Book ideas from meetings: the source material is already being generated

If you've ever thought about writing a book, this workflow makes it tractable. Claude organizes meeting moments against your book outline. Every week.

What you'll have when you're done

A file, book-ideas.md, organized by the chapters of your book outline. Every week a script reads your meetings, finds the moments that belong to a chapter, and files each one under the right heading with the verbatim quote, the source meeting, and a note on what kind of material it is: an anecdote, a principle, a counter-example, a data point, a framework. The outline stops being empty headings and starts filling with evidence.

The book you keep meaning to write starts writing its own first draft, one meeting at a time. Not a finished manuscript. The thing that actually stops most operator books: the source material, gathered and sorted against your structure, so that when you sit down to draft a chapter the raw material is already there, in your own words, waiting. The blank chapter becomes a chapter with twenty things in it you'd forgotten you knew.

The book isn't hard to write. It's hard to start.

I'd thought about writing a book for years and never got past the same wall. It wasn't the writing, exactly. It was that every time I imagined starting, I pictured a blank document and a blank year, and the gap between "I have things to say" and "here is a chapter" felt uncrossable. So I did what everyone does. I told myself I'd do it when things calmed down, and things never calmed down, and the book stayed a someday.

The wall is real but I'd misdiagnosed it. The hard part of an operator book was never having ideas. The hard part is that the source material, the specific stories and the exact way you explained something and the customer line that proves your point, is scattered across hundreds of conversations you can't recall in enough detail to use. You remember you have a great chapter on hiring. You do not remember the four specific stories that would make it land, because each one happened in a different meeting months apart, and the detail that made them good evaporated when the meeting ended.

I tried starting the way books are supposed to be started. A document titled "Book." An outline. A Saturday blocked to "just write the first chapter." All of it failed the same way: I'd open the outline, stare at a chapter heading, and have nothing concrete to put under it, because the concrete things were locked inside meetings I'd half-forgotten. The outline made the emptiness official without filling it.

The material existed the whole time. Every story, every principle, every customer truth that belongs in the book had already been said out loud and caught by Granola. This workflow reads your meetings against your outline and files each moment under the chapter it serves. The source material stops being scattered and starts being sorted, and a sorted pile of your own best material is most of the way to a draft.

What you need first

• The [Granola][1] → markdown pipeline already running. This workflow runs on the meeting library the pipeline produces. If you haven't set it up yet, do [Granola → markdown][2] first. Thirty minutes. Comes back when done.
• Claude Code installed and pointed at the folder your Granola transcripts land in. Free tier handles the extraction.
• A book outline, even a rough one. This is the one real prerequisite. It can be five chapter titles in a text file with a sentence under each. The outline is what gives Claude the buckets to sort into. If you don't have one yet, the "When it breaks" section shows how this workflow helps you find it.
• A few weeks of meetings in the folder. A book accumulates slowly. The more conversation there is to mine, the faster the chapters fill, but this is a long game by design.

Step-by-step

Step 1Write a rough outline (or borrow one you already have)

The outline is the skeleton everything hangs on, so spend twenty minutes on it before anything else. It doesn't need to be good. It needs to exist.

mkdir -p ~/notes
touch ~/notes/book-outline.md

Put five to ten chapter titles in that file, each with a one-line description of the argument it makes. If you already articulated your principles using the [leadership doctrine workflow][5], that file is a head start: the principles that repeat are usually your chapters. A rough outline you'll revise beats a perfect one you wait to write.

Step 2Save the extraction prompt

The prompt reads your outline first, then sorts each meeting moment into the chapter it serves. Save it once.

mkdir -p ~/.claude/prompts
touch ~/.claude/prompts/book-ideas.md

Paste the prompt below into that file.

Read ~/notes/book-outline.md first to learn the chapters and the argument
each one makes.

Then read every meeting transcript in ~/notes/granola/ from the last 7 days.
For each moment that could serve as material for a chapter, append it to
~/notes/book-ideas.md under the right chapter:

## Chapter: {chapter name}
### Theme: {the sub-point from the outline it supports}
- **Source:** {filename} · {date}
- **Verbatim:** "{quote}"
- **Why it fits:** one sentence
- **Type:** anecdote / principle / counter-example / data point / framework

Rules:
- Verbatim quotes only. The exact words are the material.
- Only include moments that materially advance a chapter's argument. A vaguely
  related thought is noise. A specific story or a sharp line is material.
- If a moment fits a theme the outline doesn't have yet, file it under a new
  "## Unsorted / possible new chapters" section and flag it. That's how the
  outline grows.

