how it works: step 1: vector search gets 10 similar chunks step 2: cohere rerank scores by relevance step 3: llm uses top 5 best chunks result: better answers, less noise before: ~85% vector…

tech: python, fastapi, openai, chromadb production-ready with async, error handling, docs. what it does: ✅ crawls entire websites ✅ cleans & chunks content ✅ semantic search ✅ gpt-4 powered q&a ✅…