This has been an interesting project for which Streamlit was a huge help in developing. The chatbot uses a RAG model that retrieves and augments user messages with content from Exceljet.
I’d love to hear your thoughts—do you think it performs better than the standard LLM response for Excel-related questions?
Here are some resources I found useful while building the bot:
I stumbled across this yesterday and it was wonderful. The advice was laser focussed on my actual problem and gave me an exact solution. Unfortunately I tried it again today and it’s gone. Hopefully this sort of thing will become common in future.