Feature
Product search that understands the sentence, not the keywords.
Site search fails when a shopper describes what they want instead of naming it. "Something waterproof under £100" has no keyword match, so a normal search returns nothing and the shopper leaves.
What it does
Understands descriptions, not just nouns
Queries are embedded and matched on meaning, so "a warm coat for a toddler" finds the right products even when none of those words appear in a title.
Hybrid retrieval, not just vectors
Every query runs vector similarity, full-text search and fuzzy matching together, then merges the rankings. Vectors alone miss exact SKUs; keywords alone miss descriptions. Doing both is what makes it reliable.
Filters that come from the sentence
"Under £100", "in stock", "in blue" are read out of the question and applied as real constraints, rather than hoping the ranking happens to respect them.
Never sells what you cannot ship
Stock levels sync with the catalogue and out-of-stock items are excluded from results, so the assistant does not recommend something the shopper then cannot buy.
Learns your vocabulary
The words your shoppers use for your products are not always the words in your catalogue. Poor-outcome queries feed a nightly job that proposes synonyms specific to your store.
Answers with product cards
Results come back as cards with image, price and an add-to-cart button, inside the conversation, so the shopper never has to go back to a results page.
Questions about AI product search
How is this different from my existing site search?
Does it replace Elasticsearch or my search extension?
Can it find a specific SKU or product code?
Which platforms is AI product search available on?
Try it on your own catalogue.
Ask it the questions your shoppers actually ask and see what comes back, on your real products, in about two minutes.