What is agentic commerce? A plain guide for merchants

Agentic commerce in one paragraph

In traditional e-commerce, a shopper searches, clicks through to your store and buys there. In agentic commerce, a shopper asks an AI assistant something like "find me a waterproof hiking jacket under 200 dollars that ships to Canada". The assistant (the agent) reads product data from many stores, picks options, explains them and, depending on the platform, hands the shopper to your checkout or completes the purchase within the conversation. The shopper may never browse your collection pages.

How it works today (October 2026)

Several companies are building the plumbing. The landscape moves quickly; the points below are dated to what each company had published by October 2026.

OpenAI: Agentic Commerce Protocol (ACP) and ChatGPT

OpenAI's Agentic Commerce Protocol starts with a structured product feed shared with OpenAI so ChatGPT can index products and show accurate information. Feed onboarding is limited to approved partners who apply at chatgpt.com/merchants; OpenAI asks for a full feed daily and updates through its API during the day. In March 2026, OpenAI confirmed it was shutting down Instant Checkout, its original in-chat purchase feature, and moving to an app-based model for shopping in ChatGPT.

Google: Universal Commerce Protocol (UCP)

Google describes UCP as an open standard for agentic commerce, covering discovery through post-purchase support and designed to work with A2A, AP2 and MCP. Merchants integrated with UCP can show a checkout button on eligible listings in AI Mode in Google Search and in Gemini, paid with Google Pay. Google's help page lists eligibility in the United States, Canada and Australia for select merchants, and states that Merchant Center remains the central hub for product data.

Shopify: Agentic Storefronts and Shopify Catalog

Shopify's Agentic Storefronts put a merchant's products in AI chats such as ChatGPT and Gemini, with sales going through Shopify Checkout and orders appearing in the Shopify admin. Shopify says the feature is enabled automatically for eligible stores. Shopify Catalog structures product data so agents can find it, and Shopify now serves an agents.md file and UCP discovery endpoint for each store (see our llms.txt guide).

Microsoft: Copilot Checkout

Microsoft announced Copilot Checkout on January 8, 2026, in the United States, with PayPal, Shopify and Stripe as partners. Shopify merchants were to be enrolled automatically after an opt-out window; PayPal and Stripe merchants apply. See our Copilot Checkout guide.

An example shopping journey

1. A shopper asks an assistant for "a stainless steel kettle with temperature control, under 120 dollars, delivered this week".
2. The agent searches the sources it has: a product catalog or feed, a search index built by its crawler, or pages it fetches on the spot.
3. It shortlists three kettles and quotes a price, availability and shipping information for each, taken from that data.
4. The shopper picks one. Depending on the platform and your enrollment, the agent completes checkout in the chat, or sends the shopper to your product page or cart.
5. If the price or stock the agent quoted was wrong, the shopper finds out at that last step.

Every step depends on data you publish. That is why accuracy, not only visibility, is the merchant's main lever.

What changes for merchants

• The agent repeats your data. If your structured data, feed or catalog has the wrong price or stock, the agent passes it on with confidence. The shopper blames the store, not the agent.
• Several copies of the truth. Your product page, its JSON-LD, Merchant Center, Shopify Catalog and any partner feed can each hold a slightly different version of the same product.
• Policies matter more. Agents look for shipping and return information to answer questions. Thin or missing policies weaken your listing.
• Attribution is harder. When a purchase starts in a chat, tracking parameters can be lost on the way to your cart.

What merchants should do now

1. Check which programs you are already in. On Shopify, look at the agentic channels in your admin; Agentic Storefronts and Copilot enrollment may already be active.
2. Clean your product data. Full titles, real descriptions, images on every product, correct variants and barcodes where you have them.
3. Make sure structured data matches the page. One Product block, correct price, currency and availability for each variant. Our product schema guide shows how to check by hand.
4. Publish clear policies. Shipping, returns and refunds, with real details.
5. Keep Merchant Center healthy if you sell on Google; it remains Google's source of product data.
6. Decide your crawler policy with care; see AI crawlers and robots.txt.
7. Recheck regularly. Data drifts after sales, app installs and theme updates.

AI Sales Assurance helps with steps 3, 4 and 7: it compares what AI shopping agents receive with your real product pages, flags price, stock, schema, variant, link and policy problems, and tests the cart path up to (never through) payment. Start with a free full check.

Common misunderstandings

• "Agentic commerce is only checkout in a chat." Discovery matters at least as much: agents recommend products before any checkout happens, and they use your data to do it.
• "If I'm on Shopify, it's all automatic." Shopify connects the channels, but the content (titles, descriptions, theme schema, app-added markup) is still yours to keep correct.
• "A good readiness score means agents show correct prices." A readiness score measures setup at one moment. Accuracy has to be checked against your live pages over time.

Frequently asked questions

Related guides

Sources

• OpenAI: Commerce, get started
• Shopifreaks: OpenAI pivots from Instant Checkout to an app-based shopping model (March 2026)
• Google Merchant Center Help: Universal Commerce Protocol
• Shopify: Agentic Storefronts
• PPC Land: Microsoft launches checkout inside Copilot (January 2026)
• Shopify docs: agents.md.liquid template