RAG · Python
Ongoing
Personal book RAG
Ask my favourite health and business books a question, and get an answer drawn straight from the text.
❡
A screenshot of the question-and-answer interface will live here.
Grounded in the actual pages, not the model's vague memory.
What it does
I keep going back to the same handful of books. This lets me ask them directly — “what does this one actually say about X?” — and get an answer anchored to the real pages rather than a model's half-remembered gist.
How it works
The books are split into chunks and embedded into a vector store. A question retrieves the most relevant passages, and a language model answers using only those, so the response stays tied to the source.
It's genuinely useful day to day, and the kind of tool I find myself reaching for often.
← All projects
Next project
Parking watcher →