Cortex RAG
A local retrieval-augmented generation app that answers questions about your own PDFs, Word files and text using Ollama, citing the file and page each answer came from.
Cortex RAG is a local retrieval-augmented generation application. Users upload PDFs, Word documents or text files and ask questions in plain language; each answer cites the source file and page it came from, and it runs entirely on the user's machine through Ollama with no API key or cloud upload.
Behind each answer is a hybrid search pipeline that combines keyword and semantic search, an entity graph, a neural reranker and an LLM relevance check, and each technique can be switched on or off from the sidebar. An evaluation harness measures which techniques actually help on the user's own documents, and follow-up questions are rewritten into complete questions before searching. Indexes are saved on disk under a chosen name, reopen automatically and can be extended later.
It requires Ollama and Python 3.10 or newer, and runs through a Streamlit web interface.