Cleverbridge and Visa collaborate on trusted-agent payments for AI-initiated commerce

Cleverbridge said it is among the first companies to help enable and test Visa’s Trusted Agent Protocol and an “agentic directory” that is designed to help payments platforms grant secure access to approved software agents. The announcement positions the effort as a readiness step for “agentic commerce,” where software systems and AI agents initiate transactions on a customer’s behalf rather than through a conventional app or checkout flow.

In a release distributed July 2, Cleverbridge said it is working with Visa to prepare for secure AI-initiated transactions for software and SaaS businesses. The company did not provide additional technical specifics in the announcement, beyond describing the work as enabling and testing trusted-agent access and supporting secure connections between approved agents and commerce capabilities.

Visa’s Trusted Agent Protocol, as described in the announcement, is intended to establish trust boundaries for which agents are permitted to operate in a payments context. Cleverbridge’s role, the company said, includes participation in the testing and enablement process, with the goal of helping merchants and software companies prepare internal systems for a new transaction pattern.

The “agentic directory” reference in the same communication points to a mechanism for locating and verifying approved agents. Cleverbridge did not specify how the directory is implemented, what standards it uses, or whether it is integrated with any existing Visa programs or network endpoints. It also did not disclose which specific types of transactions were tested, such as card-not-present purchases, subscriptions, or tokenized payment flows.

Cleverbridge’s message was aimed at software providers that sell digital goods or services. For those businesses, agentic commerce could change how customers authorize purchases and how merchant systems handle payment initiation, because the initiator becomes an autonomous agent rather than a user clicking through a purchase interface.

Visa’s interest in the setup is consistent with the payments industry’s broader shift toward authentication, authorization, and partner-driven access controls. Instead of treating agent software as a generic client, a trusted-agent framework implies that permissions and identity can be managed more formally, potentially reducing the risk of unauthorized automation attempting to initiate transactions.

Cleverbridge did not disclose timelines beyond the July 2 announcement, nor did it name the particular merchants, agent developers, or test partners involved. It also did not report any performance outcomes from the testing, such as authorization rates, latency, fraud outcomes, or settlement effects, which would normally be expected if the initiative had reached an extensive trial stage.

For market watchers, the near-term question is how quickly trusted-agent capabilities translate into production deployments for merchant ecosystems and SaaS billing stacks. Companies preparing for AI-initiated transactions may look for follow-on documentation, pilot results, and integration guides that clarify the end-to-end workflow from agent authorization to payment initiation and confirmation. The practical details will determine whether agentic commerce becomes a small pilot feature or a scalable channel for routine digital purchases.