Does Shopify Sidekick Use Real-Time Store Data?

"Real-time" means two different things depending on which assistant you're asking about, and conflating them causes real confusion. When you query Sidekick, you're reading your store's present state. When a shopper's AI assistant recommends you, it's reading a published version of your store that refreshes on its own schedule. Both matter; they just run on different clocks.

Sidekick reads your live data

Sidekick queries your Shopify admin directly. For most questions, that means it's working from your current records, not a snapshot from last week. Ask about inventory and it reflects your live stock. Edit a product title and ask about it, and the new title is what it reads.

The one nuance: some analytics roll up on a reporting cadence rather than updating to the second, so a sales report might reflect a slight lag. But the underlying catalog, order, and inventory data Sidekick reaches is your live data. For practical purposes, Sidekick answers from the present.

This is part of why it's a trustworthy tool for running the store: when you ask "did that price change go through," it's reading the actual current value, not guessing.

The buyer-facing assistants run on a different clock

Here's the distinction that trips merchants up. The AI assistants your shoppers use, ChatGPT, Shop, Gemini, do not query your live admin. They read a published, crawled view of your store and Shopify's Global Catalog. That published view refreshes on its own schedule, not the instant you make a change.

So you can fix a product today, confirm it in the admin, and have Sidekick reflect it immediately, while the buyer-facing assistants still describe the old version until their copy of your store catches up. The fix is real; the propagation takes time. We map the two-readers split in full in what data Sidekick can see in your store.

This is why "I fixed it, why doesn't ChatGPT show the change" is a question about propagation, not about whether the edit worked. Sidekick confirms the edit saved. It cannot confirm the outside world has re-read your store, because that's not data it can see.

Why keeping the structured layer current matters

If the outside assistants read a published version of your store, then the freshness of that published version is its own thing to manage. Structured data that drifts out of sync with your real catalog means crawlers keep reading the stale version until it refreshes. The discipline is keeping the structured-data layer current with the catalog you actually run, so the version the assistants see matches the store you operate.

That's the ongoing job AgentReady Signal does. Signal doesn't just emit Schema.org JSON-LD, agents.md, and llms.txt once; it maintains them so they track your live catalog, keeping the published view the outside assistants read aligned with your real store. It's the difference between structured data as a one-time setup and structured data as something that stays true.

And when you want to confirm a fix actually reached the outward layer, you don't ask Sidekick, which only sees your admin. You re-check the published view. AgentReady's Sidekick extension lets you ask, read-only, where you stand and whether a recent fix has landed, covered in asking Sidekick whether you show up in AI search. The answer comes from readiness data, not from your admin alone, and nothing it returns changes your store; you confirm any action yourself.

The takeaway

Sidekick reads your live data, so its answers reflect recent changes almost immediately. The assistants that recommend you to shoppers read a published version of your store that refreshes on its own clock, so a fix shows up there only once it propagates. Knowing which reader you're asking about saves you from confusing "the edit saved" with "the world has seen it." To check how the outward-facing view reads right now, the free AI-readiness checker scans any storefront in about a minute.