Kind: captions
Language: en
This table was generated automatically
from these 12 messy documents. But that
is just the tip of the iceberg. What if
I told you that most people are only
using about 10% of what Notebook LM can
actually do? So, I'm about to show you
seven use cases that will make you
realize you've been thinking way too
small about Notebook LM's true
potential. Starting with use case one,
which is turning messy research into
structured data. Let's break down
exactly how I generated that table you
saw in the beginning. First, go to your
notebook LM dashboard and open your
project. I'm using my AI automation
notebook where I've already uploaded my
sources. Look at the top right of the
screen under the studio panel. You will
see an option labeled data table. Click
that and notebook LM will start scanning
every single source, pulling out the
relevant information and building you a
structured comparison spreadsheet
automatically. And look at this result.
It instantly created a clean table with
columns for tool name, key features, and
pricing plans. You can scroll through
right here and see that it accurately
grabbed the specific details from the
PDFs I uploaded. But what if you need
specific data points that aren't there?
For that, we're going to go back and
click the edit button right here on the
data table tab. This allows you to
define exactly what you need. In the
prompt box, describe the columns you
want. I'll write I need columns for
monthly cost, difficulty level, and
customer support quality along with the
tool name, key features, and pricing
plans. And I'll hit generate. And
instantly, it completely restructures
the entire table to match that exact
query. But to actually work with this
data, you need to get it out of notebook
LM. So for that, click the button on the
top right to export the spreadsheet to
Google Sheets and the entire table
transfers instantly. Now you have a live
spreadsheet you can share with your
team, update as pricing models change,
and actually use for decision-m. The key
is being specific about what you want.
Don't ask for a summary, ask for exact
columns, company name, pricing model,
setup time, pros and cons, whatever
matters for your analysis. So that was
use case one, turning raw sources into
structured data. But sometimes you don't
need a spreadsheet. You need to tell a
story, which brings us to use case two,
drafting publication ready content. With
our sources still uploaded, let's turn
them into an article you can post
anywhere. Go back to the studio panel
and click on the reports tab. We want to
control the output right from the start,
so click the edit icon on the blog post
immediately. This opens the description
box. All right, write a thought
leadership article focusing on the
security vulnerabilities discussed in
these reports. Use a professional,
authoritative tone and write for a
technical audience. Hit generate. And
look at this result. It didn't just
summarize the text. It actually
synthesized a narrative. Look at that
headline. It's punchy and relevant. The
structure flows logically from the
problem to the solution. And the tone
has that specific security terminology
from our papers without sounding
robotic. You just went from reading PDF
files to having a complete publication
ready draft in about 10 seconds. So now
you have the structured data and you
have the article. But sometimes before
you can explain a topic, you need to
actually understand how the concepts
connect. Text is great for explanation,
but it's terrible for visualization.
That brings us to use case three.
Generating interactive mind maps.
Imagine you are learning a new complex
topic and you have a stack of dense
papers. Reading them linearly is
impossible. You need to see how the
concepts actually connect. To do that,
all you need to do is upload all your
papers to Notebook LM. I've got a folder
here full of heavy academic PDFs. Go to
the studio panel right where we found
the data table, but this time click mind
map. Notebook LM analyzes every document
and creates an interactive visualization
showing exactly how these concepts
connect. And look at this result. The
center node here is supply chain
resilience. Connected to that, you see
branches for supplier diversification,
inventory buffers, and risk assessment.
Now click on supplier diversification.
It expands deeper into strategies,
multi-ter risks, and benefits. Select
any of those subconcepts, and Notebook
LM immediately pulls up the specific
source papers and direct quotes to back
it up. This is how you learn complex
topics fast. Instead of reading through
40page documents, you explore
connections. When you find a concept you
don't understand, click it. And Notebook
LM gives you a plain language
explanation grounded in your actual
sources. This works for academic
research, technical documentation, legal
case analysis, or strategic planning.
The more sources you upload, the richer
the map gets. Now, we've visualized how
the concepts connect, but we need to
control exactly how the AI analyzes
those connections. That brings us to use
case 4, building ultradetailed AI expert
personas. Until early December, you
could customize how Notebook LM
responded, but you were capped at 500
characters. That is about three
sentences, which is not nearly enough to
build actual expertise into the system.
