Stackreader

Stackreader indexes a folder of PDFs, Word files and notes locally and lets you ask questions across all of them. Answers quote the source passage with a page number, and clicking a citation opens the document at that spot. It runs with a local model or any API key you choose.

Researchers with long reading lists, lawyers handling discovery documents and anyone working with confidential files use it when cloud upload is not an option. Indexing a thousand documents takes about an hour on a recent laptop. Scanned PDFs need OCR first, tables are only partly understood, and answers from a small local model are noticeably less precise than from a larger hosted one.