What Is the Visa Trusted Agent Protocol?

The Visa Trusted Agent Protocol is a specification that lets a merchant cryptographically verify that an incoming request comes from a legitimate, Visa-recognized AI shopping agent — rather than an anonymous bot — using signed HTTP messages, as part of Visa's broader Visa Intelligent Commerce program.

Key facts

• Announced October 14, 2025. Visa "unveiled Trusted Agent Protocol, establishing a foundational framework for agentic commerce that enables secure communication between AI agents and merchants during every step of a transaction," per Visa's press release.
• Built on established web standards, not a Visa-only cryptographic scheme. Per Visa's technical specification, the protocol uses HTTP Message Signatures compliant with RFC 9421, requiring Signature-Input and Signature header fields, and is described as aligned with the web-bot-auth standard. Signatures use asymmetric cryptography: agents sign requests with a private key, and merchants verify using a public key retrieved from a trusted key store (for Visa's own implementation, published at a .well-known/jwks endpoint).
• The protocol verifies three categories of information. Per Visa's developer overview: Agent Intent (confirming a Visa-trusted agent is seeking product details or a purchase), Consumer Recognition (identifying an existing customer account, using merchant tokens, device identifiers, location parameters, or wallet addresses), and Payment Information (supporting key-entry with hashed credentials, API/protocol-based transfers, or IOU-style management).
• Signatures are short-lived and tagged by interaction type. The specification defines two required signature tags — agent-browser-auth for browsing/product-lookup interactions and agent-payer-auth for completing a purchase — and requires signatures to include @authority, @path, created, expires, keyid, alg, nonce, and tag fields, with a validation window of 8 minutes between signature creation and expiration, per Visa's specification page.
• Cloudflare is named as the core development partner, with early partners including Adyen, Ant International, Checkout.com, Coinbase, CyberSource, Elavon, Fiserv, Microsoft, Nuvei, Shopify, Stripe, and Worldpay, per Visa's press release. Visa's Chief Product & Strategy Officer Jack Forestell stated: "We've worked closely with sellers, issuers and partners to make sure agent-initiated transactions are as seamless and secure as any payment today."
• Availability is described as in-progress, not fully live everywhere. Visa's developer documentation lists merchants and independent developers as the intended users, availability spanning North America, Asia-Pacific, Europe, CEMEA, and LAC, and states the protocol is "in the process of development and deployment," with materials and specifications available on the Visa Developer Center and GitHub.

How it works

1. An AI shopping agent that has been onboarded to a Payment Scheme (such as Visa) generates a cryptographic key pair, with the public key published to a trusted key store.
2. When the agent browses a merchant's site to check product details, inventory, or pricing, it signs its HTTP request using the agent-browser-auth tag, per Visa's specification.
3. The merchant's server retrieves the agent's public key from the trusted key store (matching by keyid and algorithm) and verifies the signature, confirming the request is from a Visa-trusted agent rather than an anonymous bot.
4. When the agent proceeds to complete a purchase, it signs a new request using the agent-payer-auth tag, which can include Consumer Recognition data (to identify a returning customer) and Payment Information (supporting several settlement approaches).
5. The merchant verifies this second signature the same way, and — because both signature types embed created/expires timestamps with an 8-minute validity window and a single-use nonce — a captured signature cannot be replayed later to impersonate the agent.
6. Once verified, the merchant can treat the agent as legitimate for that interaction and proceed with the transaction under Visa Intelligent Commerce, rather than blocking it as suspected bot traffic.

Visa Trusted Agent Protocol vs. Mastercard Agent Pay vs. AP2

FAQ

What problem does the Visa Trusted Agent Protocol solve? It lets merchants tell the difference between a legitimate, Visa-recognized AI shopping agent and an anonymous bot, using cryptographic signatures rather than behavioral guesswork alone, per Visa's press release.

Is Trusted Agent Protocol the same as Visa Intelligent Commerce? No. Visa Intelligent Commerce is Visa's broader program for enabling AI-driven payments; Trusted Agent Protocol is a specific verification mechanism that reinforces it, per Visa's developer documentation.

What cryptographic standard does it use? HTTP Message Signatures compliant with RFC 9421, aligned with the web-bot-auth standard, using asymmetric key pairs, per Visa's specification page.

Does the protocol include spending limits for the agent? Not specified in the primary sources reviewed for this article. The protocol as documented focuses on verifying agent identity and intent (browsing vs. paying); it does not describe a consumer-facing spending-cap feature.

Who built it with Visa? Cloudflare is named as the core development collaborator, with early partners including Adyen, Ant International, Checkout.com, Coinbase, CyberSource, Elavon, Fiserv, Microsoft, Nuvei, Shopify, Stripe, and Worldpay, per Visa's press release.

Is the protocol fully deployed today? Visa's own developer documentation describes it as "in the process of development and deployment" across North America, Asia-Pacific, Europe, CEMEA, and LAC, rather than universally live.

Related terms

• Agentic payments
• Agent Payments Protocol (AP2)
• Mastercard Agent Pay
• Know Your Agent (KYA)
• AI agent spending controls