Shopify opens up to AI agents: what changes for sellers
Shop MCP lets AI agents search for products, build carts and complete purchases with Shop Pay. Payment requires the buyer’s confirmation. For sellers, the priority is consistent product data, images and configuration rules.
TO DECISION.
THE ANNOUNCEMENT
What’s new: from catalogue to payment
On 13 October 2026, Shopify announced Shop MCP in its developer changelog. MCP stands for Model Context Protocol: the open standard connecting AI assistants to external tools. A single connection lets an agent search Shopify Catalog, build a cart and complete payment with Shop Pay.
According to Shopify, the catalogue contains over one billion products. The connection provides tools for search and product details, carts, checkout and orders. Search works without an API key; one can be created in the Dev Dashboard.
The announcement fits the picture Shopify presented on 6 October: the 2026 holidays as the first season of shopping with AI agents. That is the company’s assessment. For an Italian business, the concrete development to evaluate is another point of contact between catalogue, assistant and customer.
OUR ASSESSMENT
Who decides and who completes the order
Signing in with Shop lets search use saved preferences and checkout use saved addresses and Shop Pay. Carts and checkout also work without sign-in. At Shopify stores, the agent completes the order after the buyer confirms the total and payment method.
Orders placed through an agent appear in the seller’s Shopify admin like other orders. There are no webhooks at launch: status must be checked by querying the checkout. Existing integrations remain unchanged. These details matter when deciding how to track the outcome of a purchase.
| What the agent does | What remains with the buyer | What remains with the seller |
|---|---|---|
| Searches and compares products in the catalogue | Assess the recommendation and choose the product | Maintain data, variants, availability and content |
| Prepares the cart and checkout | Confirm the total and payment method | Manage orders in their admin |
| Reads checkout status and searches orders | Approve the purchase as required by the journey | Organise support and after-sales information |
OUR ASSESSMENT
Product data: comparison starts before the visit
Our assessment: a business selling online or through retailers needs to make its products understandable even when the first comparison happens inside an assistant. The customer can describe a need and receive a selection before visiting a product page. The agent compares the information it finds.
Variants, dimensions, materials, compatibility and availability therefore become crucial information. A generic description leaves questions open that a well-structured catalogue can resolve. If the same finish has different names on the website and in retailers’ product sheets, comparison becomes ambiguous. A shared foundation is needed, with recognisable, up-to-date attributes.
For example, a lamp should specify dimensions, fitting and available finishes; a chair, the dimensions and materials of each variant. This is a criterion for preparing the catalogue, not a guarantee that an agent will recommend it. The work involves the people who know the product and those managing its sale.
OUR ASSESSMENT
Photos and 3D support the decision
Images and content also gain importance along this path. When an assistant suggests two products, the customer still needs to decide which to choose: photographs, 3D views and variants make the differences described in a product sheet visible. The useful measure of quality is fidelity to what will be delivered.
AI product photography needs reliable references: colour, texture and proportions must match the real product. A product configurator can show the options; augmented reality for e-commerce helps assess its presence in a space.
Showing every variant correctly helps reduce uncertainty and potential returns linked to incorrect expectations. An attractive image depicting a combination that cannot be purchased instead creates a gap between promise and order. Content should be checked alongside the data, rather than approved solely for its visual impact.
OUR ASSESSMENT
Configurable products: rules come first
Selecting a size differs from composing a product with interdependent options. Materials, accessories and dimensions can have permitted combinations, exclusions and prices for each variant. Before entrusting selection to an AI channel, those rules must be explicit and consistent with the system accepting the order.
The 3D configurator is the place to define and show those relationships. For example, an accessory may be available only with a particular frame: that constraint must apply both to the visual experience and to the commercial data. Our guide to planning a 3D e-commerce configurator helps distinguish assets, logic and integrations.
OUR ASSESSMENT
Support remains the company’s responsibility
External agents can bring customers; questions about manuals, compatibility, support and after-sales service still reach the company. A business AI assistant based on approved content enables answers using the correct information for its products.
The buyer confirmation retained by Shopify is also worth considering when designing your own journey. The business question is which AI channels to accept and with what information. Defining responsibilities, catalogue updates and order measurement helps make a considered decision, without having to be present at any cost.
WHERE TO START
What to do now: three steps
1. Check product sheets
Review attributes, variants, dimensions and availability. Compare the website with retailers’ materials; assign an owner to missing or inconsistent information.
2. Verify the images
Check that photographs and 3D accurately show every variant. Test the journey for a real combination, from visual selection to product sheet and cart.
3. Decide channels and measurement
With whoever manages your e-commerce, assess which AI channels to activate and how to identify orders arriving through them. Availability and conditions must be checked before activation.
Prepare products and content for new shopping journeys
Bring us a product family and your available materials: we can assess data, images, configuration and support against your customer’s journey.
Frequently asked questions
What is Shop MCP?
It is an MCP connection that lets an AI agent search Shopify Catalog, build carts and complete payment with Shop Pay.
Can an agent pay without confirmation?
At Shopify stores, the buyer must confirm the total and payment method before the agent completes the order.
Is Shop MCP available in Italy?
Shopify does not state in the changelog which markets and countries it is available in. This source therefore cannot confirm availability in Italy.
What does it cost, and can sellers opt out?
Shopify does not state costs or fees in the changelog, nor seller settings for opting out. These conditions should be checked before deciding.
Is an API key needed to search for products?
Search alone does not require an API key. A key from Shopify’s Dev Dashboard provides higher usage limits.
Sources and update
Shopify Developer Changelog · 13/10/2026 · Shopify News · 06/10/2026
Updated on 18 October 2026. AI-generated illustration; this is not a Shopify screenshot.
