Skip to main content

Enrich Products with New Attributes

How to have the AI assistant derive missing values (colour, gender, material...) per product and save them as new attributes you can map or filter on.

Sometimes a channel needs a value your store simply doesn't have - a clean colour, a gender, a material, a normalised product type. Instead of building a giant tangle of "if the title contains…" rules, you can let the assistant work out the value for each product once and store it as a new attribute. You then map or filter on it like any other store field.

Product enrichment (adding sources) requires the Additional Sources add-on on your plan. If it's not enabled, the assistant will tell you and point you to upgrade.

Just ask

Look at my products and work out a clean colour for each one from its title and description.
Save it as a new attribute called "ai_color", then let me map it to the channel's colour field.

Other examples: "Classify each product's gender (men/women/unisex/kids) into a new attribute." · "Extract the material into a field I can map."

How it works

  1. The assistant reads your products to reason over them.

  2. It works out one value per product and stores them under your new attribute name.

  3. It applies the change, and Koongo re-imports so the new field is available.

  4. You use it - map it onto a channel attribute, or reference it in a rule or filter.

Values are flat: one simple value per product (text, number, yes/no). That's exactly what a feed column needs. For big catalogues the assistant works in batches and applies at the end.

Manage your enrichment attributes

  • "List the extra attributes I've added."

  • "Delete the 'ai_color' attribute." (removes it on the next import)

Behind the scenes (for the curious)

The assistant reads products with preview_products, then stores values with set_api_source - a precomputed map of { productId: value } written straight to your data store as api_source.<code> (types are preserved; the last batch is sent with apply to materialise + re-import). An alternative, set_ai_source, delivers the same per-product values as a CSV and lands as custom_<code>. Inspect with list_api_sources / list_ai_sources; remove with delete_api_source / delete_ai_source. Both require the additional-sources add-on.

Did this answer your question?