From Dev.to
Should our LLM wiki use vector search? We benchmarked it before deciding
Every "add vector search" proposal comes with the same promise: users phrase questions their own way, keyword search misses, and embeddings fix it. A feature request on Synthadoc made that case for us...