AI Book Notes Prompt Pack

The biggest pain of reading isn't finding time-it's forgetting everything. Ask AI to "summarize this book" and you get generic fluff no better than a back-cover blurb, and two weeks later you can't even recall the title. The problem is the instruction: no reading goal, no output format, no link to your existing knowledge. This pack goes from chapter summaries to atomic Zettelkasten cards to full book-breakdown outputs, in three levels. The key is giving the model "reading goal + output format + your paraphrase." Beginner clears the structure, intermediate turns the book into reusable knowledge assets, expert turns it into publishable content.

Beginner: Chapter Summaries

Just starting out-get the structure clear and coverage solid; don't rush to extract opinions yet.

You are a reading assistant. Produce chapter summaries for {{book_title}} by {{author}}.
Requirements:
1. Chapter by chapter, 3-5 sentences each, covering the core argument
2. Keep original wording for key concepts/terms; gloss in parentheses if needed
3. No personal opinions, no extending beyond the text
4. Output as a Markdown list: ## Chapter N - Title - Summary
Reading goal: {{e.g. grasp the book's framework / prep a review / locate a concept}}
Source: {{text or chapter content}}

Intermediate: Atomic Knowledge Cards (Zettelkasten Style)

Break the book into reusable units-one idea per card. This is the essence of German sociologist Luhmann's slip-box, which let him produce 70+ books in his lifetime.

You are a Zettelkasten expert. Break the following reading into atomic knowledge cards.
Each card carries one idea only, following the "permanent note" principle:
1. Title: a declarative sentence summarizing the idea, ≤20 words, not "Notes on X"
2. Body: the idea + supporting quote from the source + your own paraphrase, ≤150 words
3. Tags: #topic #method #{{custom}}, 3-5 tags
4. Links: note linkable themes, e.g. [[cognitive bias]], [[decision models]]
5. Each card stands alone for future retrieval and reuse
Reading goal: {{e.g. build a decision-model library / collect writing material}}
Source: {{text}}
Output: 5-8 cards in Markdown

Expert: Turn the Book Into Output Content

Finishing the book isn't the end; turning it into something publishable is where compounding starts: an article, a mind map, an action checklist-three formats covering the full read-think-do loop.

You are a book-breakdown expert. Based on {{book_title}}, produce three deliverables:
1. Article (1500-2000 words): hooky but not clickbait title, pain-point opener, 3 core points with original quotes + cases, end with a CTA. Tone: {{analytical / friendly storytelling}}
2. Mind map (Markdown outline): root node = the book's thesis, expand to 3 levels, one sentence per node
3. Action checklist: convert the book's methods into 5-7 to-dos, each "action + trigger scenario + expected outcome"
Audience: {{e.g. new professionals / managers / creators}}
Source / your notes: {{text}}
Output: three sections, ready to copy-paste

Pitfalls

1. Dropping the whole book at once-context overflow, the model gives you generic fluff. Feed it chapter by chapter with a clear reading goal each time.
2. Letting AI output without your own paraphrase-"read ≠ learned." The essence of Zettelkasten is that you paraphrase; AI is just the structuring assistant.
3. No reading goal stated-the same book summarized for academic research vs. content topics is wildly different. Always state the goal in the prompt.
4. Treating quotes as conclusions-the model tends to elevate the author's rhetoric into core arguments. Ask it to separate "argument" from "rhetoric," or verify yourself.

References

• Zettelkasten.de-the canonical site for the slip-box method, with hands-on posts
• Andy Matuschak's Notes-ongoing research on "what makes a note useful to your future self"
• Niklas Luhmann (Wikipedia)-inventor of the Zettelkasten, who produced 70+ books from 90,000 cards