Add agentic applications

[tituslhy]

Is your feature request related to a problem? Please describe. The Vanna AI library is great. I just can't help noticing that the tool is best suited for easier questions (what are my top 10 customers, etc.) instead of harder questions. This can be:

1. Nested questions: Who are my top 10 customers, is there anything common between them and what products do they buy?
2. Broad questions: Describe my sales trend from 2020-2023. The SQL statement here is very simple but the challenge is in having the LLM analyze the data describe the analysis qualitatively.
3. Challenging questions that require other tools: Compute the CAGR of my sales from 2020-2024.

These question types are best addressed by LLM agents because agents can write and execute multiple SQL queries upon reflection. Examples of agentic libraries include llamaindex, langchain, crewAI and autogen, but in my opinion VannaAI's fundamental SQL analysis abilities is ahead of these libraries.

Describe the solution you'd like I'm hoping for VannaAI to implement:

1. The use of LLM agents within the library (for example ReAct), so that the agents can reflect to generate better answers, and generate multiple plots and sql tables (if needed) as well as a qualitative answer to address the user's question.
2. An integration with the agentic libraries if possible (especially langchain and llamaindex) so that users can also exploit the agentic capabilities in these libraries to extend chat assistant functionalities even further. There is currently one integration written by llamaindex but this integration can definitely be improved.

Describe alternatives you've considered Using the llamaindex integration (https://llamahub.ai/l/llama-packs/llama-index-packs-vanna?from=all), I wrapped it a LlamaIndex ReAct agent around it and attempted to use it. Worked in a notebook but there are still a few kinks to fix though.

Thank you!