Agentic Commerce: A Complete Guide to Selling to AI

For three decades, the online checkout process has been fundamentally the same: a form on a webpage. Innovations from one-click ordering to digital wallets focused on speeding up that form, but the underlying page remained. That era is now ending. Agentic commerce represents a paradigm shift where artificial intelligence handles the entire purchase journey, eliminating the traditional checkout page entirely. This final evolution removes the last point of friction between consumer intent and possession, transforming how businesses must prepare to sell in an AI-driven marketplace.

The history of e-commerce is a story of progressively eliminating steps. It began in 1994 with the first secure online transaction, removing the need to visit a physical store. The late 1990s introduced comparison shopping engines, eliminating the need to visit multiple store websites. Amazon’s recommendation algorithms, starting in 1998, reduced the need for shoppers to know exactly what they wanted. The rise of conversational commerce on platforms like Facebook Messenger and WeChat after 2015 meant users no longer had to open a store’s website. Voice assistants and social commerce feeds like TikTok Shop further embedded purchase opportunities into daily digital life. By 2024, AI began doing the product research and comparison for users. The pivotal shift occurred in 2025, when AI agents like OpenAI’s Operator began autonomously completing purchases. The final step,requiring a human to be present for the transaction,was removed.

This brings us to the core change: in traditional commerce, the seller builds and controls the checkout experience. In agentic commerce, the AI agent does. The agent presents product details, pricing, and options within its own interface, such as a chat window. The user confirms, and the agent completes the purchase via an API call. The merchant never renders a checkout page; they simply receive and process the order. As Stripe’s guide states, user experience problems are becoming protocol problems. The merchant’s role shifts to providing impeccable structured product data and reliable order fulfillment.

This shift is already operational. As of early 2026, ChatGPT Instant Checkout is live for U. S. users, processing orders for merchants like Etsy and Walmart. Shopify’s Agentic Storefronts automatically syndicate products to multiple AI platforms. Perplexity’s Instant Buy feature allows direct purchases within its chat interface. Every major AI company, including Anthropic with its Claude assistant, is building commerce capabilities, committing to ad-free, agent-driven shopping experiences.

Two competing open standards are enabling this infrastructure. The Agentic Commerce Protocol (ACP), launched by Stripe and OpenAI in September 2025, is tightly focused on the checkout flow. It uses a four-party model involving the buyer, agent, merchant, and payment provider, with the agent managing the checkout interface. Stripe’s accompanying Agentic Commerce Suite offers a low-code solution for merchants to connect their catalogs and process payments across multiple AI platforms through a single integration.

In January 2026, a different coalition led by Shopify and Google unveiled the Universal Commerce Protocol (UCP). Designed as a full-stack commerce standard, UCP covers the entire journey from discovery to post-purchase. It is protocol-agnostic and uses a layered architecture similar to TCP/IP. While ACP is optimized for fast transactions within specific AI agents, UCP is built for a future with many competing AI platforms. Critically, these protocols are not mutually exclusive; a Shopify merchant can seamlessly serve both simultaneously.

A fundamental challenge for both protocols is trust in a person-not-present transaction. When an AI agent acts on a user’s behalf, traditional fraud signals based on human behavior are absent. Stripe’s answer is the Shared Payment Token (SPT), a programmable, scoped token that allows an agent to initiate a payment without ever accessing the user’s actual card details. Payment networks are also responding: Visa launched its Trusted Agent Protocol, Mastercard introduced Agent Pay, and Google developed the Agent Payments Protocol (AP2). These frameworks aim to cryptographically verify user consent and authenticate legitimate AI agents.

Fraud detection has also evolved. Stripe built a transformer-based AI model trained on billions of transactions to analyze risk without relying on human behavioral fingerprints. However, new vulnerabilities like visual prompt injection, where malicious content in product listings can hijack an agent, present novel security challenges that the industry is still addressing.

Adoption is accelerating rapidly. Major brands like Glossier, SKIMS, Coach, and Walmart are already selling through AI agents. E-commerce platforms including BigCommerce, Wix, and Squarespace have integrated with the necessary protocols. Market projections vary but are uniformly massive, with McKinsey forecasting up to $1 trillion in U. S. retail revenue orchestrated by agents by 2030. Consumer trust, however, is still developing. Surveys indicate only about 14% of consumers currently trust AI to place an order on their behalf, though AI-driven traffic to retail sites grew by an astonishing 4,700% year-over-year in 2025.

For businesses, preparation is straightforward. Merchants on platforms like Shopify should verify their Agentic Storefronts are active and ensure their product data is clean and complete. Those using Stripe can enable SPTs with minimal code. All businesses must audit their product data, ensuring descriptive titles, complete details, accurate pricing, and high-quality images. Implementing structured data markup like Product schema is essential for agent readability. Testing visibility by asking AI agents for product recommendations in your category is a practical first step.

The overarching takeaway is that checkout is becoming a protocol, not a page. The infrastructure for AI-driven commerce is being built at a remarkable pace. The agents are already beginning to shop. A business’s success in this new landscape depends on how machine-readable it makes its products, content, and commerce operations. The stores that agents can find and understand will be the ones that thrive.