What is agentic commerce and how can your online store sell inside AI chats?

Agentic commerce is shopping where an AI assistant does the legwork for the buyer: it searches for products, compares them and, on some platforms, completes the purchase without the shopper leaving the chat. Your online store can sell inside AI chats once its product catalog, policies and checkout are connected to those assistants, through your ecommerce platform or programs such as Google Merchant Center.

This became a real sales channel in 2026. Shopify and Google announced the Universal Commerce Protocol on January 11. On March 24, Shopify opened selling across ChatGPT, Microsoft Copilot, AI Mode in Google Search and the Gemini app. Shopify then reported that orders from AI searches grew nearly 13 times year over year in its first quarter.

Store owners want a plainer answer than the headlines give. When a shopper asks an AI assistant for trail running shoes under $150, will your products show up, and can that shopper pay without leaving the chat? Three things decide it: where your customers live, what you sell and how clean your product data is.

Which AI assistants can sell your products today

Four AI surfaces now show products from online stores: ChatGPT, Microsoft Copilot, AI Mode in Google Search and the Gemini app. Only some of them let the shopper pay inside the chat. ChatGPT sends buyers to your own checkout, while Copilot and Google's two surfaces can complete the order in the conversation for eligible merchants.

Shopify's Agentic Storefronts launch put all four surfaces behind one settings screen in the Shopify admin. There are no separate integrations, no apps and no transaction fees beyond standard processing rates, and for eligible stores the channel is on by default.

OpenAI first offered Instant Checkout inside ChatGPT, then stepped back from it in March 2026. It said it would allow merchants "to use their own checkout experiences" while it focused on product discovery. Shopify's help pages now call the ChatGPT channel discovery-focused: the shopper finds your product in the chat and pays on your store's checkout, in an in-app browser or a new tab.

So on ChatGPT the sale still finishes on your website, and your product pages and checkout speed count as much as ever.

How agentic commerce works, from question to order

Agentic commerce works by letting an AI agent read merchant catalogs on the shopper's behalf. The agent matches the request to products, compares price, stock and policies, and asks the shopper to approve. The order is then placed inside the chat or on the merchant's site, and the merchant still takes the payment, ships the parcel and handles any return.

1. A shopper asks in plain language, for example "a waterproof hiking jacket under $200 that arrives by Friday."
2. The agent searches structured product data: catalogs, feeds and merchant listings that carry title, price, stock, shipping and return terms. Your homepage banner plays no part.
3. It shortlists a few options and shows them side by side with price, reviews and delivery time.
4. The shopper confirms. Where checkout runs in the chat, the payment details travel as a token and your payment provider processes the charge.
5. The order arrives in your admin like any other order, and you fulfil it.

Every step after the first runs on data you supply, and an agent can't phone your shop with a follow-up question. If your feed says a jacket is in stock and the warehouse disagrees, the order fails at checkout. The shopper moves to the next option on the list. Shopify told investors that traffic from catalog-powered AI searches converts at twice the rate of general AI searches. That is a strong sign that structured product data, rather than clever copy, decides who gets the sale.

What the Universal Commerce Protocol does

The Universal Commerce Protocol, or UCP, is an open standard that lets any AI agent find out what a store can do and buy from it, without a custom integration for each assistant. Google and Shopify co-developed it with Etsy, Wayfair, Target and Walmart. More than 20 other companies endorsed it at launch, including Visa, Mastercard, Stripe and Best Buy.

The problem it solves is plumbing. Before UCP, each assistant needed its own connection to each store, which Google's engineers call an "N x N integration bottleneck". Ten assistants and 1,000 stores would mean 10,000 separate builds. UCP replaces that with one published file. A store lists its services and capabilities in a JSON manifest at /.well-known/ucp on its own domain, and any agent can read it to see which checkout, discount and delivery options the store supports.

Google's technical walkthrough of UCP also spells out the line store owners care about most: "you own your business logic, and you remain the Merchant of Record." The agent is a new front door, and the shop behind it is still yours.

At launch, UCP covered checkout, discounts, fulfilment, identity linking and order management. A March 19, 2026 update added a cart, so an agent can add several items from one store in one go. It also added catalog access for live variants, inventory and pricing. Identity linking lets a shopper keep the loyalty or member benefits they would get on your own site. UCP runs over REST APIs, the Model Context Protocol (MCP), Agent2Agent (A2A) and the Agent Payments Protocol (AP2), which lets platforms expose it with technology they run today.

