The battle over the AI-led future of banking has begun

The writer is founder and former chief executive of Starling Bank

When I came up with the idea for a new bank 12 years ago, the response from the industry followed a predictable pattern. Those that didn’t ignore me laughed at the audacity of the idea of a mobile-first bank. The laughter turned to scepticism as customer numbers began to grow and then, finally, to worry as tens of thousands began to flock to so-called neobanks.

Over time, Starling, Monzo and Revolut raised the game for all customers because traditional banks raised their game, incorporating all the features the neobanks had pioneered.

The scale of the neobank development was underlined late last month when it emerged that Brazil’s Nubank had approached Monzo about a deal that could value the British bank at £10bn. Early-stage talks with Nubank, which boasts a $65bn valuation and 140mn customers across the Americas, did not progress. The Brazilian bank subsequently issued a statement saying it was not pursuing a deal. But the valuations pointed to the scale of the neobank industry.

What we need to look at now is the next phase in the development of banking. It won’t involve putting a better bank on your phone and won’t even involve an app. A very able AI agent is almost certain to step into the gap between banks and their customers, making all this irrelevant.

As ever, though, the sector is looking the wrong way. Today, when banks talk about adopting AI, they are mostly talking about efficiency and trimming costs. Depending on your perspective, it will improve productivity or hollow out a call centre. What banks should be thinking about is agents because the new world is being built around agentic AI. Agents can research products, compare prices, organise our diaries, send emails, book travel and perform multi-stage tasks on our behalf. Anyone who has bought into this new lifestyle will not be going back to the old way of doing things. Customers are embracing agents at a rapid rate too. Meta’s Muse, aka a “human concierge”, secured more than 6.6mn downloads and topped the US app charts within two weeks of launching on September 8.

There is, however, an issue to solve. A friend recently asked Muse to get her some vanilla ice cream. It found the produce through Deliveroo and then sent her a link to complete the payment. That’s not an autonomous agentic economy. It’s a marginally faster human economy with extra steps. Customers need agentic AI that knows their preferences, operates within the permissions we set and completes the purchase for them. We won’t need to open our Deliveroo apps, or our banking apps, or any app again. The phone screen full of little squares may soon look as quaint as the BlackBerry.

This will have profound consequences for banks. Banks, indeed the entire payments industry, have spent years pursuing every barrier possible to stop the bots. All banks have added in layer after layer of protection to ensure any payment activity is prompted by the user and not a fraudster or an automated attacker. The central point of the solution we’ve all adopted has been to insist on a human presence.

There is also the issue of scale. Human beings sleep, but agents do not. An agent managing even a small business might perform thousands of transactions across multiple merchants and currencies. Our current fraud monitoring systems, which have been carefully calibrated around human behaviour, will brand everything around that activity as suspicious.

Recommended

Almost overnight, we find ourselves in a situation where some account holders would be quite happy for a non-human to act on their behalf for certain purchases. This is not going to go away because the banking industry is slow to address it or laughs at the very idea. Just as happened with Starling, someone will find a way to show that, yes, this is what customers want.

For banks to remain relevant, whether traditional or neobanks, they need to begin working with the AI businesses that could protect, or replace, their future revenue. They face a delicate balancing act. Britain’s AI minister Kanishka Narayan warned this week that cyber attacks on the UK by rogue AI agents were “likely, if not inevitable”. Banks will be an obvious target. They must keep out the bad players while welcoming those authorised to manage it for their customers.

This sector needs an AI initiative capable of bringing banks, regulators and companies such as OpenAI and Anthropic to the same table. Making banking better is no longer the story. The battle for what form banking will take in the future has already begun.