NEW NotebookLM Update Is INSANE!

New Notebook LM update is insane. Notebook LM just got the biggest update in its three year history and as of today it's live for every single Google AI Ultra user around the world. This is the Google tool that lets you dump in your notes, your PDFs, your research and ask it questions about all of it. Up until last week that's mostly all it did.

Now it writes its own code, it builds spreadsheets, it makes slide decks, it finds its own sources on the web before you even upload a single file. Google ran its own tests, old version against new version side by side across five separate evaluation categories. The new version won more than 65 times out of 100. In one specific test finding good sources on the web it won almost 8 times out of 10.

That's a huge jump for one update and the part most people are going to skip right past, the secure cloud computer sitting inside every notebook now, that's the part you really want to hear about. We'll get there in a second. Three years ago this tool started as a small experiment out of Google labs. The whole point was simple, you feed it your stuff and it helps you understand it faster.

That part hasn't changed. What's changed is everything around it. Back then you had to show up already organized, you needed your sources ready before you even opened the tool. Today you can open a blank notebook with nothing but a question in your head.

Notebook LM will go out, search the web, pull in solid sources and drop them straight into your notebook already cited. The old notebook LM could read your stuff and talk about it. This new one reads your stuff then goes and does something about it. Quick one before we keep going, a lot of the people inside the AI profit boardroom run small teams or run things mostly by themselves.

Tools like this are exactly what we build our playbooks around. Drop in your old coaching core recordings or your training guides and ask it to pull out the questions people keep asking, then turn that into one clean guide instead of leaving it buried across a dozen separate recordings. That's the kind of automation we walk through, live every single week, built around the tools people are actually using right now. Links in the comments and description or head to aiprofitboardroom.com.

To understand why this update matters so much, it helps to know where notebook LM was just six months ago. Late last year Google folded notebook LM into its top AI ultra plant for the first time and gave those users a jump of around 10 times the daily limits on things like chats, audio overviews and reports. That alone was a big shift but this new update isn't about doing the same things more often, it's about the tool doing different things altogether. The two Google staffers who wrote the announcement, Trond Völner who leads product for notebook LM and engineer Usama bin Shafkat pointed out that this whole thing started three years ago as a small labs project and today it's used by millions of people and organizations as a research partner that helps them organize their thinking and spot connections across their own documents.

Going from a small labs experiment to this in three years is the kind of curve worth paying attention to. So how is it actually doing all this new stuff? Two names matter here, Gemini 3.5 and Antigravity. Gemini 3.5 is Google's newest model family and it's the brain running the whole show now.

Antigravity is Google's own coding tool, the same engine its software teams use to build real products. Google took that engine and wired it straight into notebook LM. Pairing the two together is what lets it move past just talking about your information into actually going and doing something with it. Every single notebook now comes with what Google calls a secure cloud computer.

Picture your notebook getting its own small laptop sitting inside it, one that only it can touch, fully locked down from everything else. When you ask it to crunch numbers or build a chart, it doesn't guess at the answer. It opens that little computer, writes real code, runs it, checks its own work, then hands you the result. That matters more than it sounds like because guessing used to be exactly how these tools messed up.

Hand an older AI model a giant spreadsheet of sales numbers and it might just predict what the answer probably looks like instead of actually doing the math. This version runs the math for real every time. Picture handing it a spreadsheet of how many new members joined a community last quarter, broken down by which video or post brought them in. It doesn't estimate the answer anymore, it actually calculates it.

Top of that little computer, Google packed in more than 100 ready-made skills. Picture a toolbox that already has every tool sorted, labeled, and sitting right where you need it. Instead of a pile you have to dig through yourself. Those skills cover things like reading messy spreadsheets, comparing documents that are written in totally different formats, pulling numbers out of long PDFs, and building clean visual reports from raw data.

You don't pick the skill yourself. Notebook LM picks the right one for the job based on what you actually asked it to do. Think about what that actually replaces. Before this, if you wanted to turn a stack of messy notes into a real chart, you'd open one app to clean the data, another app to build the chart, and maybe a third app to write up what it meant.

Now that's one ask inside one notebook and the secure computer handles every step in between without you touching a single other piece of software. The way you start a project has changed too and this part is easy to miss. Before this update you needed your sources lined up before you could really get going. Now you can start with nothing but a loose idea or a question and the chat itself will help you build out a source list as you talk.

Want material in another language to get a different angle on something? It can go find that for you. Looking for more work by a specific author you just came across? Same thing.

It leans on Google search to go find solid relevant sources out on the web and drops them straight into your notebook fully cited instead of you doing that legwork by hand. Say a member inside the AI profit boardroom wants to start selling to a different country. Instead of spending a weekend digging through forums in a language they don't speak, they can just ask and notebook LM goes and builds that research folder for them. This is also where the new output formats come in and this is the part that changes how you'll actually use it day to day.

