rustic-ai/uni-db

Uni-db attempts to unify three distinct data paradigms (Graph, Vector, Columnar) into a single embedded engine. While the value proposition is strong for complex RAG and local-first AI applications, the project currently lacks the momentum (29 stars) and technical insulation to be considered highly defensible. It is largely a sophisticated 'glue' layer sitting on top of Lance (a high-performance columnar/vector format). The primary competitive threat comes from the upstream project, LanceDB, which could easily add a graph query layer (OpenCypher) and effectively subsume Uni-db's entire value proposition. Additionally, established embedded databases like Kùzu (graph) and DuckDB (columnar) are aggressively expanding their feature sets into adjacent modalities. The project's defensibility is limited by its reliance on external query standards and storage formats without yet having a proprietary 'secret sauce' or massive community adoption that would create switching costs.