Photo based matching

Upload a photo of a sneaker or fashion item and the service returns the products it believes match, drawing on a catalog of around a million items. A Beam agent sends an incoming image, reads the returned matches, and applies your rule to attach the top result to the record that triggered the lookup, such as a resale listing or a support ticket. When no match clears your threshold, or several products come back equally likely, the agent routes the image to a person to pick the right item rather than guessing.

Confidence scored results

Each match arrives with a confidence signal, and the service reports high accuracy on the items it recognizes. A Beam agent reads that confidence, applies the cutoff you approved, and writes only matches above it into your system while tagging borderline ones for review. If a result sits just under your cutoff, or the confidence pattern looks off for a clear image, the agent holds the item and notifies a person, so a weak match never lands on a customer facing record without a human agreeing to it first.

Fashion catalog coverage

Coverage spans sneakers plus broader fashion products, so a single lookup can recognize footwear and apparel alike. A Beam agent uses this to enrich listings or tickets: it reads the image, applies your rule to fill in brand, model, and style fields from the match, and updates the record. When an item appears to fall outside the catalog, or the returned category disagrees with what the record already states, the agent flags the mismatch for a person to confirm before the enriched details are saved anywhere.