Wirex joins Visa’s AI payments sandbox to trial stablecoin‑driven agents
Wirex has entered Visa’s new Agentic Ready program, a testbed designed to explore how artificial intelligence agents can independently initiate and complete payments using stablecoins, while still operating within Visa’s existing global network and compliance framework.
Under the initiative, Wirex will act as an issuer, collaborating with Visa and other partners to prototype payment flows in which software agents, not humans, execute transactions on users’ behalf. The core objective is to develop a technical and regulatory framework that lets AI tools handle money safely, with explicit user consent and clear controls over what those agents are allowed to do.
The push comes as businesses race to automate routine financial tasks. Companies are increasingly deploying AI tools that can monitor subscriptions, manage recurring invoices, or rebalance budgets without waiting for a manager to sign off on every small decision. Wirex points out that this emerging “agentic economy” is expanding at an estimated annual rate of 44%, creating pressure for payment infrastructure that can run continuously, across borders, and with minimal human intervention.
Wirex argues that stablecoins are particularly well‑suited to power these agent‑driven systems. Unlike traditional bank rails, which are constrained by cut‑off times, weekends, and legacy messaging formats, stablecoin networks can settle value around the clock. When combined with programmable logic, they allow agents to move funds instantly once predefined conditions are met, without being slowed by manual reconciliation or interbank delays.
Within Visa’s Agentic Ready framework, Wirex and other participants will experiment with real‑world payment scenarios, rather than purely theoretical models. Early trials are expected to center on business use cases where automation can immediately reduce friction: software‑as‑a‑service subscriptions, marketing and advertising budget allocation, and procurement workflows that involve numerous small payments to suppliers.
A key focus of the program is security and user protection. According to Wirex, every test case will be evaluated against baseline requirements for transaction transparency, auditability, and user oversight. The aim is to guarantee that, even when an AI agent is clicking the “pay” button, the underlying system still respects spending limits, verifies counterparties, and keeps users informed about what is happening with their funds.
Consent and control sit at the heart of the architecture being explored. Rather than granting an AI unrestricted access to an account, users would approve specific permissions: for example, allowing an agent to pay a subscription up to a fixed monthly cap, or to manage a campaign budget within pre‑defined thresholds. Users would be able to review activity, adjust parameters, revoke access, or pause the agent entirely.
Wirex says its participation in Agentic Ready builds on a longstanding relationship with Visa, where it already acts as a principal member. The new collaboration extends that partnership into the AI domain, exploring how automated systems might handle tasks like booking flights, renewing memberships, or managing expense flows without requiring a person to approve every transaction one by one.
Pavel Matveev, Wirex co‑founder and CEO, describes a noticeable shift in behavior among the company’s business clients. Many are moving from simple analytics tools toward agent‑driven solutions that can take actions, not just generate reports. Matveev argues that for this transition to be safe and scalable, a “trusted model” is needed-combining Visa’s global network and rules engine with Wirex’s compliance record and crypto infrastructure.
Visa’s Agentic Ready initiative does not exist in isolation. It is the latest step in a multi‑year effort by the payments giant to experiment with blockchain technologies, tokenized assets, and stablecoins. Previous pilots have examined everything from stablecoin settlement for card issuers to rewards programs denominated in digital assets, signaling that Visa sees programmable money as a long‑term pillar rather than a short‑lived trend.
In one recent proof‑of‑concept, Visa collaborated with Brale and participants operating on the Canton Network to test stablecoin settlement using Brale’s SBC token. That trial focused on whether privacy‑enhanced blockchain infrastructure could handle institutional payments without exposing sensitive transaction data to the public. Instead of using open blockchains where every transfer is visible, the Canton experiment ran in a permissioned environment, tailored to financial institutions that demand granular control over who can see what.
Over the past several years, Visa has also supported settlement using prominent stablecoins such as USDC on Ethereum. Layered on top of this, newer pilots have investigated stablecoin‑funded card payments, spending from tokenized asset balances, and crypto‑based reward schemes. Taken together, these projects suggest Visa is building the plumbing for a future where digital tokens and traditional card payments coexist within the same user experience.
Recent initiatives linked to Visa demonstrate how far this experimentation has already gone. In one case, a Visa‑branded card issued via Tether and Fasset allowed customers to spend tokenized gold while simultaneously earning rewards in a gold‑backed digital asset. In another, a major Japanese financial group launched a Visa‑linked card that pays out rewards in leading cryptocurrencies such as Bitcoin, Ethereum, and XRP through its local crypto subsidiary. These use cases hint at how flexible reward and settlement models could become once tokenized assets are fully integrated into card networks.
For businesses, agentic payments promise to simplify day‑to‑day financial operations. A marketing team, for instance, could authorize an AI agent to automatically distribute ad spend across platforms, pausing campaigns that underperform and reallocating funds where returns are higher-all while respecting a global budget and compliance rules. Procurement agents could source quotes, negotiate terms within set parameters, and schedule payments once goods are received, cutting down on back‑office paperwork.
Subscription management is another natural fit. Instead of relying on employees to remember renewal dates or manually cancel unused tools, an AI agent could monitor login frequency, usage metrics, and team feedback. If a service appears redundant, the agent might flag it for review or automatically downgrade the plan, ensuring the company is not paying for software it no longer needs.
On the consumer side, agent‑driven payments could reshape everyday experiences like travel booking. A user might grant an AI agent permission to find flights and accommodation that meet specific criteria-budget, airline preferences, loyalty programs, cancellation policies-and then pay for the selected options directly from a stablecoin wallet. The user retains control over the constraints, but does not need to manage every step manually.
At the same time, the rise of agentic payments raises new questions around liability and regulation. If an AI agent makes an unauthorized or erroneous payment because it was trained on flawed data, who is responsible-the user, the software provider, the issuer, or the network? Programs like Visa’s Agentic Ready are, in part, attempts to answer these questions early by embedding risk management, dispute resolution mechanisms, and audit trails into the technical design.
Data privacy and model governance will also be critical. For AI agents to make informed financial decisions, they often need access to sensitive information: transaction histories, contract terms, pricing data, and even internal company policies. Ensuring that this data is properly segmented, encrypted, and processed under strict governance frameworks will be a prerequisite for regulators and enterprise clients alike.
From a technical perspective, blending AI agents with stablecoin infrastructure introduces unique engineering challenges. Systems must manage private keys securely, handle on‑chain confirmations and potential network congestion, and bridge between blockchain settlement and traditional account‑based ledgers. They must also provide clear human‑readable explanations for each agent‑initiated payment to maintain trust and satisfy compliance requirements.
Looking ahead, observers expect agentic payments to evolve beyond simple automation into more sophisticated financial decision‑making. Future agents could compare lending rates, optimize treasury allocations across different stablecoins and banks, or hedge currency risk on behalf of businesses operating in multiple markets. Stablecoins provide the programmable settlement layer, while AI supplies the decision engine that chooses when and how to move funds.
For the broader crypto and fintech ecosystem, Wirex’s role in the Agentic Ready program signals a convergence of three previously distinct tracks: digital currencies, traditional card networks, and advanced AI. If the trials prove successful, the outcome could be a new class of financial products where users set high‑level goals and guardrails-and software quietly handles the execution in the background, powered by stablecoins and governed by established payment rules.
Visa and Wirex are framing this phase as experimental rather than fully commercial, but both organizations appear to be positioning themselves for a world in which autonomous agents are not just drafting emails or summarizing reports, but actually moving money. The Agentic Ready initiative is an early attempt to ensure that when that world arrives, it does so within a trusted, regulated, and user‑centric payment framework.