Ecommerce Search & Discovery, Powered by Catalog AI
Catalog enrichment meets Large Language Model AI to deliver ecommerce search that is relevant, conversational & intelligent, unlocking better conversions & higher revenue per visitor.
Ecommerce-specific embeddings & models, that beat GPT-4 in benchmarks
Out-of-the-box attribute enrichment & classification for your catalog
Trivially build retrieval & ranker models tuned to your catalog & user activity
AI generated taglines for your listing results, adapted to users' query & intent
/enrich: Structured attributes & categories, from unstructured fields, reviews & the web.
/embed: Get embeddings for products, tuned to your use-case (e.g. query match or similar SKUs).
/collections: Create descriptive collections for categories, personas & trends [COMING SOON]
/attributes: Map queries to standard attribute selectors (useful for facet search).
/embed: Get embeddings for your queries to enable vector based search
/followups: Get suggestions for follow-up queries to show to users [COMING SOON]
/retrieve: Get high recall SKU candidates for an input query using embeddings + vector DBs.
/rank: Get high precision results, factoring in query relevance, past click-through data, merchandiser preferences & user personalization.
/generate: AI generated taglines for each result [COMING SOON]
What makes us different
All in on Embedding Search: Our search is built ground up for the AI era. We obviate the need for keyword search, reducing your infra costs, and eliminating the need for synonym lists & manual query tuning.
Built for Ecommerce: General purpose embeddings (e.g. OpenAI Ada) don't work out-of-the-box for Ecommerce. We've trained custom embeddings & models on catalog data from across the web, tuned for tasks like search & recommendations.
Catalog Aware: Search has traditionally only been as good as the documents you feed it, necessitating operational work to normalize & enrich catalogs. We've automated this with custom models for categorization & attribute enrichment.
Dynamic Content: Search AI has traditionally only had control over the ranking of products in response to a query. We go one step further to generate dynamic taglines for search results, enabling you to metaphorically speak to your users.
including 15+ years of experience with Ecommerce Catalog AI