Adding a document retriever

[Juliano Alves de Oliveira]

I'm new to the langchain and langgraph I need a little help.

I'm trying to add some document retriever to enhanced the model with more context, but I'm having some troubles. Anyone can help with a snippet or a how to guide to make this modification?

Without the document retriever the code works very fine. Thank you

[Martin Mashalov]

What worked for me is to make a vectorDB in Chroma for example that contains file names mapped to descriptions of those files (either generated with your own algorithm or by GPT). Then organize this code into Langchain where you have a retriever that fetched and loads your CSV files and then puts them into the list that chain of tables uses.