
Introduction
Nearly half of American shoppers already use AI tools for at least one step of their shopping journey.1 About half of that same group say they will abandon AI agents entirely if they lose visibility into what those agents are doing.2 That tension is the central challenge of agentic commerce: technology could move faster than trust.
Agentic commerce is not a future scenario. The question is whether the industry builds the right foundation before the window to shape it closes.
At Visa, we were one of the first major businesses to launch agentic commerce products and services and therefore have been watching the signals closely. As we monitor the market, we’ve been able to determine what is happening in the market today and the movements likely to help agentic commerce to grow at scale. Here are the five trends that may determine how this plays out.
1. Consumer trust, not technology, is the real constraint
The most important insight in agentic commerce today is not about technology. It is about psychology. Technology is outpacing trust.
More than half of consumers are familiar with AI shopping assistants and nearly one in three expects to use them regularly.1 But willingness to act is conditional: approximately 85% want transparency and control over how AI uses their data,2 and about half say they would stop using agents altogether if that control disappeared.2
The design question is not “how do we make the agent faster?” It is “how do we make the consumer confident enough to delegate more?” Products that answer that question will scale. Products that skip it will stall.
2. Reliable infrastructure matters more than perfect protocols
Today’s internet was built for humans, not agents. Unstructured web pages, inconsistent product data, checkout flows designed for cognitive beings: none of it was architected for AI interaction. Direct integrations and purpose-built commerce connections offer clearer paths to reliable agentic experiences than navigating the open web at scale.
Early agentic commerce experiences are already emerging through browser automation, front-end orchestration, and machine-readable commerce protocols. The implication for builders is pragmatic: do not wait for perfect infrastructure. Build with interoperability in mind, because the web will be re-architected for agents over time.
“As consumers begin to delegate tasks to AI agents, maintaining control and security in payments is critical. Visa Intelligent Commerce is designed to support these experiences by enabling solutions that help consumers authorize agent-driven payments with confidence and clear boundaries.”
Tanner RicheVP, Growth Products and Partnerships, Visa
3. Payments primitives are becoming the trust layer for agents
What makes agent-delegated purchasing secure is not a clean interface. It is the machinery underneath: tokenized credentials that help to scope an agent’s payment authority, authentication frameworks that help to verify consumer-approved intent and instruction, and real-time risk and identity systems that read new signals including agent identity, intent, and consent. These are evolutions of the capabilities already underpinning trusted digital commerce, retooled for a world where AI initiates transactions on behalf of consumers.
The companies that lead on these primitives will do more than enable innovation. They will enable innovation that is also designed to address potential systemic risk. Building capability and managing exposure simultaneously: that is the difference between infrastructure that scales and infrastructure that eventually fails under pressure. At Visa, we’re using our existing infrastructure to help fuel new waves of innovation.
“The biggest risk in agentic commerce is not that the technology fails to work. It is that we build fast and forget to build trust. Every major commerce transition in history has required a trust layer. Agentic commerce is no different, and the ecosystem that gets trust right first will be the one that scales.”
Jalpesh ChitaliaVP, Agentic Commerce, Visa
This is why the payments layer is becoming the trust layer for agentic commerce. Whoever participates in building these layers will help define what responsible delegation looks like across the entire ecosystem.
4. Merchant readiness will define where agentic commerce actually works
For agentic commerce to work at scale, merchants must do more than just accept payments – albeit this is a great starting point. They must be discoverable by agents, structure product data for machine interpretation, verify trusted agents against malicious bots, drive towards payment token maturity, and enable the integrations required for reliable agent-initiated transactions.
The signal here is encouraging: 63% of merchants are already exploring or planning to implement agentic AI payments.3 But exploring and being ready are different things. The gap between intention and implementation will determine where agentic commerce works in the near term and where it does not.
For merchants, this is both a readiness question and a competitive one. Agents will transact where transactions are possible. The merchants who build for agent interaction early will capture commerce that others cannot.
5. The ecosystem that moves together will outlast the one that moves fast
The final signal is the most consequential. In every technology transition, there is a tension between speed and coordination. The participants who move fastest capture early adopters. The participants who move together shape how the technology ultimately scales.
Short-term, custom integrations can unlock early agentic commerce adoption. But ecosystems built on proprietary connections, where one agent works only with certain merchants and one issuer supports only certain platforms, will fragment over time. Integration costs rise. Security gaps emerge. The value that comes from shared standards disappears.
Interoperability is not a slow path. With an increasing number of frameworks, protocols, and standards in the market, interoperability is the key to durable scale. The approach that preserves connectivity across networks, agents, and participants is the one that earns ecosystem-wide adoption. Moving fast matters. Moving together matters more.
What comes next depends on what we build now
Agentic commerce is not arriving. It is already live. The question before every merchant, acquirer, PSP, issuer, developer, AI platform, and technology provider is not whether to engage with this transition. It is how.
Visa Intelligent Commerce exists to answer that question with infrastructure rather than aspiration: solutions, developer tools, protocols, and ecosystem programs that are designed to give payment network participants a foundation to build, transact, and scale with trust and confidence. From cash to cards to digital wallets, Visa has helped make major evolutions in commerce trusted and accessible to our ecosystem. The agentic era is no different.
The rules of agentic commerce are being written right now. The organizations that help write them will shape what commerce looks like for the next decade.
To learn more or explore how your organization can participate in the agentic commerce ecosystem, visit visa.com/agentic .
Ready to start building? Visit the Agentic Sandbox at developer.cybersource.com/hello-world/agentic-sandbox.html.
Sources/Footnotes/Disclaimer
- Visa Consumer Research (December 2025). "Visa and Partners Complete Secure AI Transactions, Setting the Stage for Mainstream Adoption in 2026." Visa Inc. press release. Nearly half of U.S. shoppers (47%) now use AI tools for at least one shopping task.
- Visa Earning Consumer Trust in Agentic Commerce (2025). Visa Agentic Commerce Consumer Research Surveys, U.S., Australia, and New Zealand. Conducted via AYTM agile consumer insights platform among adults 18-65 who shop online monthly. Available at: corporate.visa.com/en/products/intelligent-commerce/earning-trust-report.html.
- MRC/Visa/Verifi 2026 Global eCommerce Payments and Fraud Report (March 2026). Merchant Risk Council in partnership with Visa and Verifi. 63% of merchants are actively exploring or planning to implement agentic AI payments in the near future.