American Express and Visa have both launched new payment tools and frameworks to support agentic commerce, where customers use AI assistants to source and buy products.
On 8 April, Visa Inc. unveiled its Intelligent Commerce Connect solution, which it describes as a "network, protocol, and token vault-agnostic on ramp to agentic commerce". It is designed for use by agent builders, merchants, and enablers and facilitates secure payment initiation, tokenization, spend controls, and authentication through a single integration via the Visa Acceptance Platform. Furthermore, it allows agents to pay with both Visa and non-Visa cards.
"From small businesses to the world's biggest retailers, Visa powers how people pay every day, millions of times over," said Andrew Torre, president of value-added services at Visa. "Intelligent Commerce Connect brings that same, trusted payment acceptance infrastructure into the emerging world of AI-driven commerce, so businesses can let AI agents buy on behalf of consumers, securely and at scale."
Amex unveiled its framework to support the use of its cards and membership in AI-powered transactions on 14 April.
The Amex Agentic Commerce Experiences (ACE)™ Developer Kit is a framework of technical specifications developed for "flexibility and interoperability" with existing and emerging protocols. Amex says it will enable "intent-driven transactions with end-to-end visibility across the commerce lifecycle through Amex's closed-loop network".
The developer kit was announced alongside Amex Agent Purchase Protection™, which Amex described as "an industry-first commitment to extend its backing to card member purchases made by registered AI agents across its network". It means human customers are protected in instances of AI agent error.
"AI agents are beginning to reshape how people discover products and services, plan travel and dining, and make purchases," said Luke Gebb, EVP and head of global innovation at American Express. "As these capabilities evolve, card members and merchants will expect the same level of trust and security that they always relied on from American Express. The ACE Developer Kit enables this in AI-powered commerce."
Visa and Amex join Mastercard in advancing agentic commerce
The payment systems used by merchants today were created for use by humans and clicking a "buy" button tells the system that a human is authorizing the transaction. However, when AI agents are introduced to the transaction, this assumption no longer holds – and this is why payment providers are rolling out new payment infrastructure via protocols that can verify the agent has permission to complete a transaction, authenticate the intent of the human behind the agent, and ensure accountability in the case of fraud or error.
The news from Amex and Visa follows Mastercard's 2025 announcement of Agent Pay. Mastercard's payment function enables AI agents to securely initiate and complete transactions using its advanced technologies, including network tokenization, authentication, and fraud and cybersecurity solutions.
It means more payment providers can now support agent-initiated payments and merchants can accept payment via such protocols as Google's UCP, as well as Visa's Trusted Agent Protocol, Stripe's Machine Payments Protocol (MPP), and OpenAI's Agentic Commerce Protocol (ACP).
Together, the advances are creating an end-to-end ecosystem to allow customers to make purchases through their AI agents, such as Perplexity, Claude, ChatGPT, Gemini and others. At present, the majority of these transactions are still authorized by a human, but over time they will be executed autonomously. Autonomous AI buyers are known as machine customers.
However, payment authorization is already becoming complex and without common protocols to support AI agents and machine customers, this landscape would quickly become confusing for consumers and expensive for merchants. The introduction of controls such as Agent Purchase Protection, are also essential to building customer trust.
Why consumer trust still has a long way to go
Payment protocols like these will be essential to getting the customer buy-in required for agentic commerce to become widely utilized.
Research conducted by CX Network in December 2025 and January 2026 found AI-first customer journeys is a top customer behavior and top CX trend at present. However, the research also confirmed awareness of how AI works/uses customer data was the most influential customer behavior for CX practitioners, and for the first time in 11 years, consumer privacy emerged as a top 10 CX trend. These results not only signal a shift in how customers buy, but they also signal customers need reassurance before fully trusting AI.
Several developments occurred over the course of the research period, including the launch of Moltbook, the social network for AI agents. In less than a week, the news cycle had churned through stories a number of negatively framed stories, from Moltbook's bots creating a religion, to major security flaws, and an expose on some accounts being controlled by humans posing as bots.
Mark Levy, CX Network Advisory Board member and author or The psychology of CX 101, said: "For me, Moltbook did two things at once: it made AI agents feel more real, and it made them feel harder to trust… I don't think people will be able to separate 'weird social network for bots' from 'my shopping agent has my wallet'. They connect those dots instantly. If you can't verify who's behind an agent, you can't trust what it's recommending. And if you can't trust it, you won't let it act – especially not when the action is spending money, changing settings, or making a decision you'll have to explain later."
When it comes to trusting a new way to shop, consumers have been here before. When ecommerce first took off, many were apprehensive about typing their card details into their home PC, but encryption and security measures were introduced and trust grew. Today, with the threats such as data breaches and prompt injection, the stakes are higher and scrutiny of the reach and power of AI is influencing how customers view AI in CX.
Moltbook is only one negative story to contribute to this trend. Others have included the Copilot hallucination that brought down a police chief and OpenAI's latest political scandal.
For agentic commerce and machine customers to take off, trustworthy payment frameworks are an essential first step to earning customer trust.
Quick links
- AI search and loyalty: Why the first question matters more than the last purchase
- How generative AI search is changing the customer journey
- Macy's AI shopping assistant drives 400% spending surge