Forget ‘Top Of Wallet’: AI Agents Could Pick Your Payment Method

The way consumers choose payment methods is being examined as artificial intelligence (AI) agents take on more tasks in online shopping. Instead of a person selecting a card at checkout, an AI agent could assess available payment options and make the selection based on the conditions set by the consumer.

A report based on Mastercard’s research with Opinium, titled A Short History of the Future of Shopping and Payments, has revealed that AI is already being used in shopping extensively. The research has found that 51 per cent of teenagers and 46 per cent of parents have used AI to complete a purchase at least once. It has also been found that 67 per cent of parents think their children’s generation will be more comfortable shopping with AI than they are.

From Habit To A Real-Time Choice

This is where Top of wallet, referring to the card a consumer reaches for by default, often because of rewards, brand loyalty or habit, comes into the picture. The report states that now an AI agent could instead evaluate payment options using factors such as interest rates, cashback, foreign transaction fees, fraud liability, and compliance status.

The report has projected that by 2030, AI agents will continuously re-optimise consumers’ wallets based on their objectives, preferences and constraints. These AI agents could use information from earlier transactions and feedback to refine future choices.

The report has also outlined a shift in how consumers may interact with financial services. Instead of making every payment decision themselves, consumers could provide an AI agent with instructions covering spending limits, preferences and the conditions under which it can act.

Agentic Payments Are Already Being Tested

The report’s timeline records that in 2026, Mastercard, Worldline, and ING completed Europe’s first end-to-end agentic payment transaction at Money 20/20. The development forms part of the report’s timeline for how shopping and payments may evolve as AI agents take on more transaction-related tasks.

The report also cites a Gartner projection that by 2028, a majority of business-to-business purchases could be AI-agent intermediated, with more than $15 trillion in spending passing through agent-to-agent exchanges. By 2030, one in five monetary transactions could be programmable.

What Does The Timeline Show

The report has projected that AI agents could negotiate deals in 2027. In 2028, they could review and change the payment options in a consumer’s wallet. By 2029, “payment intent” could become a standard part of financial transactions, recording what a consumer has allowed an AI agent to do, under what conditions and at what price.

By 2030, the report projects that all major price comparison websites could use agentic commerce technology. It also notes that a three-second delay online could put up to 80 per cent of retail sales at risk.

Disclaimer: In India, this is not yet a mainstream payment model; the report’s consumer research covered 13 international markets outside India.