⚡ New NotebookLM 2.0 Updates are IMPRESSIVE! (Full Tutorial)

Introduction

This is the new NotebookLM 2.0. I know, it looks the same as the old NotebookLM, but under the hood, Google has added something insane.

Now, you don’t just get answers in a block of text. You can create editable PowerPoints, Excel sheets, PDFs, and more, all directly in the chat.

I’ve been using this to analyze data, spot missed revenue opportunities, and turn insights into action faster than ever.

Google tested it against the old NotebookLM… and the new version won 78% of the time in real research tasks.

I'll show you how to use NotebookLM 2.0 step by step, including the parts most people miss but that are actually the real power behind this upgrade.

I. What's New in NotebookLM 2.0: 3 Major Shifts

NotebookLM 2.0 brings 3 changes at once: real file creation in chat, a stronger engine running the show, and a chat that researches on its own.

Key points

• NotebookLM now creates real files (PowerPoint, Excel, PDF, charts) right in chat, built from your sources and editable after generation.

• Runs on Gemini 3.5 with a secure cloud computer per notebook, reporting a 65%+ win rate over the prior version, 78.2% specifically in web research.

• Chat flags gaps in your current sources and suggests new ones, but nothing gets added without your approval.

Who Actually Has Access Right Now

This update doesn't reach every account yet. Availability sits behind a paywall, rolling out to:

• Users with Google AI Ultra

• Workspace business customers with AI Ultra Access or AI Expanded Access

If you open NotebookLM and don't see the Slide Deck button or the Data Table option from earlier sections, that's the reason. Check your subscription tier first before assuming something went wrong on your end.

Shift 1: Real Files, Built Right in Chat

NotebookLM now creates real files inside the chat window, all ready to download from the studio panel:

• PowerPoint decks

• Excel spreadsheets

• PDFs

• Charts

You give it instructions, and it pulls context from your sources to build the file. You can edit the output afterward too.

Shift 2: Upgraded, More Thoughtful Chat Experience

This output upgrade rides on a new engine. NotebookLM moved to Gemini 3.5, and Google reports a win rate against the prior version across its core evaluation categories:

• Average win rate of over 65%

• 69.9% win rate when analyzing large documents

• 78.2% win rate in web research and source discovery

Every notebook also got its own secure cloud computer, letting NotebookLM write and run code for deeper research and more complex analysis, backed by over 100 curated software skills.

You don't pick which skills activate. NotebookLM figures out which ones fit the task and runs with them.

Practically speaking, this turns NotebookLM into something closer to a data analyst than a document reader.

Upload 3 years of invoices, and instead of summarizing what's in them, NotebookLM writes a script, runs it in its own sandbox, and hands back a finished chart or report.

Shift 3. An Agentic Research Companion

The chat behaves differently too. Faced with a complex request, NotebookLM decides:

• Which steps to take

• Which tools to call

• Which format fits the output

Ask it what your notebook is missing, and it points to gaps in your current sources, then suggests new ones to fill them.

You can start a notebook by just entering your loose ideas and questions into the chat. NotebookLM will then guide you through building out your repository, even finding and adding (with your permission!) relevant sources from the web.

Bonus: Flashcards Now Fully Customizable

Flashcards are now fully customizable. You can edit questions, tweak answers, and add brand-new cards to create the ultimate set of study tools.

Share them with your friends, classmates, academic rivals, etc, just be prepared for the endless shower of praise and accolades that follows.

II. Getting Started: Build Your Notebook's Foundation

Everything NotebookLM 2.0 builds for you starts with the same first step: a notebook full of solid sources.

Key points

• Every output starts the same way: build a notebook with solid sources first, either added manually or pulled in through the research panel.

• NotebookLM treats all sources as one connected research base, not separate files in different folders.

• Weak sources produce weak output; NotebookLM 2.0 amplifies whatever foundation you give it, for better or worse.

1. Add Your First Sources

Open a new notebook and add sources either manually or through the research panel on the left, which pulls in relevant material for you.

