arthuryangx/nano-notebooklm

nano-NotebookLM is a self-hosted, open-source study assistant you run on your own computer or server. You point it at course materials (PDFs, PowerPoint files, Word documents, or Markdown files) and it builds a searchable knowledge base from them. You can then chat with the material and get answers that link back to the exact page where the information came from, so you can verify what the AI says. The tool handles several study-related tasks from one interface. It can produce structured notes in LaTeX format, which is a document format common in academic and technical fields. It generates practice quizzes and an exam-preparation mode that tracks which questions you answered incorrectly and automatically creates follow-up variants to help you practice your weak spots. There is also an editable knowledge graph, a visual map of concepts extracted from your documents, which you can rearrange and annotate by hand. A key design choice is flexibility in which AI model does the work. You can connect it to OpenAI, Anthropic Claude, Google Gemini, DeepSeek, or more than a dozen other cloud providers, or run a local model through tools like Ollama or LM Studio. Swapping providers happens through a settings panel in the browser without restarting the server. This means your study materials stay on your own machine rather than being sent to a fixed third-party service. Setup is a standard Python install. After cloning the repository, you install dependencies, copy an environment file, fill in at least one API key, and start the server. Optional extras include a heavier PDF extractor for scanned documents and a LaTeX compiler if you want to export notes as PDFs. The project is Apache 2.0 licensed and accepts community contributions. It is aimed at students and researchers who want the citation-linked chat and quiz features of Google NotebookLM but prefer to keep their data local and choose their own AI provider.