Magento 2 AI Catalog Search: search that understands the shopper
The module plugs AI into Magento's standard search box: a query like "red jacket for men under $50" turns itself into color, category and price filters.
TypeOpen-source Magento 2 module
StackPHP 8.1, Magento 2.4, Elasticsearch/OpenSearch, LLM APIs
LicenseMIT
Year2026
Task and solution
Magento's native search matches words literally: it doesn't know that "red" is a color, "for men" is a catalog section and "under $50" is a price cap. Shoppers have to open the filters themselves, and some just leave.
The goal was for a plain text query to return filtered results right away, without breaking anything in the theme or catalog settings.
- AI parses the query into keywords, attributes, category and price, in any language
- Native mode: Magento's regular results page with the filters already applied
- AI mode: a dedicated page with refinement chips built from the products actually shown
- Semantic matching: "warm winter jacket" finds an "insulated parka"
- If the AI is unavailable, search works as usual and the shopper never notices
- Parse caching for identical queries and a query log with the cost of every call
- The module only reads the catalog's filter settings and never changes them
The module in numbers
2result display modes
3filter types from one query: attributes, category, price
0changes to catalog settings
Want the same result?
Tell me about your task or estimate the cost in the calculator.