Let’s say you're planning to launch an online course on freelance copywriting. You could drop in:

• Competitor course pages and pricing tiers

• Reddit threads where freelancers ask about getting started

• Your own notes from past client work

• Industry reports on freelance income trends

NotebookLM will treat all of it as one connected research base, not separate files sitting in different folders. So don’t worry.

2. Don't Skip Source Quality

A stronger engine doesn't change one basic rule: NotebookLM 2.0 only works with what you give it.

Weak or thin sources produce a weak deck, a thin spreadsheet, or a chart built on shaky numbers.

So please spend a few extra minutes gathering solid material upfront, and every file that comes out of the chat later reflects that effort.

NotebookLM 2.0 amplifies whatever foundation you hand it, for better or worse.

III. Creating PowerPoint & Excel Files From Chat

This is the change that hits daily use the hardest: chat now produces real files instead of just answering in text.

1. Generating a PowerPoint Deck

Go to the Studio panel on the right side of the screen and click Slide Deck. Select your sources first, then add specific instructions so NotebookLM knows exactly what you want.

Based on the sources in this notebook, create a 10-slide pitch deck for investors. 

Focus on the business model, three-year revenue projections, and competitive advantage. 

Skip the company history section.

A real PPTX file will show up a few minutes later. Every deck comes with a Revise button at the top. To change a slide, type the request there, don’t regenerate the whole deck from scratch.

2. Generating an Excel File

Numbers work the same way, just start with the Data Table button instead of Slide Deck. Select sources that hold your figures, ad spend and sales data, for example, then give your instructions.

Create a table with monthly revenue projections for the next 12 months, broken down by campaign.

NotebookLM pulls scattered numbers from your sources into one spreadsheet, ready to download and keep editing in Excel or Google Sheets.

3. What You Can Actually Edit (Important Limitation)

Excel files edit normally. The numbers and formulas are real data, not images, so changing one number updates the totals automatically.

PowerPoint works differently. NotebookLM 2.0 can export slide decks as PPTX files, but that does not always mean the file will behave like a normal PowerPoint deck.

Earlier NotebookLM users on Reddit reported that some slide exports worked more like full-page images than separate editable PowerPoint objects.

So after exporting, open the PPTX and test it yourself: click into the text, move individual elements, and check whether the layout is actually editable before sending it anywhere.

The Revise button helps with this by letting you edit through a prompt. But that's still not the same as typing directly the way you would in a normal PowerPoint file.

Here's a quick way to check: download the file, then click into any line of text.

• Types directly → that's real text

• Won't let you type → that section is still an image, and you'll need Revise or a manual fix in PowerPoint

IV. Finding Research Gaps While Staying in Control

This is where NotebookLM 2.0 becomes really useful. It can look at your current sources and tell you what is still missing, so you don’t build a deck, table, or chart from incomplete research.

Step 1: Ask What Your Notebook Is Missing

Open the chat and ask NotebookLM directly what your research still lacks.

Based on what we've covered so far, what gaps still exist in this research? 
What topics should I add as sources?

We’ll use the same freelance copywriting course example above, NotebookLM reviews the sources you already uploaded and the conversation so far. Then it points out what your research is still missing.

• Missing financial data → sources cover competitor pricing and tiered models, but nothing to build a 3-year revenue projection from (customer acquisition cost, average conversion rates for course sales pages)

• A blind spot in focus → all the research explains how future students will acquire freelance clients, but none of it explains how you'll acquire those students in the first place

Pick one gap to chase down and tell NotebookLM directly.

Let's look into customer acquisition cost and conversion rates for online course sales pages.

NotebookLM runs the research on its own, searching the web for relevant sources, then comes back with a tailored report along with a list of sources tied to it.

Step 2: Approve Sources First

New material it finds never gets added automatically.

NotebookLM hands you the report and the suggested sources first, then waits for your decision. If you like what you see, click Import to pull both the report and sources directly into your notebook.

This setup protects what makes NotebookLM useful in the first place: every answer stays grounded in material you've reviewed and trust.

The chat takes on more of the searching, but you keep the final say on what counts as a real source for your project.

V. Comparing Data and Turning It Into a Chart

Next, we’ll test NotebookLM with something more practical: comparing 2 files.

