The banking industry is entering a new phase driven by advances in artificial intelligence, with a particular focus on the role of AI agents in transforming customer interactions. Anne Boden, founder and former chief executive of Starling Bank, outlined how the sector has evolved from skepticism toward mobile-first “neobanks” to widespread acceptance and innovation, but now faces a fresh challenge as AI begins to reshape the landscape.

Twelve years ago, Boden experienced industry dismissiveness when proposing a mobile-first bank, a concept that has since gained significant traction through the emergence of neobanks such as Starling, Monzo, and Revolut. These digital banks have influenced traditional financial institutions to integrate new features, raising customer expectations across the board. The growth and scale of the neobank sector were highlighted recently by reports of early-stage discussions between Britain’s Monzo and Brazil’s Nubank, a firm valued at $65 billion with 140 million customers. Though Nubank denied pursuing a deal, the talks underscored the size and value of the neobank market.

Looking ahead, Boden suggests that the future of banking will move beyond mobile applications and traditional user interfaces to rely on sophisticated AI agents. These agents will act autonomously on behalf of customers by researching products, comparing prices, organizing schedules, managing emails, booking travel, and completing multi-step transactions. The adoption of such agentic AI, she argues, is already accelerating, with examples such as Meta’s Muse—a so-called “human concierge” app—that achieved over 6.6 million downloads and topped US app charts shortly after its launch in early September.

However, significant challenges remain. Current banking and payment systems are designed with the assumption of human involvement in every transaction, deploying multiple layers of fraud prevention calibrated around human behavior. AI agents, which can operate continuously without the limitations of human fatigue, may conduct thousands of transactions across multiple vendors and currencies for even small enterprises. This scale and automation risk triggering fraud detection systems, complicating the integration of AI into banking operations.

Moreover, reliance on AI agents raises security concerns. Britain’s AI minister, Kanishka Narayan, recently warned that cyberattacks by rogue AI agents are “likely, if not inevitable,” with banks positioned as prime targets. Financial institutions must therefore find ways to differentiate between legitimate autonomous agents acting on behalf of customers and malicious actors seeking to exploit vulnerabilities.

Boden emphasizes the need for collaboration among banks, regulators, and AI companies, including OpenAI and Anthropic, to develop frameworks that balance innovation with security. She argues that banks, whether traditional institutions or neobanks, should proactively engage with AI businesses that may either protect or disrupt their future revenue streams.

The ongoing debate centers on how banking will be redefined by agentic AI technologies. While some view AI integration primarily as an efficiency tool to reduce costs and automate back-office functions, Boden points to a broader transformation that could render conventional banking apps obsolete as AI agents become primary intermediaries between customers and financial services. Industry hesitation or skepticism, she warns, will not halt this evolution, as market forces and customer demand are set to drive the adoption of AI agents in everyday financial activities.

In this shifting environment, the critical question for banks is not simply how to “make banking better” but rather how to remain relevant in a future where AI agents play a central role in managing financial interactions. The contest over this future has already begun.