All Three US Card Networks Now Support Agentic Commerce

Visa, Mastercard, and American Express have all launched dedicated agentic commerce infrastructure in the span of six weeks. Morgan Stanley projects US agentic commerce will reach $385 billion by 2030, and the payment networks are racing to own the rails beneath it. The payment layer is no longer the bottleneck. The real question: are your products visible to the AI agents that will use these rails to buy?

TL;DR: All three major US card networks now offer dedicated agentic commerce tools: Visa's Intelligent Commerce Connect, Mastercard's Verifiable Intent, and Amex's Developers Kit. The payment infrastructure is ready. The bottleneck has shifted to product data quality, because AI agents can only buy products they can find.

What Did Each Card Network Actually Build?

The three announcements came fast. Visa launched Intelligent Commerce Connect on April 8, enabling AI agents and merchants to handle agentic transactions through a single integration. The pilot includes partners like AWS, Highnote, and Mesh, and connects into Visa's existing Intelligent Commerce portfolio alongside its Trusted Agent Protocol with Anthropic, Microsoft, OpenAI, and Perplexity.

Mastercard embedded Verifiable Intent directly into its payment flow: an open, standards-based trust layer purpose-built for agentic payments. Live agentic transactions are already running across Latin America and ASEAN. Mastercard also partnered with FIS and Visa on "Know Your Agent" authentication for issuing banks, giving identity verification to AI-initiated purchases.

Then came American Express with its Developers Kit on April 14. The kit covers agent registration, consumer card enablement, intent analysis, and digital shopping cart context for risk management. Amex has already completed thousands of test agentic payments. Their most aggressive move: Amex will cover errors made by AI agents. As EVP Luke Gebb put it, "We're standing behind the agents."

Key Stat: 70% of banks say agentic AI will have a "significant or game-changing impact" on financial services, per SoundHound and American Banker research.

Why Is Amex's "Closed Loop" a Big Deal for AI Agents?

Most payment networks operate in pieces. One company issues the card, another runs the network, a third processes the merchant's payment. Amex handles all three. That closed-loop model gives them unique data depth when an AI agent initiates a purchase: they can see the consumer's spending patterns, verify the merchant, and assess transaction risk in a single pass.

For agentic commerce, this matters because AI agents need fast trust signals. An agent buying a $200 jacket on your behalf can't wait for multiple parties to negotiate authentication. Amex's architecture collapses that latency. Their kit also lets consumers set spending rules for agents ("if the cost is below $300, buy it"), which reduces disputes before they happen.

Visa and Mastercard are building trust layers across their multi-party networks, requiring coordination with issuers and acquirers. Amex just needs to coordinate with itself. That's a structural advantage that's hard to replicate.

How Does Stripe Fit Into This Picture?

Alongside the card networks, Stripe launched the Machine Payments Protocol (MPP) in March 2026 with Tempo. MPP is an open standard giving AI agents an internet-native way to pay. Any business already on Stripe can accept MPP payments with a few lines of code.

Stripe also develops the Agentic Commerce Protocol (ACP) with OpenAI, which powers product discovery and merchant redirect in ChatGPT. Their Agentic Commerce Suite includes merchants like URBN (Anthropologie, Free People, Urban Outfitters), Revolve, Coach, Kate Spade, and Etsy. Platform partners include Wix, WooCommerce, BigCommerce, and commercetools.

Between the card networks and Stripe, every major payment rail now has a dedicated path for AI agents. The infrastructure question is answered.

If the Payment Rails Are Ready, What's the Actual Bottleneck?

Here's the part most retailers aren't ready to hear. The payment side is solved. AI agents can authenticate, transact, and handle disputes across all major card networks. But none of that matters if the agent never discovers your product.

Think about how AI shopping works today. A consumer asks ChatGPT or Google AI Mode for a product recommendation. The AI evaluates structured attributes, descriptions, pricing, availability, and recommends products with the richest data. Then it sends the shopper to the merchant's site to buy. ChatGPT alone processes 50 million shopping queries daily with 900 million weekly active users.

But only 18% of e-commerce product pages have complete schema markup as of April 2026. That's 82% of products essentially invisible to AI shopping agents. You can have the best payment integration in the world and still not generate a single AI-driven sale because agents never found your product.

Walmart captures 20% of ChatGPT referral traffic. Amazon, which blocks AI crawlers, sits below 3%. Same payment rails. Completely different product data strategies. Completely different outcomes.

Key Stat: Brands that audited and rebuilt their product feed schema saw a 23% revenue lift by becoming discoverable across ChatGPT Shopping, Google Shopping, and Perplexity Shopping simultaneously.

What Should Retailers Do Right Now?

The payment networks have done their part. Google is telling merchants directly that "product data is infrastructure, not just a feed." New "conversational attributes" are coming to Merchant Center, purpose-built for how products get retrieved in AI conversations. Gap just became the first major fashion retailer to launch Instant Checkout in Google Gemini using the Universal Commerce Protocol.

While you're evaluating whether agentic commerce applies to your business, your competitors are already shipping enriched product feeds to ChatGPT, Google AI Mode, and Perplexity. During Cyber Week 2025, retailers with AI agents saw 7x higher sales growth (13% vs 2%) compared to those without. That gap only widens as payment infrastructure makes transactions smoother.

If your product pages have 5 to 8 structured attributes when AI agents need 30 or more, start there. Run a free AI readiness audit at paz.ai to see exactly what agents see and what's missing. The paz.ai AI Readiness Report scores your products in 30 seconds across four layers and shows the specific data gaps preventing your products from being recommended.

The payment rails are built. The checkout protocols work. Product data is now the rate-limiter, and the brands that fix it first capture a disproportionate share of every AI-driven transaction flowing through those new rails.

Frequently Asked Questions

What is agentic commerce?

Agentic commerce is the practice of AI agents autonomously discovering, evaluating, and recommending products on behalf of consumers. The AI handles research and comparison, then directs shoppers to the merchant's website to complete the purchase. The merchant retains the full customer relationship, including login, loyalty data, and checkout.

Which card networks support agentic commerce payments?

As of April 2026, all three major US card networks offer dedicated agentic commerce infrastructure. Visa launched Intelligent Commerce Connect, Mastercard embedded Verifiable Intent into its payment flow, and American Express released its Agentic Commerce Developers Kit. Stripe's Machine Payments Protocol (MPP) and Agentic Commerce Protocol (ACP) provide additional rails for AI agent transactions.

Do AI agents complete purchases inside ChatGPT?

No. OpenAI discontinued Instant Checkout in March 2026. ChatGPT now operates as a discovery and recommendation engine, redirecting users to merchant sites for purchase. Brands rejected being reduced to anonymous fulfillment centers where they lost traffic, customer data, and loyalty engagement.

How big is the agentic commerce market?

Morgan Stanley projects US agentic commerce will reach $190 billion (base case) to $385 billion (bull case) by 2030, representing 10 to 20% of all US e-commerce. During Cyber Week 2025, AI agents influenced $67 billion in global sales, accounting for 20% of all orders.

Why does product data quality matter more than payment integration?

AI agents recommend products based on the richness and completeness of structured product data. If your product pages lack schema markup or conversational context, agents cannot find or evaluate your products. Only 18% of e-commerce product pages have complete schema markup, making product data the primary bottleneck for agentic commerce even though payment rails are now fully built.

How can I check if my products are visible to AI agents?

Run a free AI Readiness Report at paz.ai. Enter any product URL and receive a score from 0 to 100 with a breakdown across Product Mapping, Structured Attributes, Attribute Context, and Product Context. The report shows exactly what AI agents see and which data gaps are preventing your products from being recommended.