Visa Launches Intelligent Commerce Connect for Agentic Payments

Visa announced Intelligent Commerce Connect, a new component of the Visa Intelligent Commerce portfolio that provides a single integration path for merchants, agent builders, and enablers to accept and process AI agent-initiated purchases. The solution integrates with the Visa Acceptance Platform and exposes Visa Intelligent Commerce APIs while also connecting to other card networks, enabling agents to pay with both Visa and non-Visa cards. Pilots are running with partners such as AWS, Aldar, Diddo, Highnote, Mesh, Payabli, and Sumvin, and Visa expects wider availability later this year.

The offering is designed to be network, protocol, and token vault-agnostic, reducing vendor lock-in for agent platforms. Core capabilities include secure payment initiation, card tokenization, spend controls, and authentication. The integration surface unifies Visa Intelligent Commerce APIs and external network APIs to let agents route payment authorization across multiple networks. The product explicitly supports major agent protocols including Trusted Agent Protocol, Machine Payments Protocol (MPP), Agentic Commerce Protocol (ACP), and Universal Commerce Protocol (UCP). Key developer and platform touchpoints:

• •Works with major token vault providers so agent platforms can reuse existing credential infrastructure
• •Makes merchant catalogs discoverable to AI platforms, enabling product metadata, pricing, and inventory exposure for agent-driven browsing and selection
• •Supports enablers that process agentic transactions on a merchant's behalf via the Visa Acceptance Platform

The announcement is Visa staking infrastructure claims in the emerging agentic commerce stack. As AI assistants and autonomous agents move from research demos to consumer use, payments will be a gating factor for real-world adoption. By offering a single integration that is both protocol- and vault-agnostic, Visa reduces the engineering friction for merchants and agent builders to accept agent-initiated checks. That keeps Visa positioned as the transaction rail while allowing interoperability with competing networks and third-party token vaults. For payments teams and platform engineers, this is less about a new machine learning model and more about standardizing a payments interface for an AI-native shopping surface.

Expect new integration work around catalog APIs, metadata hygiene, and authentication of agent identities. Merchants will need to expose structured product data and error-handling patterns for agents that expect conversational flows rather than traditional synchronous checkout UIs. Risk and fraud teams should revisit agent authentication, spend controls, and token lifecycle policies since agents may make repeated autonomous purchases on behalf of consumers. The product also suggests growing demand for standardized agent protocols and governance to distinguish legitimate agent behavior from abusive automation.

Adoption by large merchants and platform partners will determine whether Visa simply adapts existing rails to AI or helps define the de facto payments standard for agentic commerce. Track pilot results, token vault partnerships, and any cross-industry governance efforts to authenticate agents and manage liability.

For AI practitioners building commerce experiences, Intelligent Commerce Connect removes a major integration barrier and clarifies a payments path for agent-driven UX. For payments architects, it is a signal to prepare product catalogs, tokenization flows, and fraud controls for autonomous agent interactions.