Before notebook LM mostly gave you text on a screen or maybe an audio summary you could listen to. Now you can ask it for an actual finished file, a pdf report with charts and tables already laid out for you, a word document or a plain markdown file if that's what you need, a spreadsheet fully built with the formulas already in place, a slide deck ready to present to your team, raw data files too like csv or json if that's what your other software needs to read. You can even ask for a chart as an image or a simple picture using Google's own image tool baked right in and once it builds the file you're not stuck with whatever it gives you first. You can ask for edits the same way you'd ask a person to fix one slide.

The image piece is worth a second look too. It's powered by Google's nano banana image tool sitting right inside the same notebook so if you've got a notebook full of community feedback or survey answers you can ask it to turn the main finding into a simple graphic you could actually post instead of writing a paragraph nobody scrolls past. Google shared a few of its own examples when it announced this and they're worth knowing because they're not made up they're straight from the announcement. One example was a data analyst who gets sales numbers from different countries all in different formats all a mess.

Some countries track sales weekly others monthly some use different currencies or date formats entirely. Notebook LM can search the web for context on each country's numbers write the code to clean and line them all up the same way then hand back a finished chart and a written report explaining what actually changed and why. Another example was a program manager handed a pile of dense technical specs the kind of document most people open once and then quietly avoid. Instead of reading every line themselves they can ask Notebook LM to turn that into a simple guide and a slide deck the whole team can actually use without a single follow-up meeting just to explain it.

Picture that exact move inside the AI profit boardroom take last month's coaching cool notes hand them to Notebook LM and ask for a one-page onboarding guide for brand new members. Same move just pointed at your own community instead of a customer spec sheet. The direction here is pretty clear Google already said more output formats are coming on top of the ones live today. Gemini's separate notebooks feature which Google rolled out earlier this year now syncs straight into Notebook LM too meaning a source you add in one shows up automatically in the other.

Google is clearly stitching its AI products into one connected system piece by piece and Notebook LM is turning into one of the main doors into that whole system. Every update like this one makes the gap between people who use these tools daily and people who haven't touched them yet a little bit bigger and that gap shows up fastest in small businesses where one person is usually doing five jobs at once. One more thing worth knowing if you run a business and you're cautious about where your information goes and you probably should be. The word secure in secure cloud computer isn't just marketing language sitting there for show.

Each notebook gets its own separate locked off computer that only that notebook can use. Your customer list, your sales numbers, your call notes, none of it sits mixed in with anyone else's data while the code runs. That matters if you're the kind of business owner who'd never pay sensitive numbers into just any random tool online. So what do you actually do with this starting today?

If you run a business by yourself or with a small team start by feeding Notebook LM your messiest pile of information first, not your cleanest one. That's where it saves you the most time right away. If you manage other people hand it your driest document like a policy file or a long process doc and ask for a one-page guide your team will actually sit down and read. If you sell things online drop in a few months of order and refund data and ask it to point out which product is actually worth pushing harder instead of guessing from a gut feeling.

If you create content for a living or coach people for a living try feeding it a handful of your past videos or call transcripts and ask it to pull out the three questions people keep asking you then build a short guide answering all three. If you handle customer questions for any kind of business drop in your last month of support messages and ask it to build you a simple FAQ document you can hand to a new hire on day one and if you're already running some kind of community or group the way we run the AI profit boardroom try feeding it a batch of member questions or old cool transcripts and ask it to turn those into one simple list of answers. If you're just curious and don't have a clear use yet open a blank notebook type one honest question you actually want answered and let it go find the sources for you instead of hunting for them yourself. Tools like this are moving fast enough now that keeping up alone is genuinely hard.

Every few weeks there's another update another new way to do the same task and it's easy to feel like you're always one step behind whatever just shipped. Nobody expects you to read every announcement test every feature and figure out what's actually useful on your own. That's exactly why having other people around you people actually testing this stuff inside their own businesses and telling you straight what worked and what didn't matters more now than it used to especially when the tool itself will look different again in another month or two. That's what we built the AI profit boardroom around the moment something like this notebook LM update goes live we're already putting together the walkthrough for it how to set it up how to feed it the right sources how to turn your own messy business documents into the kind of clean reports and guides we just talked about.

There are four live coaching calls every single week where you can bring your own setup and ask questions about it directly with people who are already deep into using tools like this for their own businesses. There's a daily new tutorial too walking through exactly how to set features like this up the day they ship plus a prompt library you can pull straight from instead of starting from a blank notebook yourself. There's a member map as well so you can find and connect with other business owners near you who are already using notebook LM and tools just like it. Links in the comments and description or go to AIprofitboardroom.com and if you want something free first before you commit to anything else go check out the AI success lab it's a free community and inside it you'll find the process and the notes from videos exactly like this one plus more than 100 other AI use cases already laid out and ready for you to use on tools way beyond just notebook LM.

You'll also be alongside more than 75,000 other people in there all figuring this stuff out together one update at a time instead of trying to keep up with all of this alone. Links in the comments and description for that one too.