Shopify launches · Jul 2, 2026 · 2 min read
Shopify's AI Knowledge Base Metaobjects: The Brand Facts Agents Quote
Shopify's Knowledge Base metaobjects let you write structured answers — shipping, returns, sizing, brand story — that AI agents serve verbatim. How they work, what to put in them, and the sync trap to avoid.
When an AI assistant answers "what's their return policy?" about your store, it's going to quote something. Shopify's Knowledge Base metaobjects exist so that the something is written by you — structured Q&A-style brand facts, managed in admin, served to agents through Shopify's agentic surfaces.
This is the highest-leverage, lowest-effort piece of the whole agentic stack, and on most stores it's sitting empty.
What they are
Metaobjects are Shopify's custom structured-content primitive. The Knowledge Base flavor is a set of entries — think "Shipping," "Returns," "Sizing," "About the brand," "Care instructions" — each holding a canonical, plain-language answer. Agents and AI surfaces that integrate with Shopify can pull these directly instead of parsing your policy pages and guessing.
The design insight is that agents don't want your beautifully art-directed shipping page. They want the answer: "Free over $75, ships in 1–2 business days from Ohio, 30-day returns, final-sale items excluded." The Knowledge Base is where that sentence lives.
What to write into them
Prioritize the questions that show up in your support inbox and your pre-purchase chats:
- Shipping: costs, thresholds, carriers, real delivery windows, international rules.
- Returns and exchanges: window, condition rules, exclusions, who pays return shipping.
- Sizing and fit: how your sizing runs, conversion guidance, the one sentence your reviews repeat.
- Product care and materials: the facts that trigger "is this real leather?"-type questions.
- Brand: founding story, where products are made, what you're known for — the identity facts assistants otherwise pull from whoever wrote about you last.
Write like you're answering a customer, not drafting legal copy. Agents quote what they're given; give them sentences you'd be happy to see in a ChatGPT answer with your brand's name on it.
The sync trap
The failure mode isn't empty Knowledge Bases — it's stale ones. The day your holiday shipping cutoff changes on the policy page but not in the metaobject, you have two truths, and the agent will confidently serve the wrong one. Assign ownership: policy changes update both places or they don't ship.
This is also where a managed layer earns its keep. Kinect distills brand knowledge from your store, keeps it synced as things change, and serves the same facts to shoppers through the on-site sales rep — one knowledge base, both audiences. When a merchant updates a policy, the rep and the agents hear about it together.
Frequently asked questions
Do Knowledge Base metaobjects affect ChatGPT's answers about my store?
They feed Shopify's agent-facing surfaces, which is one of the channels assistants draw on. They don't override what crawlers find elsewhere — which is why keeping them consistent with your public pages matters.
How is this different from my FAQ page?
An FAQ page is prose an agent must parse and may misread. Metaobjects are structured fields served as data. Keep both, but treat the structured version as canonical.
Who should own updating them?
Whoever owns policy changes. The workable rule: no policy or shipping change is 'done' until the Knowledge Base entry matches. Kinect automates this sync for the brands we run.
Related reading
See Kinect on your store
An AI sales rep on your storefront and an agent-ready store behind it — trained on your catalog and policies, live the same day, measured against your own baseline.