When a shopper asks the AI Chat Bot about a product, the bot doesn’t make up an answer, it pulls from a snapshot of your store data called AI Memory. This article explains which content the chat indexes, where you can see what’s in the index, and how to force a sync after you’ve updated a product so the bot answers from the latest data.
Prerequisites
- StoreAgent AI for WooCommerce installed, activated, and connected to your StoreAgent.ai account.
- The AI Chat Bot enabled in StoreAgent → Settings → AI Chat → General.
Which content the chat can access
The chat reads from three post types by default, and you can enable others. To configure what’s indexed, go to StoreAgent → Settings → AI Chat → Post Types:
- Products: names, descriptions, prices, categories, tags, attributes, and any custom meta fields you’ve added via the Metadata Keys to Include field on the Products panel.
- Pages: your static pages (about, FAQ, shipping, returns). Off by default; enable per-post-type via the toggle.
- Posts: blog posts. Off by default; enable to let the chat answer from blog content.
For each post type you can also exclude specific items via the Exclude [Post Type] search field on its panel, useful for hiding archived, internal-only, or restricted-content posts from the chat.
Checking what’s indexed
The Post Types tab shows an indexing summary next to each enabled post type: a fraction (e.g. 19/19) that tells you how many items of that post type are in the index against the total available. When the two numbers match, every published item is indexed; when they don’t, some are still being processed or have failed.
For Products specifically, you also get a per-product view: the WooCommerce Products list has an AI Memory column showing each product’s current sync status (Pending, Synced, Failed, or Excluded). See How StoreAgent AI Memory works for the full breakdown of statuses and what to do about each.
Triggering a manual sync
StoreAgent re-uploads a product to AI Memory automatically a few minutes after you save it (a 3-5 minute debounce window). If you can’t wait, for example you’ve just fixed a typo a customer flagged in chat, trigger a manual sync.
Sync a single product
- Go to Products → All Products.
- Tick the product you’ve just updated.
- From Bulk actions, choose Add to AI Memory, then click Apply. The status flips to Pending, then Synced once the upload completes.
Sync an entire post type
- Go to StoreAgent → Settings → AI Chat → Post Types.
- Click the sync icon (a circular-arrow icon) next to the post type’s enable toggle. This queues an upload for every published item of that post type.
Manual sync also helps when the index is partially out of date after a long period of bulk imports, for example after migrating a catalog or importing a product CSV.
What the chat can and can’t reference
The chat answers from indexed content only. That means:
- Yes: any product, page, or post you’ve enabled in Post Types settings, including custom meta fields you’ve explicitly added to Metadata Keys to Include.
- Yes (with the Wholesale Suite integration on): wholesale-role-specific pricing per the customer’s role. See How to integrate StoreAgent with WooCommerce Wholesale Suite.
- No: draft, pending, password-protected, or trashed posts. The index only stores published items.
- No: custom-field data unless you’ve added the meta key to the per-post-type Metadata Keys to Include field.
- No: external websites or third-party services. The chat does not browse the web or pull answers from outside sources.
How the chat gets exact prices and order details
Most answers come from the indexed content described above. For a few details that change often, the AI Chat Bot also does a live lookup in your store at the moment a shopper asks.
- Variable product prices, stock, and variation details. A general answer may quote a starting price or a price range from the index, while the exact price, stock status, and SKU of a specific variation are fetched live when the shopper asks about that variation.
- The shopper’s own order. When a logged-in shopper asks about their order, the chat looks it up live. It only ever returns that shopper’s own order.
These live lookups are read-only and stay within your own store. The chat still never calls external websites or third-party services.
Troubleshooting
The chat returns outdated product details
The 3-5 minute debounce means there’s a small window where chat answers lag behind product saves. If a stale answer persists beyond 10 minutes, sync the product manually via Bulk actions → Add to AI Memory. If it’s still stale, check the AI Memory column on the products list, a Failed status means the upload errored and won’t retry automatically.
The chat answers from a page I don’t want it to reference
Go to StoreAgent → Settings → AI Chat → Post Types, find the post type’s panel, and add that specific page to the Exclude [Post Type] search field. The chat will stop referencing it on the next sync. Alternatively, disable the whole post type’s toggle if you don’t want any of its items in the index.
Custom field values aren’t appearing in chat answers
By default, only standard WooCommerce product fields (name, description, price, etc.) are indexed. To include custom meta fields, list each meta key on the post type’s panel in the Metadata Keys to Include text area (comma-separated). After saving, run a manual sync on affected products so the new fields land in the index. Custom fields are read from the product or post itself, not from individual variations, and content is indexed as written, so text that only appears after a shortcode or page builder renders on the front end is not part of the index.
Frequently asked questions
Does the chat see customer personal data?
Customer accounts and personal data are never added to the index. The chat can look up a logged-in shopper’s own order when they ask about it, but only their own order, and only coarse location details (city, state, postcode, and country). It never returns a customer’s name, email, phone number, full street address, payment details, or anyone else’s orders. It does see the logged-in user’s role, so it can apply wholesale pricing.
What data does the AI Chat Bot actually use?
Two things: the store content you choose to index (your products, and any pages or posts you enable), and the messages in the current conversation. That is separate from your website’s general privacy policy, which covers site-wide topics like cookies, IP addresses, and analytics. Those tracking topics are not what the AI Chat Bot works from. It answers from your published store content and the conversation itself.
How big is the index?
It depends on your plan’s vector item allowance. The free StoreAgent Lite plan includes 100 memory items; paid plans include larger allowances. See How StoreAgent calculates your usage for the math, and your StoreAgent.ai dashboard for current usage.
What if I exceed my vector item allowance?
New items aren’t indexed once you’re over the limit, and chat answers may reference older versions of products that have changed. You’ll see a notification on the StoreAgent dashboard when you’re approaching or have exceeded the limit. The fix is to either remove items from AI Memory (Bulk actions → Remove from AI Memory) or upgrade your plan.
Need help?
If you have a question or run into any issues, we’re here to help.
- Premium users: Open a support ticket
- Free users: Visit our community forum
