SurfSense
An open source, privacy focused alternative to NotebookLM for teams with no data limits. Join our Discord: https://discord.gg/ejRNvftDp9
SurfSense is a self-hostable AI knowledge base where you connect documents, cloud drives, and local folders, then ask questions and get cited answers, written reports, and audio summaries.
SurfSense is an open source tool for building a personal or team knowledge base that you can then question with an AI. The README positions it as an alternative to Google NotebookLM, a tool that lets you add documents and ask questions about them. The main differences the project claims are: no limits on how many documents or notebooks you can have, no lock-in to one AI provider, and the option to run the whole thing on your own server so your data never leaves your control.
You add sources by connecting cloud storage services like Google Drive, OneDrive, Dropbox, or Notion, or by uploading files directly. The system indexes everything and lets you ask questions that come back with citations pointing to the original source material. Beyond question-and-answer, it can generate written reports in formats like PDF or Word, produce podcast-style audio summaries, and create images. A real-time collaboration feature lets multiple team members share a notebook and chat together.
A desktop app is available alongside the web interface. It adds a global keyboard shortcut to bring up an AI assistant from any application, a mode where you select text on screen and ask the AI to explain or rewrite it, a screenshot mode where you can capture part of your screen and ask questions about it, and a folder-watching feature that automatically syncs a local folder to your knowledge base as files change. The README suggests pointing the folder watch at an Obsidian notes vault as one use case.
Self-hosting uses Docker and takes a single install command on Linux, macOS, or Windows. The cloud version is available at surfsense.com. The app supports multiple languages including English, Spanish, Portuguese, Hindi, and Chinese. You can configure which AI models handle text, images, speech-to-text, and text-to-speech, including models you run locally.
Where it fits
- Connect Google Drive or Notion and ask an AI questions about your documents with source citations
- Point a folder-watch at your Obsidian vault to auto-sync notes and chat with them via AI
- Generate a podcast-style audio summary or a PDF report from a collection of documents
- Share a notebook with teammates to collaboratively question a shared knowledge base in real time