I am a payment expert – this is what happens when you use an agent to buy on your behalf
- Advertisement -
Related Posts
- Advertisement -
One of the biggest differences between being rich and poor is that people do things for you. When a person moves from arm to the middle class, he can get a shon or bring his clothes to a dry cleaner; When they get rich, they can get a driver, a private cook – a whole entourage that focuses on making their lives easier.
There is a growing excitement about the rise of AI -driven agents who act on behalf of consumers – not just during product Discovery, but until the purchase. Search -like questions on chatgpt perhaps only 1% of those on Google, but 1% of the colossal global search market is a large number of searches, with possibly millions of income.
Originally conceived as digital caretakers to simplify the search, these agents now make actual purchases, and they do it without ever giving control back to the shopper.
Big payments Players lay the foundation for AI-based trade. Visa recently launched a digital innovation hub for references to explore new identity models for agent -based transactions, while Streak Confirmed that it develops safe transaction options for AI agents. And yesterday, Google revealed plans for an AI agentic checkout for shopping – a movement that confirms that this shift is no longer speculative, but is imminent.
But beyond the optimistic headlines is a more complicated image. What happens if an AI agent makes a purchase for you? And more importantly, what can go wrong?
Managing Director, PSE Consulting.
What really happens behind the scenes (usually)
Let’s be clear: there is no “standard” for how agent shopping works – the process is still evolving and different platforms take different approaches. That said, here is a common stream that we have observed in early implementations.
When a consumer uses an AI agent for shopping, the process is superficially simple, but technically complicated. Firstly, the user stores their payment card data – including full pan, CVV, expiration date, invoicing and delivery addresses – with their chosen AI platform.
It is unlikely that shoppers buy very cheap items, such as a pizza or very expensive items such as a new car. They will probably not use it for goods with a heavy visual emphasis, whereby some of the pleasure is browsing until something touches you – Clothing is the best example. They are initially likely to use agentic AI to help them decide between relatively expensive products that are difficult to understand for non-experts: let’s use a good pair Bluetooth -headphone As an example.
The agent, who can be powered by Chatgpt, Google, Tiktok Shop, or Amazon‘s AI initiatives, use natural language to respond to the request of a shopper. Just like a shop assistant, it will ask questions to refine the results: how much do you want to spend? Do you want headphones or in-ear? Are there functions such as noise reduction or waterproofing that you need? It can then refine results and present purchase options.
As soon as the shopper decides that a payment process starts, it is usually invisible for the shopper:
- The shopper stays in the AI interface and never visits the site of the trader.
- A “buy” command within the UI agent activates the agent to automatically make the cash register form on the site of the trader.
- The trader receives the entire card details as if a human shopper typed them.
- The agent submits the order and the confirmation is sent via both the agent and the trader.
It is unlikely that the trader knows that they are dealing with an agent instead of a person. This introduces risks, because if something goes wrong – an incorrect item, a delivery mixture or price error – the shopper has to solve it directly with the trader, although they have never had interaction with the trader’s website.
In other words: don’t talk to me – talk to my agent.
Known risks (so far)
Several emerging pitfalls are already clear:
Vulnerabilities of security: In January 2025, the Chinese AI platform Deepseek was hacked, so that the stored references of users were exposed. The centralization of payment data in AI agents makes them lucrative purposes.
Sensitivity to scams: Fraudists can design sites specifically to mislead agents to complete fake checkouts.
Doubleness in Liability: If an agent enters an order out of place or incorrect details, it is unclear whether the AI provider or the consumer shows responsibility.
Poor compatibility:
- It may not support alternative payment types such as PayPal, digital portfolios or bank transfers (which are good for around 45% of e -commerce volume in the EU).
- It cannot easily process additional cash register steps (eg selection of seats, delivery slots).
- Struggle with card decreases, especially with international transactions, where the decrease percentages can be of 5 to 30%.
In markets such as the EU or Japan, legal requirements for strong customer verification (SCA) mean that consumers must approve any card transaction, which means that AI-led flows are problematic or non-compliant.
The bigger picture: are we witnessing a trade revolution?
In addition to the immediate risks and logistics, the rise of agents raises fundamental questions about the structure of digital trade.
Will this model get a grip with consumers? It could bustle like speech trade and Amazon’s Dash buttons, which do not take off because of problems with trust and usability. Or it could explode, just like the rise of market places or in-app mobile buy. The answer depends on how much value consumers put at ease and how well AI agents can overcome trust and control.
If AI agents become the preferred interface for e -commerce, the web can fragment as we know it. Why visit a trade site when your agent can do the work? This shift could stimulate the development of Model Context protocols (MCPS) – AI optimized data strokes that replace Websites absolutely. Some sellers can respond by blocking well -known agent IPs or designing the cash register that frustrate automated systems to force direct interaction. Industries such as marketing Would fundamentally change because it becomes more important to get in touch with AI agents than people.
In the meantime, platforms such as Chatgpt ways must find money to make money with their new influence. This may mean that referral costs for charging sellers, which causes the rise of a new one SEO-For-AI ecosystem. But such an income introduces new questions about trust: if your agent takes traders’ agent, how unbiased are the recommendations?
While we are on the edge of this transformation, one thing is certain: the infrastructure of digital payments is being rewritten. The question now is whether consumers – and traders – are ready to follow their agents in this new era.
And for example, I will keep a close eye on – directly or through my agent.
LINK!
This article was produced as part of the TechRadarpro expert insight channel, where today we have the best and smartest spirits in the technology industry. The views expressed here are those of the author and are not necessarily those of TechRadarpro or Future PLC. If you are interested in contributing to find out more here: https://www.techradar.com/news/submit-your-story-techradar-pro
- Advertisement -
