Player not loading? Watch on YouTube
Open Notebook is presented as an open source, MIT-licensed NotebookLM alternative for research on your own server or hardware. The video covers version 1.9.0, released June 2, 2026, and compares its focused notebook workflow with broader AI workspaces.
The walkthrough describes importing PDFs, web pages, YouTube transcripts, PowerPoint files, audio, markdown and plain text. Full-text and vector search support conversations about those sources, with clickable inline citations. Generated notes can include summaries, flashcards and outlines. Podcast generation adds multiple voices, configurable speaker personalities and adjustable length.
For self-hosted deployment, the speaker describes downloading a Compose file with curl and starting Docker Compose. The stack includes a frontend, backend, vector store and optional audio synthesis services. Initial startup takes a few minutes to download images and initialize the database, according to the narration.
Users can configure provider API keys or connect Ollama. The speaker says an Ollama configuration can provide an offline workspace with zero external API calls; that claim applies to the local configuration, while external providers receive data for their tasks. Different models can handle indexing, chat and podcast scripts. The speaker recommends stronger models for script quality and describes Open Notebook as more narrowly focused than Odysseus.
