Product training lets you feed your chatbot structured product information so it can answer customer questions about your catalog — pricing, availability, descriptions, and more.
How to get there: Go to Setup → Chatbot in the top menu → click your chatbot → Products tab (under the Training section in the sidebar).
Adding Products
There are three ways to add products:
Manual Entry
- Click Add Product
- Fill in the product details:
- Source key (required) — A unique identifier like a SKU or product code
- Title (required) — The product name
- Description — Detailed product description
- URL — Link to the product page
- Thumbnail URL — Link to a product image
- Price — Product price
- Availability — In stock, out of stock, or preorder
- Active — Whether the chatbot should use this product
- Brand and Category — For organization
- Extra attributes — Additional details (e.g., "Origin: Colombia, Weight: 250g")
- Click Save
CSV Import
- Click Import CSV (or download the template first)
- Upload a CSV file (max 10MB, up to 10,000 rows)
- Required columns:
source_key and title
- Optional columns:
description, url, thumbnail_url, price, availability, is_active, brand, category, extra_attributes
- Review the import results — any errors are shown with details
Products are matched by source_key. If a product with the same source key already exists, it is updated.
Google Sheets / CSV URL Sync
- Open the Sync from Google Sheets or CSV URL section
- Paste your Google Sheets URL or a direct link to a CSV file
- Click Sync now to import
- Optionally set an auto-sync frequency (hourly, daily, weekly, or monthly) to keep products up to date automatically
When syncing, new products are created, existing ones are updated by source key, and products no longer in the sheet can be removed automatically.
Managing Products
Product Table
The main view shows all products with:
- Thumbnail — Product image preview
- Title — Product name
- Brand and Category
- Price and Availability
- Active toggle — Quickly enable or disable a product
- Actions — Edit or delete
Use the search bar and filters (availability, active/inactive) to find specific products.
Embedding Status
At the top of the page, the embedding stats show:
- Indexed — Products the chatbot has learned and can use
- Processing — Products currently being indexed
- Failed — Products that had an indexing error
- Not indexed — Active products waiting to be indexed
If you see failed or unindexed products, click Re-index to retry.
Bulk Actions
Select multiple products using the checkboxes and use Bulk Delete to remove them.
Export
Click Export to download all products as an Excel file.
What the Chatbot Learns
The chatbot indexes the product title, description, brand, category, URL, and extra attributes for answering questions. Price and availability are included in responses but don't affect search matching — this means updating a price won't require re-indexing.
Training Credit Usage
Products count toward your training pages quota. Every 5,000 characters of product content uses 1 training page credit (minimum 1 per product).
Tips
- Use descriptive titles — The title is the primary way the chatbot identifies products
- Add extra attributes — Details like dimensions, materials, or specifications help the chatbot give more complete answers
- Keep products active — Inactive products are removed from the chatbot's knowledge. Use this to hide discontinued items.
- Use auto-sync — If your product catalog changes frequently, set up auto-sync from Google Sheets to keep the chatbot current
- Source keys matter — Use consistent, unique identifiers (like SKUs) so imports and syncs can match products correctly