Every Agentic Commerce Feature Shopify Has Shipped (and What Each One Actually Does)

Shopify's agentic commerce work sorts into four layers. Each answers a different question an AI agent asks about your store.

• Discovery — Catalog, Shopify's cross-store product index that agent partners can query, plus the auto-generated llms.txt file that points crawlers at your key pages.
• Live answers — the storefront MCP endpoint, which lets an agent search your catalog, read product details, and pull your policies over a standard protocol instead of scraping.
• Brand facts — Knowledge Base metaobjects, structured answers about shipping, returns, sizing, and the brand itself, editable in admin and served to agents.
• Transaction — agent checkout paths, including Shopify's participation in ChatGPT's Instant Checkout via the Agentic Commerce Protocol, so a conversation can end in a paid order.

The default coverage is real. A standard Shopify store today exposes machine-readable product data, a working MCP endpoint, and a generated llms.txt without the merchant doing anything. If an agent knows your domain, it can already query you.

That default is also the ceiling of the automatic story. Shopify syndicates what it has: titles, descriptions, prices, availability, images, policy pages. It cannot invent the things shoppers actually ask about — how the fabric runs, what pairs with what, whether it ships before Friday, why your version costs more than the knockoff.

Three gaps show up on nearly every store we probe. First: attribute depth. Agents answer comparison and fit questions from structured attributes, and most catalogs carry maybe a third of what their products actually offer — the rest is trapped in photos and PDF size charts.

Second: brand knowledge. The Knowledge Base metaobjects are only as good as what someone types into them, and on most stores nobody has typed anything. Agents fall back to crawling old blog posts and third-party chatter, which is how a store's 2023 shipping policy ends up quoted in 2026.

Third: measurement. None of these surfaces tell you what agents asked, what they were told, or which conversations became orders. The infrastructure sells; the reporting layer doesn't exist yet.

Treat the Shopify launches as the floor. Fill the Knowledge Base metaobjects and keep them synced with your actual policies. Push your real product attributes into structured fields, not just prose. Then test it from the outside: ask ChatGPT, Perplexity, and Gemini the questions your shoppers ask, and see what comes back.

That outside-in test is exactly what Kinect's agent-ready storefront work automates — it reads what agents can see about your store, fixes what they get wrong, and puts an AI sales rep on the storefront itself so the same intelligence sells on your site too. Brands with Kinect see 3–6% more revenue, measured against their own baselines.