This is useful when your data is split across different places. Instead of opening both files yourself and checking the numbers one by one, you can ask NotebookLM to compare them and explain what the data is showing.

1. Comparing 2 Files Side by Side

Upload the files you want to compare into your notebook, then select only those 2 in the Sources panel.

If you're running Facebook ads for an online course launch, you can upload one file with monthly ad spend by campaign and another with daily enrollment numbers from your course platform. Like this:

Then, just ask the question directly:

Compare my course enrollments to what I spend on ads each month. 
Which campaigns are driving the most sign-ups?

NotebookLM returns a written breakdown along with the cost per enrollment for each month: about $79.59 in January, dropping to $31.91 by June. The advertising clearly gets more efficient over time.

For the question on which campaign drives the most sign-ups, NotebookLM says outright it can't answer that, since the enrollment data only tracks total daily sign-ups, not sign-ups tied to specific campaigns.

It works with what the data actually shows, without guessing when the information falls short.

2. Turning the Comparison Into a Chart

A text answer covering several months of numbers gets hard to read fast. NotebookLM often suggests turning the comparison into a chart on its own, but you can ask directly too.

Turn this into a chart showing ad spend against enrollments by month.

A chart appears within a minute, and you can download it right away. The nice part is that it makes the pattern much easier to see than a long text answer.

• Ad spend climbed from $3,900 to $9,000 over six months

• Enrollments grew from 49 to 282, a 475% increase

• Both lines rise together, but enrollments grow faster than spend

The chart makes it easy to see: ad spend went up, but enrollments grew even faster. That means the campaigns became more efficient over time.

3. Other Comparisons Worth Running

The same workflow works for any 2 datasets you want to put side by side:

• Weekly website traffic against newsletter sign-ups

• Time spent on content production against engagement results

• Department budgets against actual spending for a quarterly review

• Customer support ticket volume against response times

Whenever a written answer feels too dense to act on quickly, just asking for a chart, a table, or a downloadable file to turn that density into something you can actually decide from.

VI. A Workflow You Can Copy for Your Own Project

Example scenario: You run a small online store deciding whether to expand your product line.

Then follow the same steps from this guide:

• Start a notebook with what you already have: supplier catalogs, sales reports, customer reviews, competitor pricing

• Ask what gaps exist, and approve any new sources before adding them

• Generate a Slide Deck through the Studio panel for a pitch to suppliers or investors

• Build a Data Table comparing sales numbers against ad spend

• If the numbers start feeling hard to follow, ask NotebookLM to turn them into a chart.

Strong sources at the start mean a stronger deck and a more useful chart by the end.

Every file saves real time, but none replace a careful check before sending:

• Test whether the PowerPoint text boxes actually let you type

• Verify the Excel numbers against your original source

• Confirm the chart matches what the data shows

Conclusion

NotebookLM 2.0 fixes the exact part that used to eat up research time. Instead of just reading and answering in text, it now builds real PowerPoint decks, Excel files, and charts right in chat, generated straight from the sources you select in the Studio panel.

Chat also takes more initiative, spotting gaps in your notebook and suggesting where to research next, though the final call on which sources get added still stays with you.

A few things to keep in mind before diving in: PowerPoint files don't edit as freely as one built from scratch, and this update only reaches Google AI Ultra and certain Workspace plans for now.

Still, the core of it holds up, since strong sources in lead to useful output out. NotebookLM 2.0 doesn't think or decide for you. It just closes the gap between understanding a problem and holding an actual file you can act on.

If you are interested in other topics and how AI is transforming different aspects of our lives or even in making money using AI with more detailed, step-by-step guidance, you can find our other articles here:

• Browsers Are Dying: Codex & Claude Code 'Super Apps' Just Showed What Comes Next

• The New Way to Build Profitable AI Websites With Gemini 3 (It Starts With One Page)

• 3 Easiest Methods to Make Money With Claude AI (Why Most People Ignore?)*

• Google AI Can Now Build Your Entire AI Marketing Content System From One Brand Folder*

• Learn These 5 AI Skills to Determine Your Job Security & Future-Proof Your Career*

  *indicates a premium content, if any