iStock.com/fcafotodigital
May 2, 2025
Will AI Agents Be Trusted at Checkout?
Visa and Mastercard both introduced payment tools that promise to see AI agents not only help shoppers discover products based on their preferences but also make purchases for them as well.
In announcing the launch of Mastercard Agent Pay, Mastercard stated, “This means that for a soon-to-be-30-year-old planning her milestone birthday party, she can now chat with an AI agent to proactively curate a selection of outfits and accessories from local boutiques and online retailers based on her style, the venue’s ambience, and weather forecasts. Based on her preferences and feedback, the intelligent agent can make the purchase, and also recommend the best way to pay, for example using Mastercard One Credential.”
Introducing Visa Intelligent Commerce, Visa said, “Millions of people will soon rely on AI to find the perfect sweater, research a new vacation spot or fulfill their grocery list. Visa will remove friction from the payment itself, making it possible to transact in an AI-driven world in a way that is secure and trusted.”
In both cases, the use of tokenized digital credentials, already commonly used to support e-commerce transactions, will help keep consumers’ credit card data private while providing transparency and supporting trust for banks and merchants.
The payment processors stressed that consumers remain in control, including having the option to stipulate that any final call on purchases be approved by on-device biometrics or another authentication method. For AI-automated purchases, users can set purchase limits, allowed categories, or other conditions.
Visa’s chief product and strategy officer, Jack Forestell, told Fortune that for repetitive, chore-like tasks such as ordering groceries, consumers may want an agent that “just powers through it and automatically goes and does stuff for us.”
For sophisticated search and decision-making capabilities, such as travel booking, consumers may rely on agents with some limits initially. Over time, the agents may gain more autonomy to “go spend up to $1,500 on any airline to get me from A to B,” Forestell said.
PayPal also recently joined Amazon, OpenAI, Google, and Perplexity in introducing AI-shopping tools in what many are terming agentic commerce.
It’s still unknown how open consumers are to using AI even for discovery. KPMG’s Summer 2025 Consumer Pulse survey of more than 1,500 U.S. consumers found only 34% comfortable with companies using GenAI to analyze personal data, while 63% had not used AI shopping tools and don’t plan to.
Discussion Questions
How comfortable will consumers eventually become in letting AI agents manage the shopping journey from discovery through purchase?
Do you see agentic commerce supported by automated shopping transactions impacting a small or large number of purchase categories?
Poll
Loading …BrainTrust
Doug Garnett
President, Protonik
Cathy Hotka
Principal, Cathy Hotka & Associates
Brian Numainville
Principal, The Feedback Group
Recent Discussions
More Discussions
Has Maverik Made a Mistake on its Kum & Go Rebrand?
Maverik has completed its rebranding of Kum & Go, folding the stores into its own banner…Is Guilt Messaging the Path To Influencing Consumer Behavior?
A university study has found that telling consumers their returned items will be “kept out of…How Great of an Advantage Do Low-Price Leaders Like Walmart and Aldi Have?
Low-price leaders like Aldi and Walmart, among others, are vying for an increasingly savvy consumer. How…
People will be very comfortable with allowing agents to do things like searching for the best options, finding the best prices, and even arranging delivery and so forth. They may not be quite so happy in passing over their payment details and integrating this into the process. Consumers like to remain in control and payment companies need to ensure that this is put in place along with safeguards
Neil’s right. This can work only if consumers remain in the driver’s seat.
Despite the convenience offered by artificial intelligence agents, many consumers are concerned about the privacy and security of their data.
AI’s widespread adoption could be hindered by concerns that it might misuse personal information or make biased purchasing decisions. Transparency and reliability will be essential to overcoming these barriers to trust.
It is important to implement clear and transparent data usage policies that ensure users are aware of how their information is handled and protected in order to build consumer trust.
In addition, AI agents can be designed to offer users the ability to customize their privacy settings, giving them greater control over their data. It is also important to regularly audit and update AI systems to prevent biases and improve accuracy so that consumers feel confident that these agents are trustworthy and reliable.
Companies need to be quite wary of big claims about AI. My own experience as a shopper has been that AI agents are worthless for customer experience. AI enabled search looks to be useful — but that’s much different from claims about AI. Companies need to take this one step at a time — and reject the temptation to try to impress shareholders with big claims about new technologies.
There are a couple elements at play here. One is the willingness to allow the AI agent to process a payment for a purchase. We allow all sorts of automatic payments to take place today, so that isn’t really the issue. I think the bigger issue is allowing the AI agent to do research and then make a purchase on our behalf. What is it isn’t exactly as we specified…the wrong size, color, and so on. Once the AI agents can prove to accurately shop for our needs, and this hurdle is resolved, this agentic AI process will likely become more mainstream and accepted. And, of course, there has to be a trusted payment process with the appropriate privacy and security issues for all of this to work.
This sounds like a smarter Alexa. I’m sure consumers will give this AI a try at coordinating outfits and other curated shopping, but when it comes to control of payment details, I am with Neil and Cathy.
Search for me, sure. Find great deals, absolutely. Manage my credit card and make purchases for stuff I didn’t even know I needed? Not a chance in H E double toothpicks. As an example, I would consider the adoption of subscription services that retailers are offering, and those are for known, replenishable items. We love them until that monthly fulfillment order comes to the house, and we realize that we actually haven’t burned through the last two months of orders. Consumers are starting to be far more concerned about privacy. There’s also some very well warranted skepticism about AI. If they can’t get my Wordle answer right, why would I allow it to manage my financial transactions?
I don’t see this becoming a major channel for a long time.
There’s a lot of potential, but trust will determine how far it goes. Shoppers may be open to using it for everyday buys, but for bigger or more personal purchases, brands need to prove the value outweighs any loss of control.
All of this is great—definitely in the future. The question will always be, “Is the public ready?” The answer will only become a “Yes” after the public has tried it and liked it. In the short term, there may be some bumps on the way to the tipping point where the average consumer embraces AI agents.
I go back to this example a lot, but it is applicable. Remember when airlines introduced online reservation booking? This is almost the same. It will take time before the public gets comfortable. Mastercard, Visa, and others who push the technology must find ways to get customers to use it and like it enough to keep using it. That’s the start of knowing this technology is here to stay (at least until something new comes along).