Chat bot

Since you have an existing pinecone index, you could start with a pinecone vector index.

There's an example notebook here:

Just pass in an empty array [] instead of any actual documents

https://github.com/jerryjliu/llama_index/blob/main/examples/vector_indices/PineconeIndexDemo.ipynb

Then, you could use that index as a tool in langchain. A few ways to do that
https://gpt-index.readthedocs.io/en/latest/guides/tutorials/building_a_chatbot.html

That guide uses multiple indexes in a graph, but you could skip that part and use a single pinecone index.

https://github.com/jerryjliu/llama_index/blob/main/examples/langchain_demo/LangchainDemo.ipynb

This one uses the index as a tool in langchain more directly