AI agents can now buy products from your store without a human ever visiting your website. In April and May 2025, Visa, Mastercard, and PayPal each launched agentic payment platforms within weeks of each other, and Perplexity became the first AI search engine to complete end-to-end purchases through PayPal’s checkout. If your ecommerce store is not structured for AI agents to discover, evaluate, and purchase your products, you are already losing sales to competitors who are.
This is not a prediction. It is happening now. Perplexity Pro users in the US can ask “find me running shoes under $120” and complete the purchase with one click inside the chat, powered by PayPal’s tokenized wallet and passkey checkout. Visa partnered with Anthropic, OpenAI, Microsoft, and Perplexity to let AI agents browse, select, and pay using tokenized credentials. Mastercard launched Agent Pay with Microsoft and IBM. The payment infrastructure for agentic commerce is built. The question is whether your store is plugged into it.
The Agentic Payment Landscape: Three Platforms, One Shift
Three major payment networks launched agentic commerce capabilities between April 30 and May 14, 2025. This is not coincidence. It is coordinated industry infrastructure.
PayPal + Perplexity: The First Live Agentic Checkout
On May 14, 2025, Perplexity announced that PayPal will power agentic commerce across its platform. Starting summer 2025 in the US, Perplexity Pro users can check out with PayPal or Venmo directly inside the chat interface when asking Perplexity to find products, book travel, or buy tickets.
Key details from the official announcement:
- 430+ million active PayPal accounts across approximately 200 markets will be able to transact through Perplexity
- Payment, shipping, tracking, and invoicing are all handled behind the scenes
- PayPal’s passkey checkout eliminates passwords, reducing the purchase to a single query or click
- Perplexity answers over 150 million questions per week, many of them product-related
Perplexity’s VP of Business Ryan Foutty framed it clearly: the vision is assistive AI that handles decisions and purchases wherever users are. This is not a future roadmap item. It shipped.
Visa Intelligent Commerce: AI-Ready Cards for Agent Purchases
On April 30, 2025, Visa launched Visa Intelligent Commerce at its Global Product Drop event. The core concept: AI agents get tokenized digital credentials (called “AI-Ready Cards”) that let them purchase on behalf of consumers within pre-set spending limits.
Visa’s partners include Anthropic, IBM, Microsoft, Mistral AI, OpenAI, Perplexity, Samsung, and Stripe. The scope is significant:
- Tokenized credentials replace raw card details, so agents never see the actual card number
- Consumers set spending limits and preferences, and Visa manages the rest
- Built on 30+ years of Visa’s fraud detection and machine learning infrastructure
Jack Forestell, Visa’s Chief Product and Strategy Officer, put it directly: “Soon people will have AI agents browse, select, purchase and manage on their behalf.”
Mastercard Agent Pay: Agentic Tokens for Trusted Commerce
Also in April 2025, Mastercard launched its Agentic Payments Program, called Agent Pay. The program introduces Mastercard Agentic Tokens, extending existing tokenization that already powers mobile contactless payments and secure card-on-file systems.
Mastercard is collaborating with Microsoft (Azure OpenAI Service, Copilot Studio) and IBM (watsonx Orchestrate), plus acquirers like Braintree and Checkout.com to bring tokenized agent payments to merchants.
The common thread across all three platforms: tokenized credentials replace raw payment data, consumers set boundaries, and AI agents execute purchases within those boundaries. The trust layer is solved. What is not solved is whether your store can be found and understood by those agents.
What Agentic Payments Mean for Ecommerce Stores
The payment layer is just the last mile. Before an AI agent can pay, it needs to:
- Discover your product through search, comparison, or recommendation
- Evaluate your product against alternatives using structured data
- Select your product based on price, availability, shipping, and trust signals
- Complete the purchase through an agentic payment flow
Most ecommerce stores are built for human eyes, not agent parsing. Agents do not browse your hero banner. They read your structured data, product feeds, and API responses. If those are missing or incomplete, you are invisible to the agentic commerce pipeline regardless of which payment platform the consumer uses.
The Agentic Commerce Readiness Gap
Here is what the data shows about the current state of ecommerce readiness:
- Only 31% of ecommerce stores have complete product schema markup (Product, Offer, AggregateRating) according to schema coverage audits
- Zero-click searches hit 69% on Google as of 2025 (Similarweb data), meaning most product discovery now happens in AI-generated answers, not on your website
- AI Overviews reduce clicks by 58% for position-1 results (Ahrefs, Dec 2025 data), pushing traffic further down the funnel toward AI-mediated purchases
The stores that get cited by AI agents are the ones that will get purchased through agentic payment flows. The ones that are not cited will not even be in the consideration set.
