helix-db

Why we included this project

Most teams working on RAG pipelines, agent memory, or knowledge-graph applications end up wiring together a relational database, a vector index, and a graph store, then writing glue code to keep them in sync. HelixDB flips that setup: one OLTP engine written from scratch in Rust handles graph, vector, and full-text workloads directly on object storage, so you query a single place instead of coordinating several. That is useful when you want persistent, queryable memory for coding agents or conversational assistants without bolting a vector index onto an existing application database. The project offers typed query DSLs for Rust, TypeScript, Go, and Python, and its CLI can spin up a local instance or point at a hosted cloud service. A one-shot bootstrapper also scaffolds a working project quickly, which helps if you are trying HelixDB on a first AI application rather than migrating a mature stack.