AI Agent Payments Transition from Theory to Real-World Commerce

A new research report from Visa and Artemis details how AI-powered payment agents are moving beyond conceptual models into active commercial use. The study categorizes these payments into two distinct operational realms.

Macro-Commerce: AI as a Personal Assistant

In this scenario, AI agents act on behalf of human users to complete purchases like subscription renewals, travel bookings, or online shopping. The process mirrors traditional e-commerce, with the AI executing the transaction.

Micro-Commerce: Machine-to-Machine Micropayments

This represents a more transformative arena involving frequent, tiny-value payments between software, services, or IoT devices. The report notes these transactions are typically under $1 but occur at immense scale, demanding unprecedented efficiency and low cost from payment rails.

Protocol Data Reveals Reality: Average Transaction Values in “Cents”

The report examines two protocols built for machine-native payments, with live data underscoring the micro-transaction trend.

  • Protocol A: Since its May 2025 launch, it has processed approximately $150 million in adjusted volume across a staggering 109.6 million transactions, primarily on networks like Base, Solana, and Polygon.
  • Protocol B: Built by Stripe and Tempo, it has settled around $25,000 across roughly 115,000 transactions since mid-March 2026.

Both datasets point to the same reality: the average payment value is merely a few cents. This is the blind spot for traditional payment systems.

Stablecoins vs. Credit Cards: Cost Structures Define Use Cases

Why do traditional card networks struggle with micro-commerce? The answer lies in cost. Card networks often have fixed minimum processing fees that become prohibitively high for a $0.50 payment. The report clarifies that while traditional cards remain suitable for macro-commerce, stablecoins and digital assets, with their low variable cost structure, hold a natural advantage for high-volume micropayments.

The future is not a winner-takes-all battle, but a landscape where card payments and stablecoin payments converge and complement each other based on specific use cases.

The Critical Hurdle Before Scale: The Unresolved Question of Liability

Despite promising technical pathways, widespread adoption of AI agent payments faces significant legal hurdles. The report highlights the liability gap as a primary concern: when an AI agent autonomously makes a faulty payment decision, who is accountable? The user, the agent's developer, or the infrastructure provider? Current legal frameworks lack clear guidelines for autonomous agent behavior, making this a foundational issue that must be addressed before mass-scale deployment.