Trieve

Trieve offers an AI search infrastructure for applications that combines search language models with tools for fine-tuning ranking and relevance. The service integrates various features such as full-text search via a retrieval model called SPLADE, semantic vector search, hybrid search which combines full-text search and semantic vector search with re-ranker models, and merchandising relevance tuning which boosts search results based on attributes like sales or popularity. Other capabilities include date recency biasing, sub-sentence highlighting and duplicate content handling for more effective knowledge management. Users can host sensitive data themselves with Trieve's platform ensuring data protection. Trieve also supports building search, recommendation and RAG features allowing users to focus on creating unique features for their applications. Further, it provides private managed embedding models and allows hosting of open-source models on their servers. The platform enables users to set up an industry-leading search UX quickly by adding existing data, integrating their API, and initiating search, recommendation or generation.