ArthurYangX / nano-NotebookLM

What is nano-NotebookLM?

nano-NotebookLM is a self-hosted AI study assistant built in Python that transforms course materials (PDFs, PPTX, DOCX, Markdown) into an interactive learning environment. Upload your lecture notes and it automatically builds a knowledge graph, indexes everything for semantic search, and lets you chat with citations that point back to specific pages. Beyond Q&A, it generates structured LaTeX notes with math support, creates practice quizzes, and includes an exam prep mode that grows a question bank based on your mistakes.

Why is it gaining traction?

The killer feature is the self-evolving exam prep. When you get a question wrong, the system automatically generates new variants targeting that weak spot. Combined with page-accurate citations and an editable knowledge graph, it offers a deep reading loop that NotebookLM itself does not provide. The flexibility to run on any OpenAI-compatible endpoint (including local models via Ollama or vLLM) means privacy-conscious users never have to send their study materials to the cloud.

Who should use this?

Students and researchers working with technical material (math, CS, engineering) who want to extract structured notes from dense PDFs. Self-hosters who need an open-source alternative to Google NotebookLM. Educators building study tools around their own course materials. Anyone who wants full control over their data while having AI assistance for understanding documents.

Verdict

This is a promising but early-stage project with only 42 stars and a 1.0% credibility score. The feature set is impressive and the code structure suggests thoughtful engineering, but it is still in beta with limited community backing. Worth trying if you need the specific capabilities it offers, but do not expect polished UX or production-ready stability yet.