Managing a new world of bots and bargain hunters

Imagine the perfect personal assistant. It spends the night haggling over your broadband bill, shifting your savings into a higher-interest bank account and cancelling streaming subscriptions just before their free trials turn into paid plans. By breakfast, it has ordered the anniversary present you almost forgot, resolved a scheduling clash you hadn’t spotted and switched your home insurance to a cheaper provider. All without a tedious form to fill in or a lengthy phone call with sarcastic retorts from a customer representative. The rise of autonomous AI agents could soon put such a personal concierge in everyone’s pocket.

In August, US-based start-up Instinct launched its AI assistant, then Meta followed with Muse last month. Unlike conventional chatbots, which largely answer questions, these agents can take instructions and act on them. They navigate websites, interact with apps and complete tasks independently via smartphones and computers. They are set to upend how the online economy works.

For consumers, their appeal is obvious. The internet has made information plentiful. AI agents can now sift through it, helping households save money, avoid hidden charges and find the best deals. Nonetheless, rapid adoption is far from guaranteed. A recent survey by Mastercard found that 85 per cent of consumers are open to working with an AI agent, but only 10 per cent are willing to let an agent complete a purchase autonomously.

Giving bots access to the personal accounts and data needed to optimise their decisions risks fraud, privacy breaches and costly errors. Some users are limiting their exposure by linking their AI assistants to separate digital wallets rather than their main bank accounts. Indeed, tech companies will have to earn consumers’ trust.

Businesses need persuading too. Last month, Amazon blocked Muse from shopping on its platform. Last week, Citigroup, Bank of America and JPMorgan Chase restricted AI agents’ access to customer accounts. Retailers worry about fraudulent purchases, disputed transactions and costly returns. Banks fear that mass withdrawals could lead to destabilising runs. Also, there are unresolved questions over liability if bots make mistakes, while websites designed to keep automated visitors out must now be able to handle a surge in traffic.

The plumbing of the digital economy will need to evolve. Banks, retailers and tech companies will need common standards to identify agents, grant them permissions and manage automated traffic. Clear rules must establish who is liable if things go wrong. Consumers should be able to set spending limits, approve transactions and revoke access at will. Regulators, meanwhile, must keep pace, particularly in banking, where AI wealth managers risk making retail deposits flightier.

If these hurdles can be overcome, AI agents would provide a welcome boost to consumers’ bargaining power and a much-needed jolt to companies previously reliant on customer inertia. Retailers and banks will have to compete harder to retain them, whether through better deals, loyalty rewards or improved service. Some easy profits may disappear, forcing firms to charge more explicitly for services they once subsidised. But, overall, greater transparency is welcome. Better-informed consumers would make markets more efficient, while fiercer competition should keep businesses on their toes and spur innovation.

With the right safeguards, businesses, bots and bargain hunters can thrive alongside one another. But companies will have to get used to shoppers who can spot hidden fees and pursue refunds without losing patience. The customer service department may soon meet its match.