When AI Stops Advising and Starts Buying
For the past couple of years, the conversation around AI in e-commerce has mostly been about discovery: shoppers asking ChatGPT or Perplexity for product recommendations, then clicking through to buy. That model is already changing faster than most merchants have noticed.
Agentic commerce is the next stage. Instead of an AI answering a question, an AI agent completes a purchase autonomously on a shopper's behalf. The user says something like "order me a 1kg bag of Colombian single-origin coffee, same quality as last time, under £20, delivered by Thursday" and the agent handles everything: searching, comparing, selecting, and checking out. The human never visits a product page.
This is not a distant future scenario. OpenAI's Operator, Google's Project Mariner, and Perplexity's Buy with Pro feature are all live or in limited rollout. Shopify announced its own agentic commerce API roadmap in late 2024. If you are running a Shopify store and you are not thinking about this now, you will be scrambling to catch up by mid-2026.
How AI Agents Actually Choose a Product
Understanding what an AI agent does when it makes a purchasing decision is the most important thing a merchant can internalise. It is not a keyword ranking. It is not a paid ad. It is a multi-step reasoning process that looks something like this:
- The agent interprets the user's intent, including constraints like budget, delivery date, and brand preferences.
- It queries one or more data sources: search engines, retailer APIs, structured product data, and its own training data.
- It evaluates candidates against the stated criteria and any inferred preferences from past behaviour.
- It selects the best match, then either presents it for one-click approval or completes the purchase directly.
Notice what is not in that list: the agent is not reading your beautifully designed homepage. It is not watching your brand video. It is parsing structured, machine-readable data about your products. If your product listings are missing key attributes, if your return policy is buried in a PDF, if your pricing is only visible after a customer logs in, the agent either skips you or gets the details wrong.
The merchants who win in agentic commerce are the ones whose data is clean, complete, and easy for a machine to read with confidence.
Structured Data Is the Language Agents Speak
This is where schema markup moves from a nice-to-have SEO tactic to an operational necessity. Schema markup, specifically JSON-LD implemented on your Shopify store, tells AI systems exactly what your products are, what they cost, whether they are in stock, what your returns policy is, and dozens of other attributes that a purchasing agent needs before it will recommend or buy your product.
Let's be specific about what agents are looking for:
Product and Offer schema
Every product page should have Product schema with a nested Offer object. This needs to include price, priceCurrency, availability (using the correct Schema.org values like InStock or OutOfStock), sku, gtin13 or mpn if you have them, and priceValidUntil. Missing any of these reduces the agent's confidence in your listing. An agent that is not confident in a data point will either default to a competitor whose data is complete, or flag the purchase for human review, which breaks the whole autonomous flow.
You can read more about implementing this correctly in our guide on how to use Offer schema to win AI shopping recommendations.
MerchantReturnPolicy schema
Return policy is a major trust signal for AI agents. If a shopper's agent cannot confirm your return terms, it will route the purchase to a retailer whose policy is explicitly machine-readable. Implement MerchantReturnPolicy schema with returnPolicyCategory, merchantReturnDays, and returnMethod clearly defined. Our detailed breakdown of MerchantReturnPolicy schema for AI shopping trust walks through exactly how to do this on Shopify.
Brand schema
Agents reason about brands, not just products. If you are a brand rather than a generic retailer, Brand schema helps AI systems associate your products with your brand identity, which becomes increasingly important as agents develop user-specific brand preferences. See our post on using Brand schema to protect your identity in AI search for the implementation detail.
What Shopify Merchants Need to Audit Right Now
Before worrying about agentic commerce APIs or headless storefronts, get your fundamentals right. Here is a practical audit checklist for any Shopify merchant:
Product data completeness
Go through your top 20 products and ask: does every one have a GTIN or MPN? Is stock availability updated in real time and reflected in your schema? Is the price displayed to a non-logged-in user, or is it hidden behind a login wall? Agents cannot retrieve data that requires authentication. If your wholesale or B2B pricing is gated, make sure your retail pricing is publicly accessible and schema-marked.
Delivery and fulfilment data
Delivery speed is one of the most common constraints in agentic purchase requests. Shopify merchants need to expose estimated dispatch and delivery windows in a machine-readable way. This is not yet standardised in Schema.org, but you can include it in your Offer schema using shippingDetails with deliveryTime properties. If you use a third-party logistics provider, ensure your shipping data is synced and accurate, because an agent that promises Thursday delivery based on stale data is a merchant's liability, not the platform's.
