Walk through shopperGPT.

How to open shopperGPT

Click the shopperGPT search icon in the store header. The search panel opens as a single unified interface with four components:

1Natural Language Search

shopperGPT understands intent, not just keywords. The more detail you give it, the better the results.

Notice: each result includes a personalized conversational response — shopperGPT addresses the shopper directly based on what they described. This is the concierge experience that builds brand affinity.

Example searches (built-in)

Click in the text box without typing anything. A dropdown appears with 5 rotating example searches — automatically generated from this store's catalog during setup. These give new shoppers a starting point and show off what natural language search can do.

Voice search

Tap the microphone icon and speak your search. Try: “midi dress” or “maxi dress.”

shopperGPT automatically builds a unique synonym and antonym database for each merchant, derived from their specific catalog during enrichment — not a generic word list. Try “mini dress,” “midi dress,” and “maxi dress” — then try “short dress,” “medium dress,” and “long dress.” Results match across all of them, even if no products are tagged with those exact terms.

Price ranges in plain language

No need to use the price filter. Just include it in your search:

shopperGPT parses the range from natural language and filters accordingly.

2Color Search

Click the Color tab to open the full-spectrum color picker.

• Select any shade from 16 million colors
• shopperGPT creates an embedding of your selection in real time and finds the closest color matches in the catalog
• No product tags required — it works from the actual colors in the product images

The smart color-text override

1. In the text tab, search: black mini dress
2. Open the color picker and select a bright red
3. Watch the results — you'll get red mini dresses, even though the word “black” is still visible in the text field

shopperGPT detects the conflict between the typed color and the selected color, removes the color word from the text query, and gives you what you actually mean. This is multimodal search: text + color working together intelligently.

3Image Similarity Search

Click the Image tab.

• Upload any photo or screenshot — a product image, an Instagram post, a magazine clipping
• Select the area you want to match (crop to the item)
• Hit Find Matches

shopperGPT evaluates the image and searches the catalog for the closest visual matches.

⏱️ Image searches typically return results in 6–8 seconds. Text and color searches return in ~3 seconds or less. Repeat searches are cached and usually return in under 1 second.

4Filters

Use the filter panel alongside any search to narrow results:

• Gender — women's / men's / unisex
• Price range — drag the slider or type in natural language (see above)

Search performance

Searches are cached both per-user and universally — so popular queries get faster over time for everyone.

Time to discovery is a core shopperGPT value proposition. Traditional keyword and category-based navigation typically takes shoppers 2–3 minutes of browsing to reach a product that might match what they had in mind. shopperGPT gets there in seconds — less friction, larger carts, higher conversions.

What merchants get on install

When a merchant installs shopperGPT, the platform automatically:

• Processes every product through an AI enrichment pipeline (text + image analysis)
• Generates embeddings for natural language, color, and visual similarity
• Creates 10 store-specific example searches (5 rotate per session)
• Builds a unique synonym and antonym database from each merchant's catalog

No manual tagging. No developer work. One-click install.