kindle-gpt

Kindle GPT is an open-source application that enables users to interact with their Kindle highlights using artificial intelligence. It transforms static reading notes into a searchable, conversational knowledge base by leveraging embeddings and large language models.

Key Features

• Highlight Import & Parsing — Imports Kindle highlights from exported .html notebook files and parses them for processing.
• Embedding Generation — Uses OpenAI's text-embedding-ada-002 to create vector embeddings for each highlight, combining chapter/section names with highlighted text for context.
• Semantic Search — Queries highlights using cosine similarity to retrieve the most relevant passages (up to ~2k tokens across 10 results).
• AI-Powered Q&A — Generates answers by feeding search results into GPT-3.5-turbo, allowing conversational interaction with your notes.
• Local Data Storage — All data is stored locally without a database; users can export/import embedded highlights as .csv files for portability.

Philosophy

Kindle GPT prioritizes privacy and user control by keeping all data local and providing open-source transparency, enabling users to own and reuse their embedded highlights.