Visa and Fiserv have launched an AI payment framework for autonomous shopping that lets AI “agents” securely complete purchases, combining Visa Intelligent Commerce with Fiserv’s merchant acceptance tools ahead of broader rollouts in 2026.
What launched, and why now
Visa said it has already completed hundreds of secure, agent-initiated transactions with partners, signaling that agent-driven checkout is moving from pilots toward mainstream use.
Fiserv said it will collaborate with Visa to bring these capabilities into its acceptance ecosystem so merchants can safely recognize and process agent-driven purchases without rebuilding their checkout flows from scratch.
The push comes as Visa research found that 47% of U.S. shoppers now use AI tools for at least one shopping task, such as price comparisons or personalized recommendations.
Visa also projects that millions of consumers will use AI agents to complete purchases by the 2026 holiday season, which raises the stakes for payment security, consumer consent, and bot prevention.
Who is involved, and where it will scale
Fiserv described the initiative as a strategic collaboration with Visa to enable Visa Intelligent Commerce and deploy Visa’s Trusted Agent Protocol across Fiserv’s interoperable “agentic” ecosystem.
Visa positioned the work as part of its broader Visa Intelligent Commerce effort, which it says is grounded in decades of AI use in payments and is intended to deliver secure, personalized AI-enabled commerce by early 2026.
Scaling is central to the announcement because Fiserv said it will deploy the Trusted Agent Protocol across its acceptance ecosystem, targeting use cases where AI agents discover, compare, and purchase products on a consumer’s behalf.
Fiserv highlights its broad footprint, including serving over six million merchant locations globally, which helps explain why the partnership matters beyond small pilots.
How the AI payment framework works
At the center of the collaboration is Visa’s Trusted Agent Protocol, which Fiserv and Visa said will be used to authenticate, retrieve, accept, and process agentic transactions.
Fiserv said the protocol is designed to distinguish trusted agents from malicious bots, ensure each interaction is authorized by a consumer with verified intent, and validate that payment information at checkout remains unaltered.
Visa described Trusted Agent Protocol as an open framework built on existing web infrastructure, introduced with more than 10 partners, to help merchants separate legitimate AI agents from malicious bot traffic during checkout.
Visa also said Akamai is supporting Trusted Agent Protocol, pairing it with bot and abuse protection and related behavioral intelligence to help merchants confidently welcome agents that have real “commerce intent.”
Beyond “acceptance,” Visa’s Intelligent Commerce positioning emphasizes tokenized payments and authentication APIs to protect transactions while reducing friction during checkout.
Visa also says intelligent agents can apply promotions and loyalty points at checkout and help support post-purchase protections such as dispute resolution—features aimed at keeping automation compatible with consumer trust and merchant protections.
Key building blocks (plain-English)
- Authentication: Verifies that a real consumer authorized the agent’s action and that the agent is legitimate.
- Tokenization: Replaces sensitive payment credentials with tokenized equivalents, reducing exposure of raw card data in automated flows.
- Agent integrity controls: Adds checks that help prevent “tampering” between what the consumer approved and what is presented at checkout.
What Visa brings vs. what Fiserv brings
The partnership is structured around Visa providing the agent-focused payment and trust layer, while Fiserv provides merchant distribution and operational integration across existing acceptance products.
| Component | Visa’s role | Fiserv’s role |
| “Agentic” payments platform | Visa Intelligent Commerce and its APIs/capabilities for AI-enabled commerce. | Makes it usable inside merchant tools and workflows already used in day-to-day commerce. |
| Trust framework | Trusted Agent Protocol to help identify legitimate agents and reduce malicious bot risk. | Deploys Trusted Agent Protocol across its acceptance ecosystem for merchant use. |
| Security services | Authentication and tokenization services aligned with Visa’s network capabilities. | Connects merchants to Visa services and supports adoption with scalable integration paths. |
| Operations | Standards and ecosystem partnerships to expand agentic commerce globally. | Operational intelligence that can automate dispute resolution and optimize routing in real time, per Fiserv. |
Real-world pilots that shaped this launch
Visa said it is working with more than 100 partners worldwide across the commerce ecosystem, with dozens building in its sandbox and multiple agents and “agent enablers” integrating directly with Visa Intelligent Commerce.
Visa also listed early U.S. pilots from partners (including Skyfire, Nekuda, PayOS, and Ramp) that have executed end-to-end consumer and B2B purchases in closed beta environments.
These pilots matter because “autonomous shopping” is not only about AI recommendations; it requires a payments layer that can safely move from “find” to “buy” with clear authorization and reliable fraud controls.
Visa framed the milestone as evidence that controlled, real-world agent-initiated transactions can work in live production settings—provided trust and authentication are built into the flow.
What merchants should expect next
Fiserv said the goal is to let merchants, ISVs, and ISOs embed AI-driven agentic experiences into workflows without disrupting existing operations, which is critical for adoption at scale.
Fiserv also positioned itself as a bridge between traditional payments and agentic commerce by linking merchants to Visa Intelligent Commerce authentication and tokenization services while enabling issuers and acquirers to participate.
Visa, for its part, said pilots in Asia Pacific and Europe are anticipated to start in early 2026, while additional readiness efforts are underway in Latin America and the Caribbean and other regions.
Visa also highlighted a Middle East example (working with Aldar in the UAE) focused on using AI agents to pay repetitive fees, showing that autonomous payments may expand beyond retail shopping into recurring commerce tasks.
Timeline to watch
| Date | Milestone | Why it matters |
| Oct 2025 | Visa introduced Trusted Agent Protocol with more than 10 partners. | Establishes agent identity and bot-vs-agent separation as a baseline requirement. |
| Dec 18, 2025 | Visa reported hundreds of secure agent-initiated transactions and projected millions of AI-agent purchases by Holiday 2026. | Moves the narrative from concept to measured pilots and a near-term adoption target. |
| Dec 22, 2025 | Fiserv announced collaboration to deploy Trusted Agent Protocol across its acceptance ecosystem. | Signals distribution to large merchant bases rather than limited pilots. |
| Early 2026 (expected) | Visa anticipates new pilots in multiple regions and aims for broader secure, AI-enabled commerce delivery. | Sets a market clock for merchants, issuers, and platforms to get “agent-ready.” |
Final thoughts
The Visa–Fiserv effort frames autonomous shopping as a security and trust problem as much as a convenience upgrade, because AI agents must be identifiable, authorized, and resistant to manipulation at checkout.
If the Trusted Agent Protocol deployment scales as described, merchants may gain a standardized way to accept agent-initiated transactions while keeping consumer intent and payment data protections intact.
The next milestones to track are how quickly merchants can enable the protocol inside existing acceptance stacks and how consistently issuers, acquirers, and commerce platforms align around shared rules for “trusted” agents.
