A solo founder built an enterprise knowledge graph without a graph database
TheRedfather · reddit · 2026-07-21
- The founder of QX Labs describes how they built an enterprise knowledge graph without using a graph database.
- Their argument is that GraphRAG and Neo4j-style systems are often overkill for most company-brain use cases, especially when a regular database plus search index can do the job.
- The post explains why plain top-k RAG is not enough for questions like “tell me everything about X” or corpus-wide counts, and why graph-heavy indexing can be expensive and operationally awkward.
- Their alternative uses a small fixed ontology, waterfall entity resolution, alias tables, embedding-based matching, and a cheap LLM only for ambiguous merges.
- Instead of typed graph edges, the system stores co-occurrence counts as top-N lists on entity records, and entity summaries are generated lazily on first request.
Related event: Low-Cost Enterprise Knowledge Graph Ditches Graph Databases(2 posts)→
More from Infra
- Graph workload 854.graph500 enters SPEC CPU 2026 as a new CPU benchmark — Prof_DavidBader · 2026-07-22
- HilbertRaum open-sources a fully local AI chat and document analysis app for private use — Vladowski · 2026-07-22
- Hybrid and local inference are emerging as a response to AI energy and token costs — dmitry140 · 2026-07-22
- NVIDIA details Vera CPU with 2x performance claims and a 22,000-core rack — ryanshrout · 2026-07-22
- NVIDIA says Vera Rubin NVL72 delivers 10x more tokens per megawatt than Blackwell — nvidia · 2026-07-22
- AI economy is running out of cheap compute as data-center and power costs surge — Scobleizer · 2026-07-22