The Highlight Extraction Experiment
LinkedIn Post

How I rescued 400 book highlights without endless typing or scanning.
That moment when you realized youβve highlighted 400 passages in a paper bookβ¦ a challenge I wrote about yesterday:
270 pages with approximately 200-400 highlights. In a paper book!
And I want to transfer my highlights so that I can work with AI building on them (like I did here: https://lnkd.in/gNKFWWRg).
βββ
This is what I came up with:
βββ
β Took a photo for each page with a highlight (12β)
β All automatically saved to my Google Photos
β Created an Album (2β)
In batches of 10 images:
β Inserted in a Google Document (2β)
β Added to NotebookLM as source (1β)
β Ask for extracting the highlights (2β)
β Saved as new note in NotebookLM (some seconds)
βββ
Learnings so far
βββ
β NotebookLM can read Google Documents containing images, analyze them and extract my highlighted passages
β Sometimes it extracts a little more. Iβm fine with that
β Rarely it skips a highlight. I can live with that too
β It canβt yet process all 140 images at once, but works well with 10-image chunks
βββ
What you gain
βββ
β Speed
β Highlights that you can now use with AI
β A Google Doc with all images as reference
Thoughts?
Content preparation
Posted on LINKEDIN on 2025-03-25_Tue