LatticeDB: A SQLite-like embedded graph database with vector and full-text search
LatticeDB – Like SQLite but for graph databases
LatticeDB is an embedded, single-file property-graph database that combines graph traversal, HNSW vector similarity, and BM25 full-text search in one query layer. It targets local-first applications like Graph RAG, agent memory, and knowledge tools, offering zero-config operation and durable event streams. Benchmarks show sub-microsecond node lookups, 0.83 ms vector search at 1M vectors with 100% recall, and up to 2,800x faster graph traversals than SQLite's recursive CTEs.
LatticeDB at 1M achieves 0.83 ms mean with 100% recall@10 — faster than FAISS single-threaded HNSW and competitive with Weaviate and Qdrant server-based systems (which add network overhead in practice).