Visa has moved its agentic-commerce programme from the card issuer to the merchant’s checkout. The expanded Agentic Ready programme launches with more than 140 merchants, payment service providers, acquirers and commerce platforms testing how a trusted AI agent is recognised, authenticated and processed — the step where Know Your Agent stops being a policy question and becomes a live transaction check.
What happened
On 18 November 2026, Visa announced that Agentic Ready, part of Visa Intelligent Commerce, is opening to the acceptance side of the payment chain. The first phase, earlier in 2026, worked with issuers: more than 175 institutions across 45 markets testing agent-initiated transactions on their own credit and debit cards. The new phase gives merchants, PSPs, acquirers and commerce platforms a common framework to assess how trusted agent-initiated transactions can be “recognized, authenticated, and processed”.
The launch cohort includes Adyen, Antom, Billtrust, Checkout.com, Fiserv, Global Payments, PayPal and Stripe, alongside more than 140 initial participants. Visa says more financial institutions, merchants and enablers are expected to join through 2027. In Europe the programme starts with more than 85 acceptance participants and more than 40 issuers; the European list runs from Barclaycard Payments and Nexi to Klarna, Mollie, Revolut and Worldline.
“Agentic commerce is already taking shape, but scaling it requires more than innovation. Merchants need confidence in who they’re transacting with, and payment providers need to know these transactions can move securely and reliably through existing infrastructure,” said Rubail Birwadker, senior vice president for growth products and partnerships at Visa.
The testing is not only about whether a payment authorises. Visa’s regional releases describe partners assessing their existing systems, running agent-initiated transactions in real-world environments and identifying gaps across agent-led discovery, checkout and post-purchase interactions, including how payment credentials are handled. Visa Acceptance Platform supplies the surrounding pieces: credential management, fraud protection and dispute management.
Why it matters for KYA
An issuer-issued agent credential answers one question: did the cardholder authorise this agent to spend? That is necessary and not sufficient. At the merchant’s end, the question is different — which agent is this, who is accountable for it, what is it allowed to buy, and can I show afterwards that it stayed inside those limits? A payment token proves authority to pay. It does not, by itself, tell a merchant which principal stands behind the agent or whether the agent is the same one the issuer verified.
That gap is why “recognised” and “authenticated” appear together in Visa’s description. Recognising agent traffic — for example by verifying a signed request from a registered buying agent rather than a browser session — establishes something about the software making the request. Authenticating the transaction establishes that the credential is valid. Neither is the same as an authority check against a named principal, and a merchant that treats the first two as the third will approve transactions it cannot later defend.
The programme is also a reminder that KYA infrastructure is being negotiated by pilot rather than by standards body. More than 140 acceptance participants testing in production will produce de facto requirements — which fields a merchant needs, how fast a revocation has to arrive, what evidence a disputed agent purchase requires — long before a formal specification exists. Organisations that wait for the specification will inherit whatever the pilots settled on.
For practitioners, four things are worth building now, whatever the programme’s outcome: record the agent identifier alongside the payment authorisation in the same transaction log; keep the delegation record (principal, agent, scope, limit, expiry) queryable at dispute time; make revocation reach the merchant during the session rather than at the next statement; and separate the fraud signals for agent traffic from the bot-management rules built for scrapers, because a legitimate agent and a malicious one can look identical at the network layer.
What to watch
- Whether “recognised and authenticated” becomes a published, implementable specification that non-participants can adopt — and whether it is open or Visa-only.
- Whether the same agent identifier can be correlated across issuer, network and merchant, or whether each link sees a different token with no common handle.
- Whether dispute and liability rules end up distinguishing an agent’s mistake from the cardholder’s, and who carries the loss in each case.
- Whether the 2027 cohort publishes approval-rate and fraud data. That data, not the partner list, will show whether agent verification actually works at the till.
Sources
- Visa, “Visa Expands Agentic Ready to Help Scale Trusted AI Commerce Globally”, 18 November 2026 — investor.visa.com
- Visa Europe, “Visa Expands Agentic Ready to Help Scale Trusted AI Commerce across Europe”, 18 November 2026 — mynewsdesk.com
- Visa, “Visa Launches Agentic Ready Program in CEMA” (first-phase programme description) — visa.com
- FX News Group, “Visa expands Agentic Ready program to acceptance ecosystem”, 18 November 2026 — fxnewsgroup.com
Know Your Agent (KYA) explains agent identity, verification and accountability. This is an explainer, not legal or compliance advice — see our Sources & methodology. New to KYA? Start with What is Know Your Agent? and the glossary.