Reviews and ratings
AI agents use social proof as a tie-breaker. AggregateRating schema on your product pages feeds this signal directly to AI systems. If you have Shopify product reviews and they are not exposed via structured data, you are leaving a key differentiator invisible to agents. A product with 4.7 stars from 230 reviews, expressed in proper schema, will outcompete an equally priced competitor whose ratings are only visible in their store's HTML.
Page speed and crawlability
AI crawlers operate differently from Googlebot, but they still need to access your pages. If you have accidentally blocked AI crawlers via your robots.txt or a Cloudflare setting, your products will not be indexed for AI-powered systems at all. This is more common than merchants realise. Check your robots.txt and your Cloudflare firewall rules. You can get a quick read on your current AI visibility status with a free AI visibility audit.
The API Layer: What Is Coming in 2026
Shopify has signalled that it will offer native agentic commerce APIs, allowing AI agents to query product availability, initiate carts, and complete purchases programmatically. This is significant. It means that for Shopify merchants, the platform will eventually handle a lot of the heavy lifting at the transaction layer. But the data layer is still your responsibility.
Think of it this way: Shopify can provide the checkout API that an agent calls to complete a purchase. But the agent still has to decide to call your checkout API in the first place. That decision is made based on product data, trust signals, return policies, pricing, and brand reputation. None of that is automated for you.
Early adopters who have their structured data right, their trust signals in place, and their product data complete will be the ones that AI agents learn to prefer. Agents are trained on patterns of successful purchases. If your store consistently has accurate data and fulfilled orders, you build a kind of machine reputation that compounds over time.
Common Mistakes Merchants Are Making Right Now
Having looked at a lot of Shopify stores over the past 18 months, a few patterns come up again and again:
- Using Shopify's default schema and stopping there. Shopify's built-in structured data is basic. It covers the minimum but misses return policies, brand signals, detailed shipping data, and aggregate ratings. It is a starting point, not a solution.
- Updating product details on the front end without updating schema. Price changes, stock updates, and promotional pricing get changed in the Shopify admin but the schema either lags behind or was hard-coded and never updates. Agents that find a discrepancy between schema data and on-page data lose trust in the merchant fast.
- Assuming Google Search Console is the whole picture. GSC does not tell you anything about how AI systems read your store. You need to test your schema against AI-specific tools and understand how large language models are processing your product pages, not just how Googlebot indexes them.
- Ignoring brand entity establishment. If your brand is not well-defined as an entity across the web (Wikipedia, Wikidata, your Google Business Profile, consistent NAP data, and schema), AI agents have less confidence in who you are. For newer brands especially, entity establishment is a foundational step before anything else.
How FlinnSchema Approaches This for Shopify Merchants
At FlinnSchema, we work specifically on making e-commerce brands visible and trustworthy to AI systems. For Shopify merchants preparing for agentic commerce, that means a structured process: auditing what schema is currently live, identifying the gaps that matter most for AI purchasing decisions, implementing the missing structured data, and then monitoring how AI systems respond to the changes.
The merchants who see the biggest uplift are typically those who had been relying on Shopify's default schema and had no return policy, brand, or rating schema at all. Getting those three elements right often produces a measurable shift in how AI tools cite and recommend their products within a few weeks. You can see what that looks like in practice on our client results page.
Frequently Asked Questions
What is agentic commerce, in plain terms?
Agentic commerce is when an AI agent makes a purchase on a user's behalf, without the user visiting a website or interacting with a checkout. The user states what they want, and the agent finds, selects, and buys the product autonomously. It is the logical next step from AI product recommendations.
Do I need to do anything with Shopify's settings to prepare for AI agents?
The most important steps are outside your Shopify admin. You need to implement complete, accurate structured data (schema markup) on your product pages, ensure your store is not blocking AI crawlers, and make sure your product data (pricing, stock, shipping, returns) is accurate and publicly accessible without requiring a login. Shopify's native tools do not handle all of this for you.
Is schema markup only relevant for Google, or do AI agents actually use it?
AI agents use schema markup directly. JSON-LD structured data is one of the most reliable signals an AI agent can use to extract product attributes, pricing, availability, and trust signals without misreading the page's visual layout. Merchants with complete, accurate schema are significantly more likely to be selected by AI purchasing agents than those relying solely on HTML content.
How quickly will agentic commerce become mainstream for retail?
The honest answer is: faster than most projections. OpenAI's Operator and similar tools are already live for early adopters. Shopify's API roadmap suggests full agentic commerce support is a 2025 to 2026 priority. Merchants who wait until agentic commerce is mainstream to prepare will find themselves competing against stores that have been building machine trust for a year or more. The time to act is now, not when it becomes impossible to ignore.