Step 3Run it across everything you have

Open Claude Code in your notes folder:

cd ~/notes
claude

For the first run, mine your whole history, not just the last week. Paste the prompt but change "last 7 days" to "everything in the folder." Claude reads it all, sorts the moments into your chapters, and builds the first version of book-ideas.md.

Open it and read one chapter that came back full. That's the moment the book stops feeling impossible: a heading you'd been staring at empty, now sitting under a dozen specific things you actually said, with the source and the exact words. Check that the verbatim quotes really fit the chapter they're filed under, and skim the "Unsorted" section, because the moments that didn't fit your outline are often telling you what the outline is missing.

Step 4Let the outline and the file grow together

This isn't a one-time extraction. It's a long accumulation that quietly updates your thinking about the book.

Two things happen as it runs. The chapters fill, week by week, until the thin ones have enough to draft from. And the "Unsorted" section grows, and when a theme shows up there three or four times, that's the book telling you it has a chapter you didn't plan. Move it into the outline and the next run starts filing material under it. The outline you wrote in Step 1 was a guess; the file turns it into something shaped by what you actually keep talking about.

Step 5Make it a standing weekly run

The whole point is slow accumulation, which only works if it keeps running. Two ways:

Option A · One saved command (lowest effort). Each week, open Claude Code and say: "Run ~/.claude/prompts/book-ideas.md against this week's transcripts in ~/notes/granola/ and append to ~/notes/book-ideas.md." Ten seconds, and the chapters get a little fuller.

Option B · A Claude Code [Routine][3] (fully automated). Get Claude to set up a Routine that runs the extraction every Sunday, appends the week's material, and sends you a note saying which chapters grew and what landed in "Unsorted." Set it once and the book's source material gathers itself in the background, month after month, while you run the company.

How you'll know it's working

The signal is a chapter crossing a threshold.

The first time a section of your outline has five or more solid entries under it, you can draft that chapter, and the workflow has done the thing that makes a book tractable. You sit down to write, and instead of a blank page you have five specific stories in your own words, sorted, sourced, ready to shape into prose. The chapter that felt impossible in the abstract is suddenly a half-day of editing what you already said. That's the unlock: the book was never one giant act of creation, it was a hundred small captures you were already making and throwing away.

The second signal is that the outline starts changing on its own. The "Unsorted" section surfaces a theme you keep returning to, you realize it's a real chapter, and the book gets truer to what you actually have to say than the outline you guessed at in Step 1. The book finds its own shape from the evidence.

When it breaks

• You don't have an outline yet. The most common blocker, and the workflow has an answer. Run it with a near-empty outline (even just "## Unsorted") and let it file everything into the unsorted section for a few weeks. The themes that show up over and over are your chapters. The workflow surfaces the book's structure from what you keep talking about, which is a better outline than one you'd invent at a desk.
• Too much volume. Mine your whole history on the first run and a popular chapter can come back with fifty entries, most of them mediocre. The fix is a monthly prune: keep the five to ten strongest entries per chapter, cut the rest. A chapter needs its best material surfaced, not all its material hoarded.
• Everything lands in one chapter. If your meetings skew heavily toward one topic (say, fundraising for a quarter), that chapter floods and the others stay empty. That's accurate, not broken: it reflects what you actually spent the season on. The thin chapters fill when your attention moves, and the imbalance itself tells you which parts of the book your daily work is and isn't feeding.
• The quotes don't quite fit the chapter. Early runs file moments under a chapter on a loose association. Tighten the "materially advance the argument" rule and add an example of a strong fit versus a weak one. The bar should be high: a book chapter is built from material that proves a point, not material that's vaguely on topic.

Where this fits in your harness

This is part of the content layer of your [harness][4]. The [Granola → markdown pipeline][2] made every conversation searchable. This workflow points that searchability at the longest-horizon project most operators carry and never start, and gathers its source material against a structure, week after week, until the book is mostly assembled before you've formally written a word.

The siblings most relevant to this workflow:

• [Content ideas from meetings][6] · the short-form sibling. Both mine your meetings for what's worth saying; content ideas feeds the post you ship this week, book ideas files the same kind of material against a chapter outline for the long project. The posts you write from the content file are often early drafts of the book's sections.
• [Your leadership doctrine][5] · the upstream feeder. The doctrine workflow captures the principles you lead by; those principles are frequently the chapters of the book. Build the doctrine file first and your outline half-writes itself.

See the [Granola pillar][7] for the full pipeline and the other workflows that compound on top.

[1]: /workflows/what-is-granola [2]: /workflows/granola-to-markdown [3]: /workflows/claude-routines [4]: /workflows/what-is-a-harness [5]: /workflows/leadership-doctrine-from-meetings [6]: /workflows/content-ideas-from-meetings [7]: /blog/granola-for-ceos-highest-roi-ai-install