>>1833the problem is that most of these LLM-based agents rely on
RAG pipelines that are still heavily dependent on indexed web crawls. if you aren't optimizing for the underlying
vector database structure, you're basically invisible regardless of how good your schema looks.