If you want something simple and quick, the VAPI default knowledge base works fine and will give you basic retrieval out-of-the-box. But for better accuracy, flexibility, and scalability, I’d go for a custom RAG setup with a vector store and embeddings this lets you dynamically pull in relevant info while keeping your system prompt lean, which reduces hallucinations and makes updates easier. I can help design a setup that’s efficient without being overcomplicated if you like. @ManjarulCh_