Product analysis with AI
Product analysis with AI Find products by image, URL, or text

📖 Documentation

How to use Product analysis with AI to find and compare products by image, URL, or text.

Getting started

Product analysis with AI lets you search across connected stores using three input types:

  • Image — Upload a product photo or drag & drop; we find visually similar items.
  • URL — Paste an image URL; we fetch the image and search by it.
  • Text — Type a product name or keywords; we search by semantic similarity.

Results are grouped by store. Use the contact info shown when available to reach sellers.

Search by image

Best when you have a photo of the product (e.g. from your phone or a screenshot).

  1. Open the By image tab on the home page.
  2. Click Choose file to choose an image, or drag and drop a file onto the drop zone.
  3. Optionally add more images (we use the first for search).
  4. Click Search to run the query.

Tip: Clear, well-lit product photos usually give better matches. Cropping to the product helps.

Search by URL

Use this when you have a direct link to a product image (e.g. from another site).

  1. Open the By URL tab.
  2. Paste the image URL into the field (e.g. https://example.com/image.jpg).
  3. Use Preview to preview the image if needed.
  4. Click Search to search.

The URL must be publicly accessible. CORS or hotlink protection on the target server can block loading.

Search by text

Search by product name or keywords when you don't have an image.

  1. Open the By name tab.
  2. Type the product name or descriptive keywords (e.g. "Xiaomi Mi Vacuum Mop 2").
  3. Click Search to see semantically similar products.

Understanding results

Results are grouped by store. Each card shows:

  • Product image and title (linked to the product page when available).
  • Brand and store name.
  • Price in TMT when available.

If no matches are found, try a clearer image, a different photo angle, or a text query with different keywords.

Tech & API

The platform uses:

  • Visual search — Image and URL queries are embedded with a vision model; we search a vector index of product images.
  • Text search — Name queries are embedded and matched against text/metadata in the same index.
  • Stores — Data is indexed from configured store scrapers; results are grouped by source.

For API access, bulk export, or custom integrations, contact the platform maintainers.