The technical infrastructure for AI agents to autonomously conduct payments is now fully functional, yet merchants and consumers remain deeply skeptical about letting machines handle transactions. AWS AgentCore Payments reached general availability on August 18, 2026, providing the execution layer that allows AI agents to discover, access, and pay for APIs, content, and services using the x402 protocol and Stripe's Machine Payment Protocol (MPP). Despite this milestone, agentic commerce currently represents less than 1% of U.S. e-commerce transactions, even as AI-generated traffic to retail sites has surged 4,700% year-over-year.

The disconnect reveals a fundamental challenge in emerging technology adoption: building the plumbing is far easier than convincing people to use it. For developers, AWS AgentCore Payments has dramatically simplified the engineering burden. Ampersend, a company working with Edge & Node, completed full integration in under two weeks. Without the managed wallet and spending control infrastructure, the team estimated the same work would have required three to four months of development.

"Managed wallets, spending guardrails and x402 settlement let us demonstrate fully autonomous agent-to-agent micropayments in a matter of days," said Rodrigo Coelho, CEO of Edge & Node.

Rodrigo Coelho, CEO of Edge & Node

What Is the Three-Layer Stack for Agentic Commerce?

Industry experts have identified a three-layer architecture needed for autonomous commerce to function at scale. Each layer solves a different problem in the agent payment ecosystem:

• Identity Layer: Solutions like Alchemy's AgentCard provision each AI agent with a unique email address, phone number, stablecoin wallet, and tokenized payment credentials, establishing who the agent is in the transaction.
• Trust Layer: Initiatives like Visa's secure agent transaction projects aim to scale from hundreds of closed-beta transactions to millions by the end of 2026, addressing whether the transaction can be trusted.
• Execution Layer: AWS AgentCore Payments handles the actual settlement, managing spending limits, wallet custody, and payment orchestration across multiple protocols.

"The internet was built for human interactions, but the infrastructure of the future must be built for autonomous ones," stated Stephanie Cohen, Chief Security Officer at Cloudflare.

Stephanie Cohen, Chief Security Officer at Cloudflare

Why Are Merchants and Consumers Hesitant About Agent Payments?

The readiness-adoption gap reflects significant trust deficits on both sides of the transaction. Consumer confidence in AI payment handling remains low. According to the Visa Earning Trust Report, only 23% of U.S. consumers trust artificial intelligence to handle payment transactions. On the merchant side, the skepticism runs even deeper. Ninety-three percent of merchants believe that AI providers should bear the financial loss for incorrect purchases made by agents, and only 28% are willing to offer their full product range to autonomous systems.

A secondary problem compounds the trust issue: the "shadow agent gap." Sixty-seven percent of U.S. workers already use unapproved AI tools in their daily work, creating enterprise risk that formal identity layers do not address. This means that even as companies build official agent payment infrastructure, unauthorized AI systems are already operating in the background, outside any governance framework.

How to Navigate the Protocol Proliferation Problem

One major barrier to widespread adoption is the fragmentation of competing payment standards. Rather than a single unified protocol, the industry has splintered into at least five competing checkout protocols, each requiring separate integration work:

• Visa Intelligent Commerce: Visa's proprietary agent payment standard requiring integration costs between $5,000 and $500,000 depending on complexity.
• Mastercard Agent Pay: Part of Mastercard's Agent Pay for Machines program, which includes over 30 industry partners building agentic commerce infrastructure.
• Stripe ACP (Agent Commerce Protocol): Stripe's standard, which also supports the Machine Payment Protocol (MPP) launched in March 2026.
• Google UCP: Google's unified commerce protocol for agent transactions.
• Meta Muse: Meta's proprietary agent payment framework.

This fragmentation creates what industry observers call the "protocol proliferation tax." Each integration can cost between $5,000 and $500,000, forcing merchants to choose which standards to support rather than implementing a universal solution. The result is that only 3% of current transactions involve agents, despite the infrastructure being technically ready.

How Is Ripple Positioning Itself in the Emerging Agent Payment Ecosystem?

Ripple has taken a strategic approach to the fragmented standards landscape by integrating into multiple emerging frameworks simultaneously. The company launched the XRPL AI Starter Kit on June 10, 2026, enabling AI agents to autonomously create wallets and conduct payments using XRP or Ripple's RLUSD stablecoin through the x402 protocol. The kit allows AI clients, including Anthropic's Claude, to interact directly with XRP Ledger documentation and perform wallet creation, transaction monitoring, and real-time payment processing.

Ripple's technical advantages are concrete. XRP Ledger settlement times run between 3 and 5 seconds, fees are predictable rather than volatile, and the ledger includes a native decentralized exchange. The x402 protocol has already processed a substantial volume of machine-to-machine transactions, with estimates placing throughput between 75 and 120 million transactions in recent months, worth tens of millions of dollars in aggregate value.

Beyond x402, Ripple also joined Mastercard's Agent Pay for Machines program, a consortium of over 30 industry partners, positioning XRPL across multiple emerging standards. This dual-track strategy hedges against the risk that one protocol will dominate while others fade. Stripe and Coinbase, both major players in the x402 ecosystem, are also members of Mastercard's initiative, suggesting that multiple standards may coexist rather than consolidate.

What Does Real Machine-to-Machine Transaction Volume Tell Us?

Despite the low consumer and merchant adoption rates, the underlying infrastructure is already processing meaningful transaction volume. The x402 protocol alone has handled between 75 and 120 million machine-to-machine transactions in recent months, representing tens of millions of dollars in aggregate value. This suggests that agent-to-agent commerce is already happening at scale in specific use cases, even if consumer-facing agentic commerce remains negligible.

The distinction matters. Machine-to-machine payments for API access, model inference, and digital services are already economically viable and operating at volume. The bottleneck is not technical feasibility but rather consumer trust and merchant willingness to expose their full product catalogs to autonomous buyers. Until those sentiment gaps narrow, the infrastructure will continue to outpace the market it was built to serve.