Trieve Search Platform

Example Use Cases

• 1

  Perform semantic dense-vector and hybrid dense+sparse retrieval over document collections for high-quality search results.

• 2

  Run RAG pipelines with selectable LLMs via OpenRouter and manage topic-based memory for conversational context.

• 3

  A self-hostable, enterprise-grade search & recommendation backend (Qdrant integration, custom rerankers, filtering, and grouping) deployable in a VPC or on-prem.

Description

Trieve provides semantic dense-vector and neural sparse (SPLADE) search, hybrid search with cross-encoder reranking, recommendations, and convenient RAG routes that integrate with OpenRouter/LLMs. It integrates with Qdrant and supports bring-your-own models (embeddings, rerankers, LLMs), plus features like sub-sentence highlighting, recency biasing, filtering, and grouping. The project includes frontends, local development tooling, and guides for self-hosting in VPC/on-prem or using hosted services.