🟢 Beginner–Intermediate ⚙️ Type: AI Research Assistant / NotebookLM Alternative 💸 Free & Open Source (MIT) ⭐ Trending on GitHub
What is Open Notebook?
Open Notebook is a powerful, open-source, and privacy-focused alternative to Google’s wildly popular NotebookLM. It allows you to upload massive amounts of personal data—such as PDFs, text files, YouTube videos, PowerPoint presentations, and web links—and instantly chat with that data using Retrieval-Augmented Generation (RAG).
Google’s NotebookLM captured the world’s attention with its incredible “Audio Overviews” (AI podcasts). Open Notebook completely replicates this magic with its own highly customizable Podcast Generator. It can synthesize your uploaded documents into an engaging, multi-speaker audio discussion.
The crucial difference? Total privacy and model freedom. Instead of being locked into Google’s Gemini ecosystem and handing over your personal research, Open Notebook lets you use any AI model you want. You can plug in OpenAI, Anthropic, Groq, or even route everything entirely offline through your own local hardware using Ollama.
Who is it for?
- Students and Researchers who need to synthesize dozens of academic papers, lectures, and YouTube videos into easily digestible notes and podcasts.
- Privacy-Conscious Professionals dealing with strict NDAs, legal documents, or sensitive corporate data that cannot be uploaded to consumer cloud AI services.
- Homelabbers and Open-Source Advocates who want to run a massive RAG architecture and local Text-to-Speech (TTS) engines entirely on their own hardware.
- Writers and Content Creators seeking a personalized “thinking partner” to catalog ideas without training massive corporate AI models on their unreleased work.
What makes it special?
- Multi-Speaker Podcast Generation — Like NotebookLM, it generates deep-dive audio discussions. Unlike NotebookLM, you can customize the podcast by selecting up to four speakers, giving them distinct personalities, and choosing custom Text-to-Speech voices (via OpenAI, Kokoro, or local TTS engines).
- Bring Your Own Model (BYOM) — You are not trapped in a walled garden. You can set up OpenAI for text generation, Anthropic Claude for deep reasoning, and a local Ollama model for fast drafting—all at the same time.
- 100% Local Capable — If you have the hardware, you can disconnect your computer from the internet entirely. By integrating with `llama.cpp` or Ollama, every document read and podcast generated stays on your physical drive.
- Encrypted Database — It uses a local instance of SurrealDB to store your notes and embeddings, secured with a custom encryption key you control.
Requirements before you start
Open Notebook is designed to run in a containerized environment to keep its database and backend services clean:
- Docker & Docker Compose — Essential for running the application and the SurrealDB database seamlessly.
- An API Key OR a Local AI Engine — You will need active API keys (e.g., OpenAI, Anthropic, Groq) or a local installation of Ollama running on your machine.
- Basic Terminal Knowledge — Just enough to run a few commands to start the Docker containers.
Step-by-step installation
Step 1 — Download the Configuration File
Open your terminal (or Command Prompt) and download the official docker-compose.yml file into a new folder on your computer:
mkdir open-notebook
cd open-notebook
curl -o docker-compose.yml https://raw.githubusercontent.com/lfnovo/open-notebook/main/docker-compose.ymlStep 2 — Set Your Encryption Key
Open the docker-compose.yml file in any text editor. Look for the environment variables section under the open_notebook service. You must change the default encryption key to a secure password of your choosing:
environment:
- OPEN_NOTEBOOK_ENCRYPTION_KEY=my-super-secret-key-123Save and close the file.
Step 3 — Start the Services
In your terminal, tell Docker to pull the images and start the application in the background:
docker compose up -dWait roughly 15 to 20 seconds for the database to initialize and the application to fully boot.
Step 4 — Configure Your AI Models
- Open your web browser and navigate to
http://localhost:8502. - In the left-hand menu, navigate to the Models settings page.
- Click + Add Configuration, select your preferred provider (e.g., OpenAI, Anthropic, or Ollama), and paste your API key or local endpoint URL.
- Click Sync Models, then navigate to Default Model Assignments to tell the app which AI should handle chat, summarization, and audio generation. You are now ready to create your first notebook!
Common errors and fixes
| Error | What it means | How to fix it |
|---|---|---|
| Podcast Generation fails without an error message | The Podcast Wizard requires a designated Text-to-Speech (TTS) model and an Outline Generation LLM assigned in your default settings. | Go to your Model Settings and ensure you have properly configured and assigned a TTS provider (like OpenAI’s tts-1 or a local Speaches container) to handle the audio synthesis. |
Connection Refused when adding an Ollama model | The Open Notebook Docker container is trying to reach Ollama on localhost, but “localhost” inside the container just points back to itself, not your host PC. | When configuring your Ollama provider URL in the Open Notebook UI, use http://host.docker.internal:11434 instead of http://localhost:11434. |
Bind for 0.0.0.0:8000 failed: port is already allocated | Another application on your computer is already using port 8000 (which SurrealDB uses). | Open your docker-compose.yml file and change the SurrealDB port mapping from "8000:8000" to "8081:8000", then restart Docker. |
Free vs Paid comparison
| Feature | Open Notebook (Local Open Source) | Google NotebookLM |
|---|---|---|
| Software Cost | $0 (Bring your own API keys) | Free (Currently) |
| Data Privacy | ✅ Absolute control (Data stays on your PC or chosen API) | ❌ Uploaded to Google servers |
| Podcast Customization | ✅ Choose voices, personalities, and up to 4 speakers | ❌ Locked to Google’s default male/female hosts |
| Setup Effort | ⚠️ Requires Docker and API configuration | 🟢 Instant web browser access |
Bottom line: Google NotebookLM is an incredibly polished, zero-friction tool. However, if you are working with confidential business documents, sensitive codebases, or deeply personal journals, handing that data over to Google is a non-starter. Open Notebook takes 10 minutes to set up via Docker and gives you the exact same RAG and podcasting superpowers with absolute data sovereignty.
Alternatives — 3 similar tools
1. Google NotebookLM
The commercial giant that inspired this project. It is currently free to use, highly reliable, and features incredibly smooth Audio Overviews powered by Gemini. It is the best choice if you don’t care about data privacy and just want instant results without setting up software.
2. AnythingLLM
The most robust, all-in-one local desktop application for RAG (Retrieval-Augmented Generation). You can drop thousands of PDFs into it, connect local or cloud models, and chat with your documents instantly. It is easier to install than Open Notebook but does not have a native multi-speaker Podcast Generation feature.
3. Quivr
A highly popular open-source “second brain” in the cloud. Like Open Notebook, it allows you to ingest files, links, and code to create a specialized AI assistant. It offers a very polished enterprise-grade interface but is heavier to deploy locally than Open Notebook.
🚀 Want more free AI tools like this?
We find, test, and write setup guides for the best free and open-source AI tools — so you don’t have to dig through GitHub yourself.Browse Free AI Tools at globalaiforce.com/shop →
📸 Follow us for daily AI tool tips and tutorials: instagram.com/globalaiforce