Case Study: FARFETCH boosts conversational AI performance with Milvus from Zilliz

Optimizing Conversational AI at FARFETCH

Farfetch, a leading online fashion retailer, aimed to elevate its digital shopping experience with its iFetch conversational AI system. To do this, they needed to move beyond traditional product catalogs and instead utilize high-dimensional product embeddings. Their challenge was finding a specialized vector database capable of efficiently storing and retrieving these embeddings in real-time to power instant, accurate product recommendations for users.

To address this, Farfetch conducted a comprehensive benchmark of vector databases, ultimately selecting Zilliz's Milvus. The solution implemented with Milvus provided significantly faster performance, resulting in 15x faster indexing time and 5x faster query time. This optimized their AI's ability to deliver highly relevant product recommendations, which in turn helped boost customer conversion rates for Farfetch.

Farfetch

Pedro Moreira Costa

Applied Scientist