Case Study: TheFork achieves better LLM observability and faster iteration with Arize AX

TheFork cuts p95 latency with Arize AI after eliminating duplicate embeddings

TheFork, one of Europe’s leading restaurant discovery and booking platforms, faced the challenge of shipping LLM-powered features to production. They required tight feedback loops, production-grade observability, and flexible evaluations to monitor their AI for quality, latency, and cost to boost conversions. To meet these needs, they implemented Arize AI's observability platform, Arize AX, on AWS.

Arize AI provided a solution with prompt-level tracing, automated online evaluations, and drift alerts. This implementation delivered significant results, including a material reduction in latency by eliminating duplicate embeddings and sharper visibility into costs and performance. TheFork gained faster iteration cycles and a scalable framework for cost governance, allowing them to improve AI quality without unexpected spend.

TheFork