You’re reading Excellent AI Prompts.
AI prompt chains, agent skills, and agentic workflows for professionals learning to use AI to earn more, think sharper, and live better.
Copy them. Use them. Build with them.
Where your attention goes, your energy flows.
As I sit here reading a book on sound, I’m unusually attuned to every sound in the room. The dog’s water fountain gurgling. Birds chirping. The white noise of the cooling system humming. My husband picking up his coffee mug, and swallowing a sip.
That's what people loosely call the RAS (reticular activating system). Neuroscientists will tell you the real story is more complicated, but the phenomenon is genuine: once something enters your awareness, your brain starts flagging it in places it was always present. Which is why when I sat down to read a book on sound, my whole body became an ear.
“I think I will do nothing now but listen, To accrue what I hear into myself — to let sound contribute toward me.”
Walt Whitman, “Song of Myself”
Paper before the machine
We’re going to read. And the first part of this, we’re doing on paper.
I use a Stalogy notebook as my compendium. It’s where I take notes on things I want to think about slowly. No particular system, no pressure to make it legible to anyone else, including future me. It’s a thinking surface.
You can use anything. A legal pad, a composition notebook, a single piece of printer paper folded in half. The medium matters less than the act of slowing down enough to think and write.
Handwritten notes, for me, will never be replaced. When your hand can’t keep pace with the material you’re consuming, you’re forced to choose, compress, and reframe instead of simply transcribing. Mueller and Oppenheimer documented this, finding that laptop note-takers were more likely to capture material verbatim, while longhand note-takers did better on conceptual questions. The moment you decide what is worth writing down, you’re already making your first editorial judgment.
So before we touch NotebookLM, give yourself 15 to 20 minutes with something you’ve been reading. It could be a chapter, an article, or a handful of pages. Take notes the way you feel like taking notes, not the way you think you should. Underline the line that perplexes you. Write the question that keeps coming up. Draw the arrow between two things that rhyme.
Note: Write legibly enough for a camera to read. In a moment, you’re going to photograph these pages and turn them sources.
Related Posts
Enter NotebookLM
NotebookLM’s core mechanic is this: you give it sources, and it only reasons from those sources, RAG. It rarely drifts into general training data and starts free-associating. It pulls only from the sources you give it.
Interesting article I found during by “RAG, RAS similarities” exploration: Why More Context Makes Your Agents Stupid
When you’re ready, pull out your phone. The NotebookLM app has a camera button built directly into the Sources tab on mobile. Tap it, photograph your notebook pages, and they become a source. Add the book alongside them, PDF, a web page, a YouTube link, or the passages you marked pasted into a text note. You don’t need the whole book. The sections you actually sat with are enough.
Try this prompt:
I’m reading [Book Title] by [Author]. Find: the author’s essays, interviews, or recorded talks on this subject; writing that explores [the book’s central theme] from a different angle; and any critiques or responses to this author’s ideas.One caveat worth knowing: the text recognition (OCR - optical character recognition) reads clean handwriting well and struggles with dense scrawl.
Then open the Studio panel.
9 ways to make NotebookLM an unforgettable reading experience
1. Generate the mind map first, then just wander
Click Mind Map in the Studio. The generic version builds in seconds and gives you the book seen from above, every major idea as a node, branching outward.
Spend five minutes in it the way you’d explore a new city on foot. Tap a node you didn’t expect to see. Read the summary it generates. Tap another. Let the ideas settle before you decide what you think about them.
Then go further. Before generating a second map, type into the prompt field:
Build this map around [the idea from your paper notes that stopped you most]. Show how the author develops that thread and what connects to it.
2. Generate the Audio Overview and go for a walk
Two AI hosts will have a genuine conversation about your book for 10 to 20 minutes. They’ll disagree. They’ll find things you didn’t notice. You’ll have opinions about what they got right and what they missed. Sometimes the overviews are highly insightful. Sometimes they fall flat. It’s an experience nonetheless.
3. Direct the Audio Overview before you generate it
You can fully customize the audio overview, or podcast, topic.
The Audio Overview responds to instruction. Before you click Generate, type this in the customization field:
Focus on [the tension or question that interested you most]. Make the tone [curious / playful / direct — pick one]. Skip [what you already understood well]. I want to leave the conversation thinking about [what you most want to sit with].
Example audio overview using the book I’m reading now:
My minimalist prompt: “Explore how to redesign a bedroom with sound”
4. Watch it like a short documentary
The Video Overview converts everything in your notebook into a short narrated visual presentation. This includes whatever sources you’ve assembled: the articles, interview transcripts, author talks, and your photographed notes. It works from what you gave it, not from a full text it doesn’t have.