The Agentic Payment Readiness Checklist for Ecommerce
If an AI agent is going to purchase from your store, here is what needs to be in place.
1. Machine-Readable Product Data (Non-Negotiable)
Your product pages must include structured data that agents can parse without rendering JavaScript. This means:
- Product schema with name, description, image, SKU, brand
- Offer schema with price, priceCurrency, availability, priceValidUntil
- AggregateRating with ratingValue and reviewCount
- Shipping details via OfferShippingDetails schema
- Return policy via MerchantReturnPolicy schema
Without Offer schema, an agent cannot confirm your price. Without availability markup, it cannot confirm the item is in stock. Without shipping details, it cannot compare your delivery time against competitors. Each missing field is a reason for the agent to skip your store.
Tools like Shopti’s agent discoverability diagnostic can identify exactly which fields are missing from your product pages and which ones are blocking agent visibility.
2. Product Feed Accessibility
AI platforms like Perplexity pull product data from multiple sources: web crawling, merchant feeds, and marketplace integrations. Your store needs:
- A Google Merchant Center feed that is accurate and up to date
- A llms.txt file at your root domain that directs AI agents to your product catalog and key pages
- Canonical product URLs that do not change with session IDs or tracking parameters
If you have not set up llms.txt yet, the complete llms.txt guide for ecommerce covers the exact format and placement.
3. API and MCP Integration
Model Context Protocol (MCP) servers are how AI agents interact with your store programmatically. An MCP server exposes your product catalog, inventory, and checkout as tools that agents can call directly.
Perplexity’s shopping integration already uses Shopify’s product API. Visa’s Intelligent Commerce uses tokenized APIs. The pattern is clear: agents need programmatic access to your store’s data, not just your HTML.
If you are on Shopify, some of this is handled through Shopify’s agentic commerce features, but as the Shopify agentic plan analysis explains, Shopify’s implementation covers basic plumbing while leaving significant gaps in product data quality and agent discoverability.
For stores on WooCommerce, BigCommerce, or custom platforms, you need to build or install an MCP server that exposes:
- Product search and retrieval
- Real-time inventory and pricing
- Cart creation and checkout initiation
- Order status and tracking
The MCP server guide for ecommerce covers the architecture and implementation details.
4. Payment Flow Compatibility
For agentic payment flows specifically:
- Accept PayPal and Venmo (required for Perplexity’s current integration)
- Support tokenized card payments through your payment processor (Stripe, Braintree, Checkout.com)
- Enable passkey authentication where possible to reduce friction for agent-initiated checkouts
- Do not require CAPTCHA or browser fingerprinting on checkout APIs, as these block automated purchase flows
If your checkout requires JavaScript rendering, visual CAPTCHAs, or multi-step browser-based flows, agentic payment systems will fail at the last mile. Your checkout needs an API-accessible path.
5. Trust and Authentication Signals
AI agents evaluate trust before purchasing. The signals they look for include:
- SSL certificate (non-negotiable)
- Verified business information in schema (name, address, phone)
- Customer reviews with structured markup
- Clear return and refund policies in machine-readable format
- Consistent NAP data (Name, Address, Phone) across the web
These are the same signals that build consumer trust, but they are especially important for agents because agents cannot “feel” trust. They parse it from data.
The Three Agentic Payment Platforms Compared
| Feature | PayPal + Perplexity | Visa Intelligent Commerce | Mastercard Agent Pay |
|---|---|---|---|
| Launch | May 14, 2025 | April 30, 2025 | April 2025 |
| Status | Live (US, summer 2025) | Announced, rolling out | Announced, pilot phase |
| AI Partners | Perplexity | Anthropic, OpenAI, Microsoft, Mistral, Perplexity, Samsung | Microsoft, IBM |
| Payment Method | PayPal, Venmo | Tokenized Visa cards | Mastercard Agentic Tokens |
| Consumer Controls | Account linking, passkey | Spending limits, preferences | Tokenized credentials |
| Merchant Integration | Shopify (current), expanding | Via Stripe, merchant APIs | Via Braintree, Checkout.com |
| Geographic Reach | 200 markets (430M accounts) | Global Visa network | Global Mastercard network |
The key takeaway: all three platforms use tokenized credentials, require merchant API access, and depend on structured product data. If your store is ready for one, it is close to ready for all three.
