What it does
A local vector database for retrieval-augmented generation, built on FAISS. It ingests PDF, TXT, Markdown and DOCX files, splits them into chunks, creates embeddings and stores the index on disk. An AI agent can then run semantic search over the documents, and a command line tool supports indexing and querying from the terminal.
Use cases
- 01Search policy documents with natural language
- 02Index a folder of PDFs for an internal assistant
- 03Query an index from the command line