This turned out nice. but question about expanding.

[zooteld]

I really like the way you laid this out. I am tinkering with a pdf upload function and a mysql database for a vector store. do you have any future plans or pointers for combining these with your project?

[3x3cut0r]

this project is born, because it was a requirement of my work to use streamlit for the frontend. dont ask why. otherwise i would never use streamlit for building a frontend.

so my future plans depend heavily on what my work demands of the project. and at the moment it is just in proof of concept status. so my work on the project can end at any time. so much said.

next on the list of features would probably be some kind of history for the chats. for this, sql is probably the only option for me.

a pdf upload feature would also be in my mind but could be more complex to handle ocr and stuff ...

I'll have to do some research on both features and see if I can find the time to implement them.

[zooteld]

thats really cool. simmular to my project. For my use all the data has to be private. as a pdf upload feature... I have 2 samples I have been looking at. the "streamlit docs" sample that reads the pdf's from a dir, and another that just loads the doc into memory using FAISS. I took a little different approach and built in a upload file to dir function. then probably trigger a dir reader from llama_index to read the files. as for a vectore store... sql is probably best because then you can just have it load that on startup. ... just some thoughts but llama_index has a ton of options