added Agentic RAG notebook

[taradepan]

Agentic RAG

Normally, when we use RAG (Retrieval-Augmented Generation), we provide it with a single collection containing embeddings, and we perform a vector search on that collection. However, what if we have multiple collections covering different domains and need to perform vector searches across them based on the user's query?

In such scenarios, Agentic RAG becomes invaluable. It can intelligently search across multiple collections, perform internet searches when required, or even generate answers from its own knowledge base. This makes it an integral solution for handling diverse queries in a seamless and efficient manner.

I have added a notebook that uses MongoDB as the vector database with multiple collection.