On December 4th, Google quietly
increased that limit to 10,000
characters, which is 20 times more
space. And that single change allows you
to generate results like this. Look at
the structure here. It generates a full
product manager style decision memo with
executive summary, user evidence,
feasibility assessment, blind spots, and
recommendation. It follows a strict
professional format because I told it
exactly how to think. To get that same
level of depth, start by uploading your
research. Then go to the top right,
click the settings button, and select
custom. This is where you define the
persona. I'll write you are a senior
product manager with 10 years of
experience. When analyzing documents,
provide decision memos structured as
executive summary, user evidence,
feasibility, blind spots, and
recommendation. Always site specific
data. Flag assumptions, and use clear,
direct language. But realize this, that
instruction I just wrote is only about
300 characters. You have 9,700 left, so
use that space to be specific. Define
the exact format you want. Specify how
it should handle uncertainty. Tell it
what to prioritize. With 10,000
characters, you can build personas that
actually think like specialists. The
more detailed your persona, the better
your output gets. So now you have an AI
persona that thinks like an expert. But
that analysis is useless if you can't
communicate it to a client. That brings
us to use case 5, generating client
ready presentations in minutes. Suppose
you just finished a research project and
your client needs a presentation.
Normally, you're looking at hours in
Google Slides, manually formatting
bullet points, and finding images.
Instead, start by uploading your
research documents. Go to the studio
panel and click on slide deck. Then,
click the edit button to give Notebook
LM specific instructions. I'll write a
10 slide presentation for a marketing
executive explaining our competitor
analysis findings. Focus on pricing
strategy differences and market
positioning gaps. Use a clean,
professional design. Notebook LM
generates a complete presentation,
professional design, proper hierarchy,
and visual elements, all pulled from
your sources and properly structured.
Export it, make a few edits if you need
to, and you are done. But don't stop
there. Go back to the studio panel and
click the edit button on infographic. In
the prompt box, I'll write create an
infographic comparing the three
competitors across pricing, features,
and target audience using a side-by-side
layout. Now you've got a slide deck and
a supporting visual, both created from
the same source documents. And if your
client prefers video, click video
overview in the studio panel. Notebook
LM creates a narrated video with AI
hosts discussing your findings while
showing relevant visuals on screen.
Okay, let's dive right in. I want to
talk about two things that on the
surface have absolutely nothing in
common. So, what's the big secret? Well,
it's not about tech and it's definitely
not about staying hydrated.
>> This workflow turns raw research into
three polished deliverables. Slide deck
for the meeting, infographic for the
report, and video for internal sharing.
all from one notebook. You have the
documentation and you have the
processes. But sending a PDF to your
team doesn't mean they actually read it
and it definitely doesn't mean they
understand it. That brings us to use
case six, building custom training
simulators with your training manuals,
compliance docs, or process guides
uploaded. Open the studio panel. Here
you will see specific options for
flashcards and quizzes. But don't just
click the button yet. If you leave it on
default, you get basic definition
checks. But we actually want to test
applications. So click the edit button
on flashcards. Set the difficulty to
hard and use the custom prompt box to
change the logic. All right. Create
scenario-based flashcards. Present a
real situation where an employee must
choose between process A or B. Do not
test definitions. Test decision-m. And
look at the difference. Instead of
simply checking if you memorize the
rule, it presents a complex scenario. It
forces you to apply the documentation to
a real problem and determine the correct
decision. It forces the user to think.
Every card includes an explain button.
Click it and Notebook LM breaks down the
correct decision, citing the exact page
in your training manual where that rule
exists. You can do the same for quizzes.
Ask it to generate questions that
require combining multiple concepts to
solve a single problem. Then simply
share the notebook with your team.
Everyone accesses the same interactive
training tools automatically generated
from your docs and grounded in your
actual rules. We've talked about
analyzing documents you already have.
But what if you don't have the documents
yet? What if you are starting from zero?
That brings us to use case seven, the
autonomous deep research agent. Notebook
LM added a feature called deep research
back in November. However, do not
confuse this with a Google search as it
isn't a search engine. It is an
autonomous agent. When you search
Google, you get a list of links you have
to read. With deep research, the AI
creates a plan and actually does the
reading for you. Let's look at an
example. I'll open a blank notebook with
zero sources. Select deep research from
the drop down right here and type a
complex goal. I'll write research the
pros and cons of implementing a 4-day
work week, specifically looking for
long-term productivity studies and
financial impact. Now, watch what
happens. It immediately generates a
multi-step research plan. It goes out to
the live web, scans hundreds of
articles, filters out the clickbait,
reads the technical reports, and
synthesizes everything. And this
actually changes everything. You used to
have to bring the data to the AI. Now,
the AI goes out, finds the data,
validates it, and brings it to you. It
turns Notebook LM from a library into a
librarian. So, you now understand the
true potential of what Notebook LM can
actually do. But knowing these advanced
features won't help if you are using the
tool wrong. I've seen people try
everything I just showed you and fail
because they skipped the basics. The
video on your screen covers the
foundation everyone misses. In just a
few minutes, I'll show you how to use
Notebook LM better than 99% of people
out there. It is on the screen right
now. Click it and I'll see you there.