// the find
redis-developer/ArXivChatGuru
Use ArXiv ChatGuru to talk to research papers. This app uses LangChain, OpenAI, Streamlit, and Redis as a vector database/semantic cache.
A Streamlit demo that lets you load arXiv papers on a topic into a Redis vector index and ask questions against them using LangChain and OpenAI. It's a learning project — the README says so explicitly — meant to illustrate how Redis fits into a RAG pipeline, not to be a research tool you'd actually rely on.
Docker Compose setup is the primary path and works out of the box with a single make command. Topic-scoped Redis indexes mean you can load multiple paper sets without them polluting each other. The stats page exposing index metadata and query engine internals is a useful addition for anyone trying to understand what's happening under the hood. Test coverage exists for the core modules (constants, db, prompt, stats) which is more than most demo repos bother with.
This is hardwired to OpenAI — no model swapping without code changes, and the chunking strategy is described in the README as something that needs improvement. No chat history or multi-turn memory, so each question is context-free; you're querying the papers, not actually having a conversation about them. The project hasn't been touched since early 2026 and the LangChain dependency means you're one minor version bump away from a compatibility break. No metadata filtering by year or author yet, so if you load 50 papers and only care about 2024 results, you get everything.