“Holy &#*@&!!!” moments with Google's NotebookLM
Diving into some of NotebookLM’s coolest new features and use-cases
I rarely have a revelatory moment anymore with generative AI tools and features. I did today.
Have you used Google's NotebookLM? If so, have you used it recently? It's always been a cool application with really interesting use cases, but they've made some recent updates that left me thinking about what incredible things I could accomplish with it.
NotebookLM uses AI to help organize content and research materials, analyze information, and generate creative outputs. You can throw a bunch of content into it like PDFs and blog posts, or even transcripts, and it will summarize it, let you query it, and ask it questions about the data. That's been part of the app since it launched. The next part hasn't.
Audio Overviews: the holy &#*@&!!! moment
NotebookLM will now generate a "podcast episode" featuring two AI hosts that discuss your sources. Not only do they discuss it, but they discuss it in a very real and conversational manner. Except for a few mispronunciations, it's extremely listenable and compelling (perhaps a little cheesy but what AI-generated content isn't?).
For fun, I threw a link to a 2,000-word blog post I wrote a couple of years back about content project management (with comparisons to The Beatles).
Here's the result. It's pretty wild.
Next, I tried adding the text of a whitepaper that I was a ghostwriter on a while back about the Jobs to be Done framework.
Here's that result. This one's even more compelling, in my opinion.
I'm not really sure where that ketchup bottle example came from - I think the AI came up with it. Other examples in there came straight from the text, and it matched the source material.
Some NotebookLM audio overview use cases
Audio companion
If you wanted to share an audio companion to your written content, or to explain something complicated to your audience, this could be a simple and easy way to do it without taking the time to record a podcast or a video. You could even add some video to the audio track, as I did above using Descript.
Education/learning
If you're a student or you're simply trying to learn a lot of in-depth, complicated subjects, you could throw all of your PDFs or web page articles in there and your new digital friends could chat about it for you, summarizing and pulling together some good insights and helping you to learn it and grasp it - summarization and uncovering insights is something that Google's Gemini model is quite adept at (it's not as good at things like copywriting in my experience).
More features and use-cases beyond Audio Overviews
NotebookLM’s value extends beyond the Audio Overviews function. Here’s one example:
Conference note summariztion: using Google's NotebookLM to get more value out of conferences
I go to conferences, but then get busy and never go back to the notes I took. It's been an ongoing problem, and conferences are expensive. Case in point: I have a ton of my own notes as well as presentation slides and other summaries from the recent MAICON (Marketing AI Conference from Marketing AI Institute ). I'm thinking of throwing everything I collected into a notebook, adding some context, and seeing what comes out of it as far as my top takeaways from the event. Then I can continue to query the notebook through the next several months when I want to remember something I heard or wrote down, and I'll have an "AI summary" of the event as well that I could share with colleagues, which will help me pull even more value out of an already very valuable event experience.
The audio overview wasn't necessarily what I was after in this instance. I was mainly looking for a way to easily access and query the mountain of notes and slides (and a few audio clips) that I had brought back from MAICON 2024 (Marketing AI Institute's annual conference). I had 50 files, some from sessions I didn't even attend but had the slides. Coincidentally, NotebookLM lets you load precisely 50 files. I loaded everything and let the system read and analyze. Then I started to query it to see what it could uncover.
For instance, I remembered Robert Rose talking about storytelling in his presentation "AI is not a Strategy". As you'll see in the graphic, I asked NotebookLM a question about that and it quickly uncovered his insights in a nicely formatted response.
I then asked it to uncover some of the key takeaways about copyright and generative AI that were discussed at the conference and got a similar result.
You can ask specific and detailed questions, or you can ask for general high-level info. Notebook LM will give you the response as well as linked citations to where it found the data in the files.
Other options
Can you do this with other AI tools? Sure. You could make a GPT in ChatGPT or a Project in Claude. There are likely other options too. NotebookLM just makes it really easy. Could it give you incorrect information or make faulty summaries? Of course. Google even tells you as much at the bottom of the screen. Just check those citations to verify. So far, I've found its responses to be correct.
Show the value of attending
NotebookLM used in this way will make it easy for you to take action on new insights and ideas generated from attending a conference. If you're attending a conference on your employer's dime (or even on your own dime), this is a great way to make sure that you can demonstrate the ROI of attending.
Note: be careful what you upload to NotebookLM or any generative AI tool. Refrain from uploading sensitive, personal, or proprietary information into this or any cloud-based AI platform.