ACP is the other standard to know. OpenAI built the Agentic Commerce Protocol with Stripe, and since March 2026 ACP also carries merchant product data into ChatGPT for discovery. Target, Sephora, Nordstrom, Lowe's, Best Buy, The Home Depot and Wayfair have connected their catalogs that way.

For most store owners, these protocols are the platform's job. A Shopify store gets UCP support without writing code. A custom-built store has to publish the manifest and checkout endpoints itself, which makes it a development project rather than a settings change.

What your store needs ready to be included

To be included in AI shopping, a store needs complete product data: title, description, image, price and live stock. It also needs published terms of service, privacy and return policies, a customer support contact and a checkout a platform can reach. Google's in-chat checkout excludes several product types on top of that, including subscriptions, personalised goods and pre-orders.

Product data comes first. Shopify asks for a title, description, image, price and available inventory on every product before it reaches AI channels. OpenAI's product feed spec has nine required fields, among them a stable item ID, the brand, the seller name and an availability value such as in_stock or backorder. Write descriptions as facts a machine can match: material, size range, compatibility, what is in the box. A line like "perfect for every adventure" gives an agent nothing to compare.

Price and stock have to match reality. An agent quotes what your feed says. Shopify's Google channel asks you to switch on automatic product and shipping sync with Merchant Center. On your own product pages, Google's merchant listing markup needs a name, image, price and currency, and recommends availability, shipping details, return policy and GTIN.

Policies must be published where machines can read them. Google's Merchant Center requirements for UCP checkout make a return policy mandatory, as a merchant of record requirement. It has to state the return cost and the return window and link to the full policy. You also need at least one support contact: a website, an email address or a phone number. Shopify requires terms of service, a privacy policy and a return and refund policy before you can sell through its agentic storefronts.

Not every product qualifies. Google treats each product as ineligible for checkout until you set its native_commerce checkout_eligibility attribute to true. Some products cannot qualify at all:

• subscriptions and instalment plans
• personalised or custom-made goods, refurbished items and pre-orders
• age-restricted products
• services, rentals and virtual items
• anything that needs special shipping or activation in a physical store

Products that carry a regulatory warning, such as California's Proposition 65, also need a consumer notice field. Shopify's Google channel adds a rule worth copying on every platform: put legal disclosures within the first 6,000 characters of the product description.

The checkout has to hold up after a handoff. On ChatGPT, and on Google when direct checkout is off, the shopper lands on your own checkout page straight from a chat. That shopper has already decided. A forced account sign-up or a surprise shipping charge at the last step loses a sale the agent had won for you.

Consider a homeware store with 1,200 products where 300 descriptions are one line long and 80 items show stock the warehouse no longer holds. Fixing those 380 listings will do more for its AI visibility than any protocol work. The same fixes improve its regular Google Shopping listings as well.

How to get your store into AI chats on any platform

The route in depends on your platform. Eligible Shopify stores are included by default through Agentic Storefronts, and the main condition is selling to US customers. Stores on other platforms can use Shopify's Agentic plan, which has no monthly fee. They can also connect directly with Google Merchant Center, OpenAI's product feed and, for Copilot, PayPal or Stripe.

On Shopify. Open the agentic section of the Shopify admin and check which AI channels receive your catalog. Then decide per channel whether shoppers pay inside the chat or on your store. Direct checkout is the default for AI Mode and Gemini, and you can switch it off to send every order through your own checkout. ChatGPT needs your store to sell to US customers, though the business itself can be based outside the US. AI Mode and Gemini currently need a US-based store selling to US customers.

On WooCommerce, Magento or a custom build. The Agentic plan is for brands that don't use Shopify for ecommerce. It adds your products to Shopify Catalog so they reach the same AI channels, with no monthly subscription fee. You pay Shopify Payments or third-party transaction fees when an order goes through Shopify-powered direct checkout. Setup asks for an external product URL on each item pointing to your existing store, a verified domain, and business verification with an official registration document.

