By Steve Merrill, Founder of WRKNG Digital | August 15, 2026
What does an AI shopping agent actually check on your store?
An AI shopping agent checks six things before it recommends or buys: what the product is, whether it is in stock and what the real price is, whether it can trust you, how you compare to alternatives, whether it can check out programmatically, and whether your data is machine-readable. It never sees your pretty homepage. It reads your data.
OpenAI and Stripe shipped the Agentic Commerce Protocol so agents can buy inside ChatGPT. Shopify plugged into it. That means an agent is already asking these questions of your store. Here are the six, and how a Shopify store answers each one in data.
1. What exactly is this product?
The agent needs a clean identity: title, brand, category, GTIN, variant, and attributes. It looks at your Shopify product data and your Google Merchant feed. If your title reads "Blue Shirt V2 FINAL" instead of "Bella Canvas Unisex Cotton T-Shirt, Navy, Large," the agent cannot match you to the shopper's request. Vague identity means you get skipped before the race starts.
2. Is it in stock and what is the real price?
The agent wants the live total, not the sticker. It reads your availability and price fields from the feed and your Schema.org Offer markup. If your feed says in stock but you are actually sold out, or your $30 tee becomes $47 after shipping, the agent either passes you over or completes a broken order. Accurate stock and honest all-in pricing are non-negotiable. That's it.
3. Can I trust this merchant?
Before an agent spends someone's money, it checks your risk signals: return policy, reviews, ratings, and shipping terms. It reads AggregateRating schema and your policy pages, and Google's Merchant Center feeds store-level trust data straight into the systems these agents lean on. No visible returns policy and no reviews reads as risk. Agents route around risk.
4. How does it compare to the alternatives?
The agent lines you up against every other option and picks. It compares price, shipping speed, ratings, and specs across feeds in seconds. This is where thin data kills you, because the competitor with a full spec sheet and 400 reviews wins the tie every time. Fill in every attribute the agent can compare on. The blanks are losses.
5. Can I complete checkout programmatically?
The agent needs to buy without a human clicking through your cart. The Agentic Commerce Protocol defines how an agent submits an order, and Shopify's Catalog and checkout systems handle that plumbing for stores on the platform. If checkout only works through a manual browser flow, the agent stops at the recommendation and the sale evaporates. On Shopify, this part is largely handled for you. Don't break it with custom checkout hacks.
6. Is your data machine-readable?
Every question above depends on this one. The agent needs a clean product feed, Schema.org Product markup on every page, and ideally an llms.txt file that maps your key pages in plain text. Structured data is the difference between an agent reading your store and guessing at it. Guessing means it moves on.
How we chose these six questions
These map directly to what published agent and commerce specs require to complete a purchase: identity, price, stock, trust, comparison, and checkout, all delivered as structured data. They come from the Agentic Commerce Protocol, Shopify Catalog docs, Schema.org, and Google Merchant Center, not opinion.
FAQ
Q: What does an AI shopping agent check on my store?
It checks product identity, live stock and the real total price, merchant trust signals like returns and reviews, how the item compares to alternatives, whether checkout runs programmatically, and whether your data is machine-readable through a feed, schema, and llms.txt.
Q: Do I need the Agentic Commerce Protocol to sell to AI agents?
Not if you are on Shopify. Shopify handles the agent-facing checkout plumbing. Your job is clean product data, accurate stock and pricing, and structured markup so the agent can read and trust your listing.
Q: What structured data does an AI agent read on a product page?
Schema.org Product markup with offers, price, priceCurrency, availability, and AggregateRating, plus a clean Google Merchant product feed. That gives an agent identity, price, stock, and trust in a format it can parse without guessing.
Q: Does llms.txt actually help my store get picked?
It gives AI crawlers a clean map of your key pages in plain markdown, so they parse your site with less effort. It is not a ranking guarantee, but it removes friction. Friction is what makes an agent skip you.
Q: How do I make my Shopify store ready for AI shopping agents?
Fix your product feed so titles, prices, and stock are accurate. Add Product and Offer schema to every product page. Keep returns and reviews visible. Publish an llms.txt file. Do those four things and you answer every question an agent asks.
Your store is already being read by machines that decide what to buy. Most stores are failing questions two, three, and six right now and don't know it. See how WRKNG Digital makes your Shopify store answer every one of these questions in data.
