explainability, accountability and human oversight as AI moves toward greater delegation. The Bank of England similarly warns that probabilistic AI behaviour can sit uneasily with payment systems that require predictable and legally certain outcomes.
Why the banking app will not simply disappear
There is a strong counterargument to the “post-app” thesis: customers still need a trusted place for high-stakes activity. Opening an account, reporting fraud, changing security settings, resolving a disputed payment, applying for a mortgage or reviewing an investment portfolio can require explanation, documentation and reassurance that do not fit neatly into invisible finance.
Banks are also improving their apps rather than abandoning them. Mobile channels increasingly combine servicing, advice, messaging, identity verification and security. For many institutions, the strategic response will be to make the app better while simultaneously making the bank accessible through APIs, wallets and agentic interfaces. In other words, the winning model may be “app plus everywhere”, not “app versus everywhere”.
There are also limits to delegation. Consumers may be comfortable allowing an agent to move £50 between accounts but not to refinance a mortgage or sell long-term investments. Regulators may require additional friction for high-impact decisions. Providers may deliberately keep certain actions inside authenticated bank environments because the risk of fraud or misunderstanding is too high.
What banks should optimise for in a post-app world
If the customer interface fragments, banks need to rethink what makes a primary relationship primary. Login frequency becomes a weaker metric. More durable measures may include the share of a customer’s deposits, salary inflows, payment credentials, lending relationship, identity credentials, financial permissions and recurring financial decisions that continue to route through the institution.
Banks will also need to compete as high-quality infrastructure providers. That means reliable APIs, fast consent flows, clear machine-readable product information, robust identity and authorisation controls, resilient real-time payment capability and systems that can distinguish a legitimate delegated action from fraud. The institution that is easiest for trusted agents to use safely may gain distribution even if the customer rarely opens its app.
At the same time, banks should protect the moments where human judgement and institutional trust matter most. Fraud recovery, complex lending, vulnerable-customer support, financial distress and major life events are likely to remain areas where customers value direct accountability. A bank that becomes invisible during routine transactions but highly visible when something goes wrong could still build a powerful relationship.
Implications for the wider financial system
For regulators, the central challenge is preserving consumer protection when the service chain becomes longer. A single customer action may involve an AI assistant, a data intermediary, an identity provider, a bank, a payment network and a merchant. Rules designed around bilateral bank-customer interactions will need to work across that chain without creating uncertainty over who is responsible.
For fintechs and technology companies, the opportunity is large but so is the burden of proof. Winning the interface means being trusted with intent, identity and authority. Any provider seeking that position will need to demonstrate security, explainability, operational resilience and clear boundaries around commercial incentives. A personal financial agent that quietly steers users toward the products paying it the most would quickly invite regulatory scrutiny.
For investors, the transition changes how retail-banking franchises should be assessed. Strong digital engagement remains valuable, but app usage alone may become less informative. The more important questions are whether a bank owns durable funding, can distribute through external channels without losing economics, has permissioned access to high-quality data, and can remain the trusted regulated institution behind increasingly invisible financial journeys.
Conclusion: banking may become less visible, not less important
The decline of the banking app as the default front door would not mark the end of digital banking. It would mark the next phase of it. Banking functions would become more ambient: embedded in merchants, wallets, workplace tools, accounting systems and eventually AI agents that act on a customer’s behalf.
In that environment, the strategic prize shifts from owning every interaction to remaining the institution that customers and machines are willing to trust with money, permissions and execution. Some banks will try to preserve the app as the centre of the relationship. Others will become exceptionally good at being present everywhere else. The strongest may do both.
The real risk is not that customers stop using banking services. It is that they keep using them constantly while barely noticing which bank is providing them. That is the point at which a distribution challenge becomes a business-model challenge - and why banks need to prepare for a world in which the customer relationship may survive even when the banking app is no longer where it lives.
References
1. Open Banking Limited, “Open Banking Payments Fraud Monitor – June 2026 Edition” (8 July 2026).
2. Open Banking Limited, “API performance stats” (latest statistics accessed 21 August 2026).
5. Financial Conduct Authority, “Anthropic to support FCA’s Supercharged Sandbox” (22 July 2026).
6. Bank of England, “Financial Stability Report – July 2026.”