Notebooker.ai: A Self-Hosted NotebookLM Alternative

NotebookLM is great, but the lack of control over models and data storage is a dealbreaker for some. Notebooker.ai fixes this by letting you bring your own AI keys (OpenAI, Anthropic, or local endpoints) and your own S3-compatible storage like R2 or AWS. Basically, you own the stack.

It's built on top of the open-source Open Notebook platform and acts as a centralized hub for the PDFs, links, and audio files that usually get buried in a hundred open browser tabs. Instead of just hoarding links, you can chat with your notebook to get answers with exact citations or convert a reading list into a private RSS podcast feed.

The most interesting part is the plugin engine. It handles different "creation types" as modular plugins, meaning it can generate flashcards, mindmaps, textbooks, or timelines without bloating the core code.

If you're looking for a practical tutorial on how to set up a personalized knowledge base without giving your data to a black box, this is a solid route.

Technical Highlights:

• Model Flexibility: Supports OpenAI, Anthropic, and Cloudflare Worker AI.
• Storage: S3-compatible buckets (R2, Spaces, AWS) for full file control.
• Connectivity: Full webhook support for incoming/outgoing JSON, plus RSS processing.
• Privacy: No data used for training when using your own keys.

For those who want to see it in action before deploying, there's a view-only demo at https://app.notebooker.ai/demo.

Since it's a side project, expect a few bugs, but the fact that much of it is submitted as PRs back to the open-source core makes it feel reliable. Definitely worth a look if you need a more transparent AI workflow for research.