This document intelligence implementation uses LangChain and Pinecone to take in and analyze single documents, using Streamlit to take in questions and output answers.
The keys are put in a file entitled "NO_COPY_keys_file", with the OpenAI key being represented by opn and the Pinecone key being represented by pnc. A sample document also needs to be used (though one is not provided in this case).
Running the project turns the document into vector embeddings, puts the embeddings into the vector database Pinecone, and allows the resulting vector embeddings to be queried by OpenAI. Then, it creates a Streamlit frontend allowing the user to repeatedly input questions and get back answers without having to repeatedly run the code.