What Happens If Your Store Is Not Ready
The agentic commerce shift creates a new kind of digital divide. Stores that are machine-readable, API-accessible, and payment-compatible will appear in AI agent recommendations and be purchasable through agentic checkout flows. Stores that are not will effectively not exist for a growing segment of shoppers.
Consider the numbers:
- Perplexity answers 150 million questions per week, many of them product comparisons
- Google AI Overviews now appear on 60%+ of US searches (Semrush, Oct 2025 data)
- Zero-click search rates hit 69% (Similarweb), meaning most product research happens entirely within AI answers
- ChatGPT referrals to publisher sites are up 25x year over year (PPC Land data)
The consumer does not visit your website first. They ask an AI. The AI recommends products based on structured data and agent accessibility. The consumer completes the purchase inside the AI interface. If your store is not in the AI’s recommendation set, you do not get a second chance.
Implementation Priority: What to Do This Week
Not everything needs to happen at once. Here is the priority order:
Week 1: Product Schema Audit Run a schema audit on your top 50 product pages. Verify that Product, Offer, and AggregateRating schema are present and valid. Fix any missing or malformed fields. Use Google’s Rich Results Test or Shopti’s diagnostic to identify gaps.
Week 2: Payment Compatibility Confirm your store accepts PayPal and Venmo. Check that your checkout has an API-accessible path (no JavaScript-only forms, no CAPTCHA on API endpoints). If you use Stripe, enable tokenized payment methods.
Week 3: Feed and Discoverability Set up or update your Google Merchant Center feed. Create an llms.txt file pointing to your product catalog, category pages, and key policies. Verify your robots.txt allows AI crawler access (check the robots.txt audit guide for details).
Week 4: MCP Integration Assessment Evaluate whether your platform supports MCP or has an API that could be wrapped in an MCP server. Start with read-only product and inventory endpoints. Checkout integration can follow once the payment infrastructure is in place.
FAQ
Can AI agents actually buy products from my store right now?
Yes, with limitations. Perplexity Pro users in the US can purchase products through PayPal integration starting summer 2025. The product must be discoverable by Perplexity (through web crawling, Shopify integration, or product feeds) and the checkout must be compatible with PayPal’s tokenized payment flow. Visa and Mastercard’s agentic payment programs are in earlier stages but follow the same pattern.
Do I need to change my payment processor to support agentic payments?
Not necessarily. If you already accept PayPal, Venmo, or major credit cards through a modern processor (Stripe, Braintree, Checkout.com), you are likely compatible. The key requirement is supporting tokenized payments and having an API-accessible checkout path. Check with your payment processor about their agentic commerce support.
What is the difference between agentic commerce and regular ecommerce?
In regular ecommerce, a human browses your website, adds items to a cart, and completes checkout manually. In agentic commerce, an AI agent discovers your product, evaluates it against alternatives, and initiates the purchase on behalf of the consumer, often without the consumer ever visiting your website. The consumer sets preferences and spending limits, and the agent handles the rest.
How do I know if my product schema is good enough for AI agents?
Use Google’s Rich Results Test, Schema.org’s validator, or Shopti’s agent discoverability diagnostic. Check that every product page includes Product schema (with name, image, description), Offer schema (with price, currency, availability), and ideally AggregateRating and shipping details. Missing fields mean agents cannot evaluate your product properly.
Is this only relevant for large stores or marketplaces?
No. Agentic commerce is platform-agnostic by design. Perplexity’s integration started with Shopify but is expanding. Visa and Mastercard’s programs work with any merchant that supports tokenized payments. A small DTC brand with clean structured data and PayPal checkout is in a better position than a large retailer with broken schema and a JavaScript-only checkout.
Sources
- PayPal / Perplexity Partnership Press Release (May 14, 2025): newsroom.paypal-corp.com
- CNBC: “Perplexity partners with PayPal for in-chat shopping” (May 14, 2025): cnbc.com
- Visa Intelligent Commerce Announcement (April 30, 2025): usa.visa.com
- Mastercard Agent Pay Announcement (April 2025): mastercard.com
- Similarweb Zero-Click Search Data (2025): ppc.land
- Ahrefs AI Overviews Click Reduction Study (Dec 2025): ahrefs.com
- Semrush AI Overviews Study (2025): semrush.com
Check your store agent discoverability score free at shopti.ai.
