I was going through the Agent tutorial ,

React uses prompting, function calling uses the function/tools api provided by the LLM provided

Usually function calling is more reliable

Thank you

"Should I create tools for every file and save that in database?"

I am building a RAG SaaS project where users will upload hundreds or thousands of PDFs, docs, and other file types. I need the agent to answer questions based on these documents. What is the most efficient way to create query engine tools for the agent?

Should I provide the agent with a single query engine tool and indicate that it can handle any type of question?
Alternatively, should I create and save separate query engine tools for each document upon upload and provide the agent with a list of these tools?

Additionally, does saving individual query engine tools for each document even work effectively?

Thank you for your help!

@Logan M it would be great help if you kindly answer this question. thank you again

Also this one "Could please share idea or resource regarding Incorporating Chat History in ReActAgent or FunctioncallingAgent, What i want is, after finishing a chat , the ReActAgent will get the previous 2 or 3 chats in its context .

@Logan M Sorry for asking too many question