In Studio, click Video Overview. Choose a format (Brief runs a few minutes to a bit less than an hour; Explainer goes deeper) and one of six visual styles. Before generating, paste this into the customization field:
Focus on the tension between [what the author argues] and [the strongest counterargument you found in the sources]. Frame it as a documentary about that disagreement.
Pause it when something surprises you and find the source it came from in your notebook.
5. Create three ways to see this
The Infographic feature in Studio condenses your notebook into a single designed visual summary, organizing concepts, relationships, and key points into a downloadable image. As of early 2026, there are 10 visual styles to choose from: Sketch Note, Kawaii, Professional, Scientific, Anime, Clay, Editorial, Instructional, Bento Grid, and Bricks. You can also control orientation and level of detail before generating.
Because your notebook is built from interviews, articles, and transcripts rather than the full text, more than likely, what NotebookLM chooses to visualize reveals which ideas are actually coming through in your sources.
Generate three, each with a different prompt.
First: no instructions. Let the AI decide what matters.
Example:
Second:
Show how the ideas in these sources relate to each other. Prioritize connections over hierarchy. Use a Sketch Note style.
Third:
Show the contrast between the author’s main argument and the strongest counterargument found in these sources. Use an Editorial style.
Use the third one as an outline if you’re going to write anything from this session.
6. Find the connection you wouldn’t have looked for
This one uses the chat panel, not Studio.
The sources you’ve built through interviews, essays, and critical responses give you a dimensional picture of an author’s thinking, often richer than the book alone. Use it.
Paste this:
I work in [your profession or field]. Based on everything in this notebook, what is the most unexpected connection between these ideas and the way I do my work? Not the obvious one.
The more specific you are about your work, the sharper the answer.
Example:
I publish a daily newsletter about AI for working professionals. Based on everything in this notebook about sound and attention, what is the most unexpected connection to how people actually absorb information delivered in text?
If the first answer is the obvious one, ask it to go further.
7. Quiz yourself
The Quiz feature in Studio generates multiple choice questions pulled directly from your actual sources. Because your notebook is built from interviews, articles, and your own notes rather than the full text, the questions test what came through in those sources specifically.
Click Quiz in Studio. You can set the number of questions, difficulty level, and topic before generating. In the topic field, paste:
Focus on where the sources in this notebook agree and where they contradict each other. Test me on the tensions between them, not just the facts.
When you get a question wrong, click Explain. NotebookLM cites the specific source where the answer lives.
The Explain button is more useful than the quiz itself. Use it even on the questions you get right.
8. Ask for a report that surprises you
The Reports section in Studio generates structured documents from your sources. The options are Briefing Doc, Study Guide, Blog Post, and Create Your Own. Create Your Own takes a custom prompt. The suggest formats are sometimes useful as well.
In the Create Your Own field, paste:
Write a report on the most counterintuitive finding across all the sources in this notebook. Write it for someone who already understands the basics and wants the idea that shifts how they see the topic. Lead with the surprising thing.
A surprising report is only as surprising as your sources. A good prompt won’t fix shallow material.
9. Use Deep Research to go somewhere you haven’t been
Deep Research is NotebookLM’s agentic mode. Instead of reasoning from what’s already in your notebook, it creates a research plan, searches the web autonomously, and returns a citation-backed report. You can pull those findings and their sources into your notebook with one click.
In the chat panel, paste this first to get a question worth giving it:
Based on everything in this notebook, give me one deep research prompt that would take this inquiry somewhere I haven’t thought to go. Something adjacent. Something I wouldn’t have searched for on my own.
Then use that prompt to launch Deep Research. Here’s an example of what a strong question looks like for a session built around a book on sound:
What does research on auditory attention say about whether the sounds present during a reading session get encoded alongside the information itself, and what does that suggest about how we choose our reading environments?
Deep Research finds sources you didn’t know to look for. Whether they’re the right ones depends entirely on the quality of the question you gave it.
What are you reading?
The book on sound is still open on my desk. I have three pages of Stalogy notes I haven’t finished processing and a Deep Research session running on a question I didn’t know I had until an hour ago.
There’s probably something on your nightstand, or a tab you’ve kept open for weeks, or a chapter you set down intending to return to. Pick it up, give it twenty minutes and a pen, and let your hand be slower than your eyes. Then bring it to NotebookLM.
What are you reading right now? Let me know in the comments.
Until next time,
Lea