LangChain research assistant

what is this ?

This repo holds the code for an autonomous research agent, building up on this excellent LangChain web search assistant YouTube tutorial. It can search:

• a PostgreSQL databases
• Arxiv papers summaries
• PDF documents
• the web

how to use

• cp .env.example .env (and fill in the blanks)
• pip install -r requirements.txt
• python3 main.py
• go to localhost:8000 and search for something:
  • a pdf by going to /generate-pdf-report/playground/ route
    • enter the absolute file path for the document you want to summarize4
  • a PostgreSQL database by going to /sql-qa/playground after having plugged in your database credentials in the .env file
  • on Arvix (papers summaries) by going to /generate-arxiv-search-report/playground/ route
  • on the web by going to /generate-web-search-report/playground/ route

how it works

The search feature is heavily inspired by https://github.com/assafelovic/gpt-researcher, which basically splits up one search into multiple sub searches before generating a final report =>