Q: Can I be a BETA tester? This looks really cool!

Maybe! Send me a contact note with your computer details (PC/MAC, GPU, etc,), ereader version, your LLM of choice, and why you want to help test.

​

Q: Is the AI Digital Bookshelf free?

A: The core functionality of being able to see your library, along with your highlights and notes is free. There is a one time cost to enable the AI functionality (generate a single book or cross-book analysis), plus any costs you incur depending on the AI model you run when you bring your own key (BYOK). Your AI consumption is charged through your AI model provider (i.e. Anthropic or OpenAI). For some examples of the AI output, click here.

​​

Q: What devices are compatible with the AI Digital Bookshelf?

A: All Kindles (excluding handwritten notes) are compatible. Kobo functionality has been built, but needs more testing (contact us if you want to help). Audible files that have been downloaded to your Kindle will also appear on the bookshelf, and you can easily download and import your full Audible library.

​​

Q: What languages does the AI Digital Bookshelf support?

A: Currently only English. We are looking into Spanish and other language compatibility and localization.

​

Q: I think a few books are missing from my library.

A: Some books are encrypted and can't be read directly from the Kindle. However, if you make a single highlight or note on the book in your Kindle, it will appear in your AI Digital Bookshelf. This works because it reads both from the Kindle library structure, as well as the My Clippings file.

​

Q: How much does it cost to run the AI?

A: It depends... it could be a lot, it could be nothing. For example if you run Anthropic's frontier model against multiple full text books, it could be several 100,000 tokens and relatively expensive. But if you install a local LLM, it's free.

​

Q: My AI results are slow or poor.

A: The quality of the AI results are based on two factors: the power of your chosen AI model and the quality of the input; your highlights, notes, summaries, and whether or not you use full text. Some models have very high token limits and are better suited for full text analysis. Others are very strong at reasoning even with shorter inputs (20 highlights/notes) like Anthropic's Opus 5.

​​

Q: Some expected book covers are missing or wrong. How can I fix this?

A: The metadata fetch works based on book title and author. Make sure those fields are correct. Editing and saving the book metadata, or clicking the "Refetch" button will kick off a request to one of three online metadata sources (Goodreads, Apple Books, Google Books). If it still does not return what you're looking for, right click on the book cover in the "Book Details" view and select "Remove Cover".​

​

Q: How was the AI Digital Bookshelf built?

A: Written primarily in python, it uses common open source tools including Flask, SQLite, and HTML5/CSS/JavaScript.

​

​

About the developer

Contact with bugs, issues, questions, or other feedback

Back to main