GenRec: Towards LLM-Native Recommendation at Netflix

GenRec is an innovative, LLM-backed recommendation ranker developed by Netflix that optimizes recommendation systems by verbalizing user histories and contextual information as natural language. It leverages pre-trained language models, reducing reliance on handcrafted features while achieving significant improvements in performance metrics over traditional recommendation models. This approach indicates a shift toward LLM-native recommendation systems, allowing for efficient handling of diverse content types and optimizing for long-term member satisfaction and business goals.