Agentic AI promises seamless automated payments and hyper-personalised retail, but growth depends on new trust, fraud models and rules for autonomous agents.
Agentic AI is starting to create a serious buzz. Capable of the kind of independent decision-making previously restricted to scenes from a sci-fi film, these autonomous systems are quickly being discussed as the next frontier in productivity.
Unlike traditional automation, AI agents can adapt and react in real time, using historical data and relationships to determine the best way to complete tasks.
Much like generative AI quickly took over the business world, agentic AI could soon become our personal assistants, with far-reaching implications for the retail and payments industries.
A new model for transactions
The idea that autonomous AI agents could soon handle payments entirely on our behalf is exciting, if not a little disturbing. A recent global survey found that 66% of consumers are open to AI agents making purchases on their behalf, primarily to save time and automate routine transactions.
Industry leaders are starting to respond to this growing sentiment. Mastercard and Visa released their strategies for agentic payments within a day of each other. Google also announced plans to introduce autonomous commerce tools, while PayPal’s collaboration with Perplexity is bringing AI shopping into chat-based interfaces.
It’s the next step in convenience. From needing to search, select, and pay, consumers are looking to shift to systems that can anticipate their needs, automate choices, and complete transactions—all with as little input as possible. What once took multiple taps or physical effort now happens seamlessly in the background.
We used to dream about technology taking away the friction in retail and payments. But is it still just a dream? Because, while agentic AI may be the next logical evolution for payments, it also has far broader implications for control, trust, and the speed of commerce.
Challenging regulations
Despite all of the excitement, we have to learn to walk before we run. Regulation, fraud detection, infrastructure readiness, and public trust in AI agents all require our attention.
Most existing financial systems assume a human decision-maker is making transactions. With AI agents doing the work for us, what does that mean for identity checks, consent flows, and compliance procedures? None of these are currently designed to accommodate autonomous agents.
Traditional fraud detection is built around human behaviour. These systems track activity such as location shifts, irregular form entries, or device inconsistencies. AI agents behave differently, which means that many fraud tools are ill-equipped to detect risks posed by autonomous systems.
There is also the added risk of poorly configured or malicious agents operating independently, potentially exposing users or organisations to unintended transactions. As the use of agents grows, fraud models will need to be rebuilt around how machines behave rather than how humans act.
For example, established standards such as PCI compliance do not currently address how to handle card payments initiated by non-human actors. Important questions also remain regarding authorisation, liability, and dispute management.
It’s clear then that a framework for verifying agents is necessary. As know-your-customer (KYC) protocols have become central to digital finance, a similar approach should be required for AI agents. Know-your-agent (KYA) procedures would essentially allow businesses to verify the identity and reliability of these autonomous systems before allowing them to conduct financial activity.
Transformative potential across the ecosystem
Introducing an agent into a financial workflow is not unlike hiring a new staff member. It requires trust, verification, and ongoing oversight. There are early models, such as Visa’s Intelligent Commerce plan and Mastercard’s Agent Pay, but the industry still lacks a widely accepted standard. Without this, it could prevent the wider growth and adoption of agentic payments.
But if we can find a way to navigate these concerns, the benefits of agentic payments could be substantial. For merchants, there is potential not only to reduce the likelihood of cart abandonment dramatically, but also to improve the intelligence of customer interactions on a scale we’ve never seen before.
For consumers, the lure of hyper-personalised shopping experiences and the complete automation of the checkout process could be the most significant driver of agentic payments. But, whether fully or semi-automated, getting there will mean building a new level of trust in AI.
