Projects
E-commerce · Hamburg
AI search for an e-commerce shop
Semantic product search that understands what customers actually mean.
Challenge
The existing search missed 30% of relevant products. Customers bounced.
Solution
Vector search with embeddings, LLM-powered query rewriting, A/B tested.
Results
+22% conversion, −18% bounce on the search page.
Tech
Python, LangChain, Weaviate, Next.js