Will Shoppers Trust AI Agents With Their Wallets? Some Companies Are Betting They Will

AI agents are taking on online shopping, with tech providers beginning to accommodate agent-led transactions.

Earlier this month, Stripe, a digital payment provider used by over 200 million users, announced it would be giving AI agents the ability to pay with Link’s wallet. The release will enable consumers to use personal agents like OpenClaw to make purchases on their behalf.

OpenAI, on its end, launched its agentic commerce protocol. And last year Perplexity introduced its conversational search shopping feature. Visa and Experian have now also unveiled Agent Trust, a security framework which aims to link human identities to agents and prevent fraud.

These rollouts come as consumers are beginning to turn to AI for their shopping experience. According to an Adobe survey conducted last year, more than half (53%) of US consumers use AI to conduct shopping research.

Visa’s President of Technology On Agentic Commerce

All of these product rollouts are occurring as part of a shift where interactions across the internet become increasingly machine-to-machine, rather than human-to-machine. Instead of shopping manually across multiple merchants, users can spin up an agent to go to multiple stores and shop for items on their behalf. It’s agentic commerce.

“I think this is one of the most significant changes to ever dawn in commerce,” Rajat Taneja, president of Technology at Visa told International Business Times in a video interview. He went on to say people are moving away from the era of trying to find and decide what to buy, to letting autonomous intelligence help discover things.

“A lot of the agentic work in commerce is more assisted, but consumers are using it now in an increasingly larger way to discover, to compare, to decide what to buy, and it is becoming, and it will become even more part of their journeys,” Taneja added.

However, he does note some barriers to wider adoption. “Recommending is easy in some ways, especially with the intelligence we now have with the newer models. Spending money is different, and that I think is still in the process in the very early stages of being developed because in order for us to move from discovery and recommendations to spending and outsourcing to an agent, trust and infrastructure matters.”

Taneja says that telling how agentic commerce will evolve is hard to predict, but notes that in order for this evolution to take place, the foundation of commerce has to be rock solid. He also says that Visa is “dedicating” its work “to ensure that the things that enable commerce to flourish today can evolve in this new world of more autonomous agent commerce.”

It’s not just Visa noting this shift in the B2C market. Futurum research highlights a broader evolution, noting that AWS has introduced “Agent Mode” to allow agents to interact with the marketplace’s data layer, Salesforce launching Agentforce to allow agents to share context and execute tasks, and Google Cloud’s dedicated AI Agent Marketplace integrated with Vertex AI.

The Trust Gap

Trust is the biggest factor that will determine the future of agentic commerce. At this stage, while many organizations are rolling out the infrastructure to support agentic transactions, adoption remains low.

A survey conducted by YouGov in 2025 found that just one in four (26%) of Americans say they trust AI in retail, while 33% say they don’t. In addition, while 65% trust AI to compare prices, just 14% trust it to place orders on their behalf. This research points to a significant trust gap that will halt adoption for the foreseeable future.

Users not only have to be able to trust the agent is secure, but also that it has the guardrails to follow their instructions precisely. After all, an agent misinterpreting user instructions could result in unwanted purchases.

There are also security concerns. One of the most significant is that of prompt injection attacks, where a hacker deploys malicious instructions to an agent, which instructs it to perform unintended actions. They could provide a mechanism to make fraudulent purchases.

“The same technologies, the AI intelligence that are going to power smart commerce, autonomous commerce, will also power smart crime,” Taneja said. “And so, as that gets industrialized, what we are focused on is to move faster as fraud goes from human pace to machine speed, the defenses and the expertise to combat that has to move even faster.”

Will Agentic Commerce Remain Experimental?

Agentic commerce could remain an experimental solution category if the trust gap isn’t addressed.

Although addressing these trust challenges is a tall task, Taneja pointed out that even with e-commerce consumers were initially nervous about putting their credit card details on websites and making purchases. Fraud prevention measures, authorizing only legitimate transactions, played a critical role in increasing consumer confidence in the momentum of e-commerce.