LatticeDB tags two releases in 11 hours

LatticeDB reached 163 points on Hacker News on August 25. Its maintainer answered with two releases by the next morning, v0.11.1 at 13:09 UTC and v0.12.0 at 00:05 UTC. It packs graph queries, vector search, and full-text search into a single file, with no server for RAG builders to stand up.

LatticeDB is written in Zig and pitched as SQLite for graph databases, a single-file embedded knowledge-graph store instead of a full Neo4j deployment.

At 403 stars, the project is still small. But the combination is uncommon: embedded, single-file, graph-native, with vector search baked in rather than added on. I've wanted exactly that shape more than once, usually while wiring separate stores together for a project that didn't need separate stores.

The same-day turnaround says something too. Two tagged releases inside eleven hours means someone read the incoming traffic and responded with real fixes. It doesn't prove the vector index scales past a demo, and the GitHub page doesn't say anything about performance at larger graphs yet.

The point count won't decide whether LatticeDB survives its first spike of attention. Whether the maintainer keeps releasing fixes after the traffic fades will.