How the AI searches products
Understand how Interlinked prepares catalog information for relevant product answers and what to check when a product is hard to find.
Product search uses meaning and detail
After a catalog change is confirmed, Interlinked prepares the product information so an agent can look for relevant items during a customer conversation. The agent can use more than an exact product-name match: it can relate a customer’s need to descriptions, categories, attributes, availability, and price.
Catalog visibility and search readiness are separate. A product can appear in the Catalog table while preparation is still in progress.
Examples
| Customer asks | Helpful catalog information |
|---|---|
| “What is your largest protein presentation?” | Size, weight, presentation, and description |
| “What is the most affordable option?” | Current prices and availability |
| “Do you have something for sensitive skin?” | Description, category, skin type, and other attributes |
| “I need a black formal shirt in medium.” | Category, color, size, and description |
The agent still needs a clear product, price, and availability state to give a useful answer. Good search does not correct inaccurate catalog data.
Timing expectations
After a large import, merge, replacement, or manual change, preparation can continue after the table updates. The time depends on catalog size and the amount of detail. Wait for the visible processing state to settle before treating a broad test as final.
Images are handled separately in Products → Media. A missing image does not necessarily prevent a text product answer or storefront publication.
If a product is hard to find
- Confirm the product is in the active Catalog view, not only in a staged import.
- Check its name, price, availability, SKU, and important attributes.
- Add a concise description or clearer category when the customer’s wording is not reflected in the catalog.
- Allow preparation to finish after a large change.
- Test a specific customer question again in a controlled conversation.
For data quality, read Product catalog best practices. For import state, read Review and confirm a product import.
Related in Interlinked
Connect this documentation topic to the relevant product, workflow, or commercial context.