How to Turn NotebookLM Into a Personalized Book Analyst
Most people read books the inefficient way.
They read the whole thing and hope the useful parts reveal themselves.
Sometimes that works.
A lot of the time, especially with business or technical books, you spend hours reading material that is interesting but not that relevant to what you're actually trying to do.
I've been testing a better way with NotebookLM.
The trick is simple:
Don't upload just the book.
Upload the book plus a profile of yourself.
That second document changes everything.
If NotebookLM only has the book, it mostly gives you summary behavior. Fine, but generic.
Once you add a profile that explains who you are, what you're building, what decisions you're trying to make, and where your gaps are, the tool starts doing something more useful.
It starts filtering.
That is the real value.
NotebookLM is grounded in the sources you give it. That's the feature. It is not pulling random advice from all over the internet. It is cross-referencing the book against your own context and showing you where the overlap is.
So instead of asking, "What is this book about?"
You're asking, "What in this book matters for me right now?"
That is a much better question.
The profile does not need to be long.
It just needs to be useful.
I would include:
- who you are
- what you're working on right now
- the decisions you're actively making
- the things you know you don't understand well yet
- how you prefer to learn
- what you want books to help you do
For me, the highest-leverage part is the decisions section.
That is where the tool gets sharp.
If your profile says you're trying to decide whether to build a custom system or use existing APIs, NotebookLM can surface the parts of a book that actually help with that decision.
If your profile says you're building AI workflows on a budget and keep running into reliability problems, it can prioritize the sections that speak to architecture, evaluation, or cost control.
That is what makes this feel less like summarization and more like analysis.
You are basically turning a general book into a personalized lesson plan.
The setup is easy.
Create a new notebook.
Upload the book.
Upload your profile.
Then start with a few direct prompts:
- What are the most actionable insights from this book for my situation?
- Which chapters should I read first, and what can I skip?
- I'm trying to decide X. What does this book say that helps?
I tested this with Chip Huyen's AI Engineering.
Without the profile, the output was broad.
With the profile, NotebookLM started pointing me toward the sections on agents, system design, cost control, and evaluation because those were the things that matched what I was actually working on.
It also told me what not to spend much time on yet.
That part matters more than people think.
A lot of books are not evenly useful to you at a given moment. Maybe 20 percent is highly relevant and the other 80 percent is good background for later. If a tool helps you find that 20 percent faster, the book becomes more valuable.
Not less.
This also works because most of us are not reading just to become generally smarter.
We're reading because we're trying to make better decisions.
We're trying to solve a problem.
We're trying to understand something well enough to use it.
That is why I like this setup.
It makes reading more intentional.
You still need judgment. You still need to read the actual chapters. This is not a substitute for thinking.
It is a better way to aim your attention.
Build the profile once.
Reuse it across books, research papers, reports, and long technical documents.
Update it every few months as your projects and blind spots change.
Most people use AI to summarize information.
I think the better use is to make it sort information against your actual reality.
That is when a book stops being general knowledge and starts becoming something closer to a playbook.
You stop reading with hope.
You start reading with intent.
TL;DR
1. Create a short profile document about yourself. 2. Add six sections: who you are, what you're building, active decisions, knowledge gaps, how you learn, and what you want from books. 3. Open NotebookLM and create a new notebook for the book you want to analyze. 4. Upload the book as your first source. 5. Upload your profile as the second source. 6. Ask: "What are the most actionable insights from this book for my situation?" 7. Ask: "Which chapters should I read first, and what can I skip?" 8. Ask: "I'm deciding [insert decision]. What does this book say that helps?" 9. Read the recommended chapters first instead of going through the book blindly. 10. Reuse the same profile for other books, then update it every few months as your work changes.
Note: I use AI as a writing and thinking tool. The ideas, examples, and judgment in this post are mine.