Directly with each AI platform. Google's UCP checkout is open to select merchants for now. You need a Merchant Center account in good standing with approved free listings, a Google Pay & Wallet Console account and the technical integration. Then you register interest through Google's form. For Copilot Checkout, merchants on PayPal or Stripe apply through Microsoft, and Shopify stores are enrolled automatically after an opt-out window. OpenAI publishes a product feed spec for merchants who want their catalog in ChatGPT shopping.

Our view: on Shopify, getting in is mainly settings and policy checks. Elsewhere, start with Google Merchant Center rather than a custom protocol build. The same feed powers Google Shopping today, and Google's UCP checkout builds on it.

Why agentic commerce matters for store owners

Agentic commerce matters because AI assistants now send measurable orders. Shopify's AI-driven store traffic grew 8 times year over year in the first quarter of 2026, and AI-search orders grew faster still. Microsoft says that when shopping intent is present, paths to purchase that include Copilot are 194% more likely to end in a sale.

Microsoft's early Copilot Checkout data adds that paths including Copilot led to 53% more purchases within 30 minutes than those without it. Read these numbers with care. They come from the companies selling the channel, and the growth starts from a small base. Both still point the same way.

Forecasts go further. McKinsey estimates agentic commerce could orchestrate up to $1 trillion in US retail revenue by 2030, and $3 trillion to $5 trillion globally. Treat that as a direction for planning rather than a number for your budget.

The bigger change is how a shopper reaches you. A buyer who asks an assistant for a gift under $50 sees a handful of products rather than three pages of results. If yours isn't in that handful, the shopper never sees your homepage or your discount code.

Pros and cons of selling inside AI chats

Selling inside AI chats brings shoppers who are close to a decision at a low setup cost, and you keep your role as merchant of record. The price is a US-first rollout, less control over how your product is presented and platforms that still change their plans.

On the plus side:

• Shoppers arrive close to a decision, because the assistant has compared options and answered the basic questions.
• Setup on Shopify costs little: eligible stores are in by default and pay standard processing rates.
• You stay the merchant of record on Google's checkout, and Microsoft says merchants keep the transaction, the customer data and the relationship on Copilot Checkout.
• The work carries over. Clean feeds, clear policies and accurate stock also help Google Shopping, marketplaces and your own site search.

On the minus side:

• Coverage is US-first. Copilot Checkout runs only in the US, and Google's checkout reaches three countries.
• The assistant writes the summary of your product and picks the rival shown next to it. Your brand story gets a sentence, if that.
• Whole categories, from engraved gifts to subscription boxes, are shut out of Google's in-chat checkout.
• Rules move fast. OpenAI changed its checkout plan in March 2026, and a strategy built on one assistant's current feature can date quickly.
• Attribution is young. Shopify tags ChatGPT referrals in the admin, but an AI-influenced sale that finishes days later on another device is hard to trace.

Each drawback points to the same order of work. Fix data quality first, and spend on channel-specific projects only after a channel shows up in your sales reports.

Who should start now, and who can wait

Start now if you sell physical products to US customers. Shopify stores can move fastest, because the channel is live and on by default. Start the groundwork now if most of your customers are outside the US. Hold off on in-chat checkout if most of your catalog is personalised, subscription-based or services, which Google's rules exclude.

US-facing stores with standard products. Turn the channel on, then audit the 50 products that bring in most of your revenue. Check for full descriptions, correct stock, real photos and clear return terms. Doing this before the holiday season puts the fixes in place when shopping traffic peaks.

Stores selling mostly in India, the UK, Europe or the Middle East. In-chat checkout is still closed to most of you. If you also ship to the US, ChatGPT discovery through Shopify is open even when your business is based elsewhere. Google's checkout covers products eligible in Canada and Australia, with the UK announced for later. Use the time to get product data and policies right in Merchant Center, work that helps your Google Shopping listings today.

Personalised, subscription and service businesses. Google's checkout rules exclude your main products, and ChatGPT sends buyers to your site anyway. Keep product pages and structured data accurate so assistants can still describe and recommend you, and leave checkout integration until the rules change.

Custom-built stores with developers on hand. Read the UCP specification and Google's Merchant Center guide, publish accurate merchant listing markup on product pages, and register interest with Google. Building your own UCP endpoints makes sense once US, Canadian or Australian orders are large enough to repay the work.

AI assistants can only sell what they can read.

To get your product data, policies or checkout ready for these channels, tell the Techno Mantraa team about your store through the Get a Quote form on our contact page or at info@technomantraa.com.