I Built an AI Pipeline for Kindle Highlights

A personal project is detailed to automate the summarization of Kindle highlights through a custom-built, local AI pipeline. The process begins with data retrieval and processing, where highlights are extracted from the Kindle's "My Clippings.txt" file, parsed, deduplicated, and sorted. Heuristics are then applied to differentiate between highlighted text and section titles, structuring the raw data for better context. The core of the project uses the Ollama framework to run an open-source large language model locally, feeding it the processed highlights with a specific prompt to generate a structured summary. The final output is a Markdown file that organizes the original highlights by section and appends the AI-generated summary.