Visa is rolling out a programme called “Agentic Ready” across Europe, testing how payment infrastructure handles transactions initiated not by humans, but by software agents.
The effort involves collaboration with Commerzbank and DZ Bank, two of the banks exploring how AI agents can be woven into existing payment systems without breaching compliance requirements.
According to the announcement, the programme focuses on scenarios where an AI agent is given a goal or a set of rules — and then completes a purchase autonomously, without a customer confirming each step. A system could, for example, monitor supply levels, compare prices, and trigger a transaction once defined conditions are met.
Current payment infrastructure is built around human identity. Every card transaction depends on confirming that a person authorised a purchase. If software begins initiating those transactions, banks need new ways to verify identity and intent at the system level — including how an agent proves it is acting on a user’s behalf, and what limits constrain its actions.
Compliance Is the Hard Part
The participating banks are testing whether AI-driven transactions can still satisfy fraud checks, maintain audit trails, and honour customer consent requirements. Those areas are tightly regulated, and any change to how a transaction originates must still clear existing oversight standards.
A RepRisk report, cited in the source, found that banks are already dealing with more frequent and costly AI-related incidents, with some leading to multi-million-dollar losses.
Visa‘s focus is infrastructure. The company is working through how payment networks should behave when the initiating party is software — covering agent authentication, transaction approval flows, and dispute resolution if something fails.
What Changes for Organisations
In large companies, procurement typically moves through multiple approval layers. AI agents operating within defined spending limits could compress that process for routine purchases, reducing manual steps. The trade-off is that organisations need explicit rules governing what agents are permitted to do — without them, the exposure to errors or misuse grows.
The company sees the shift as comparable in scale to the early move toward online payments, when banks had to retool for a new type of transaction flow entirely.
Regulators are paying closer attention to AI in financial decision-making, particularly in credit and fraud detection. Any expansion of agent-initiated payments will land in that same scrutiny.
For now, Visa‘s programme remains in a testing and system design phase. The question it is working to answer is a structural one: how does a payment network serve a user that does not hold a card?
Photo by Pixabay
This article is a curated summary based on third-party sources. Source: Read the original article
