kindle-gpt
FreeSearch and chat on your Kindle highlights.
FreeFree tier
About kindle-gpt
Kindle GPT is an open-source AI tool that enables you to search and chat with your Kindle highlights. It works by importing the .html notebook exported from the Kindle app, parsing the highlights, generating embeddings using OpenAI's text-embedding-ada-002 model, and then enabling semantic search via cosine similarity. The most relevant results are fed into GPT-3.5-turbo to produce natural language answers. All data is stored locally, and you can export your embedded data as a .csv file for reuse. The project is 100% open source and recommended for desktop use.
Key Features
Import Kindle notebook (.html) and parse highlights
Generate embeddings using OpenAI text-embedding-ada-002
Semantic search via cosine similarity with up to 10 most relevant results (~2k tokens)
Answers generated by GPT-3.5-turbo based on query and matched highlights
Export embedded data as .csv file for later reuse
All data stored locally, no database required
100% open source code
Pros & Cons
Pros
- Open source and free to use (only requires your own OpenAI API key)
- Data is stored locally, ensuring privacy
- Supports .csv export for portability of embedded data
- Leverages powerful OpenAI models for accurate search and answers
- Easy to run locally with simple setup instructions
Cons
- Requires an OpenAI API key, which incurs usage costs
- Desktop-only usage recommended; not optimized for mobile
- Depends on manual export of Kindle notebook (.html) from the Kindle app
- Embedding process may be slow for large highlight collections
- No built-in database; re-importing .csv is needed for persistence
Best For
Searching through personal Kindle highlights by topic or keywordChatting with your reading notes to get summaries or insightsRevisiting and querying important passages from books you have read
FAQ
How does Kindle GPT work?
Export your Kindle highlights as a notebook (.html) from the Kindle app, import the file into the app, generate embeddings using OpenAI's text-embedding-ada-002, then search or chat with your highlights. The tool uses cosine similarity to find relevant passages and GPT-3.5-turbo to generate answers.
Is my data stored on a server?
No. All data is stored locally in your browser. Kindle GPT does not use a database. You can re-import any previously generated .csv files to avoid re-embedding.
Can I export my embedded highlights?
Yes. After generating embeddings, you will receive a downloaded .csv file of your embedded notebook, which you can use elsewhere or import back into Kindle GPT later.
What models does Kindle GPT use?
It uses OpenAI's text-embedding-ada-002 for embeddings and GPT-3.5-turbo for generating answers.
Is Kindle GPT free?
The code is 100% open source and free to use. However, you need your own OpenAI API key, which comes with usage costs.