By Steve Merrill, Founder of WRKNG Digital | August 15, 2026
Is your Shopify store ready for agentic checkout?
Most aren't. The seven signs a Shopify store will choke when an AI agent tries to buy on its own are: no structured product data, price and inventory that aren't machine-readable, no llms.txt, gated product info, ambiguous variants, a checkout built only for humans, and trust signals buried in images instead of text.
Here's the thing. AI agents are starting to shop for people. OpenAI shipped the Agentic Commerce Protocol with Stripe, and Shopify is building agentic commerce into its Catalog. When an agent can't read your store, it skips you and buys from the one it can read.
I built a $10M ecommerce brand over 15 years. I've had a hard time watching store owners assume Shopify handles all of this for them. It doesn't. Here are the red flags.
1. Your product data isn't marked up with Schema.org Product data
What it looks like: your product page shows the price and specs to a human, but there's no Schema.org Product markup in the code. An agent reads code, not pretty pages. Without it, the agent has to guess your price, brand, and availability, and agents don't guess. They move on. The fix is simple: add valid Product structured data with name, price, priceCurrency, and availability on every product page.
2. Price and inventory aren't machine-readable
What it looks like: your stock level lives in a "Only 3 left!" banner rendered by a script, and your sale price only appears after a countdown loads. An agent evaluating you needs a clean, current number it can trust before it commits a purchase. Shopify's Catalog is built to expose price and availability in a structured feed. If yours is trapped in JavaScript widgets, the agent sees nothing. Feed it a real feed.
3. You have no llms.txt file
What it looks like: there's no llms.txt at the root of your domain. This file is a plain-text map that tells AI models which pages matter and where to find clean product info. It's an emerging standard, not a law, but it's a cheap signal that costs you nothing. No llms.txt means the agent crawls your site blind and gives up faster.
4. Your best product info is gated behind clicks, tabs, or scripts
What it looks like: your sizing, materials, and shipping details only appear after someone clicks a tab or opens an accordion. A human clicks. An agent often can't, or won't wait for the script to fire. If the answer to "does this ship free and fit a size 10" isn't in the plain HTML, the agent treats it as unknown. Put the load-bearing facts in visible, crawlable text.
5. Your variants are ambiguous
What it looks like: your color is "Midnight" and your size is "M/L," with no clean mapping to a standard value. A human infers what Midnight means. An agent buying the wrong variant is a returned order and a refund. Name variants clearly, map sizes to real measurements, and expose each variant as its own structured option so the agent picks the exact SKU the shopper asked for.
6. Your checkout assumes a human is driving
What it looks like: a captcha, a forced account creation, an upsell popup, or a multi-step flow that only makes sense to a person with a mouse. The whole point of the Agentic Commerce Protocol is letting the agent complete the purchase without a human clicking through. Every human-only gate is a wall the agent hits and abandons. Support agent-friendly checkout paths and kill the friction that only a person can clear.
7. Your trust and return signals aren't stated in plain text
What it looks like: your return policy is a JPG, your reviews load from a third-party widget, and your guarantee lives in the footer as an image. Agents weigh trust before they buy, same as people. If your return window, warranty, and rating aren't in readable text, the agent can't factor them in and defaults to the competitor who spelled it out. State the policy in words the machine can read.
How we scored these signs
Each sign maps to a specific thing an AI agent needs to browse, evaluate, or complete a purchase, drawn from Shopify's Catalog docs, OpenAI's Agentic Commerce Protocol, and Schema.org's Product spec. If a sign blocks one of those three steps, it made the list.
FAQ
Q: What is agentic checkout on Shopify?
It's when an AI agent browses, evaluates, and buys for a shopper instead of a human clicking through the cart. Shopify supports it through Catalog and its agentic commerce work, and OpenAI's Agentic Commerce Protocol defines how the agent and store communicate.
Q: How do I know if my store is ready for AI agents?
Check whether your product data, price, and inventory are machine-readable, whether you have an llms.txt file, and whether your variants and return policy are in plain text. If an agent needs a human to read any of it, you're not ready.
Q: Does Shopify handle agentic commerce for me automatically?
Shopify gives you the plumbing, but it doesn't clean messy product data, add missing markup, or fix vague variants. That part is on you.
Q: What is llms.txt and do I need one?
It's a plain-text file at your domain root that points AI models to your key pages in a clean, readable form. It's an emerging standard and a low-cost way to make your store easier for agents to parse.
Agentic checkout is coming whether your store is ready or not. If you want a straight answer on where your Shopify store stands, get an agentic readiness audit at WRKNG Digital.
By Steve Merrill, Founder of WRKNG Digital | August 15, 2026
