15 NotebookLM Hacks That Change How You Work

[00:00:00]Here, I have 23 sources in this Notebook LM file. I just asked it a very specific question about my journey and learnings about distributed systems, and it gave me exactly what I was looking for. That is not a generic answer. That's an AI that has read every single document that I want to read, that I have given it, and is using my sources to answer my specific questions. That [music] is Notebook LM. Hack number one, the most underused thing about Notebook LM is what it actually accepts as [music] sources. Most people know that you can upload PDFs and files, but watch this.

[00:00:36]I'm going to paste a YouTube URL, which is a 2-hour lecture on operating systems, straight into the sources panel, and it's all done. Notebook LM just pulled the entire transcript of that video, 2 hours of lecture content, [music] and indexed it as a searchable source. You did not have to copy-paste anything. There's no manual transcription required. It's one single URL, copy-paste it, and you're done.

[00:00:58]Now, you can add up to 50 sources per notebook, which means you can pull an entire playlist of lectures and query across all of them all at once. Hack number two, you can add full textbooks, not just PDFs, but entire EPUB files into Notebook LM as a source. Now, the standard format for digital books is EPUB, which means that you can load an entire textbook [music] as a source.

[00:01:23]This is me adding Designing Data-Intensive Applications, which is a 550-page book, directly into my notebook.

[00:01:30]Notebook LM becomes a tutor that has read every word of your course material or textbooks and can direct you straight to the right page. Hack number three, you can upload lecture recordings directly into Notebook LM, and it accepts audio files, [music] which means if you recorded a lecture or a team meeting or a normal meeting, for that matter, you can upload it and query it.

[00:01:50]If your professor records their lectures but doesn't post the slides, this is how you turn a 90-minute audio into a fully searchable knowledge base, not to mention you can also convert it into an actual slide deck with just one single click. Hack number four, the button panel on the right. This is probably the most visible feature in Notebook LM, and most people only ever click one or two of these buttons. But, if you look on the right, you have nine different buttons with nine unique use cases. Let me go through all of them. The first one is the reports button. You can click on that button and create detailed reports about any of the resources that you have added into your notebook. But, the little trick is to click on the arrow button and go within that, and you have different layers. You can create your own report, you have briefing docs, study guides, and blog posts. Not to mention that you can go within each of them and customize them further by choosing your own language, and also creating a detailed prompt on what you want that report to be about. This gives you a very granular control over the output of your reports. Now, my favorites are study guides and briefing docs. Study guides can give you a structured breakdown of all of the key concepts across all of your sources with review [music] questions at the end. You can use this before exams or even interviews. I really wish that I had something like this when I was a student, because it's just so helpful doing things like studying for interviews. You can add multiple sources, let's say for data science or AI/ML and other important topics, and create study guides for specific topics that you need help [music] in. The next feature that I really like is the briefing document. You can go into reports and create [music] a briefing document, which is basically a summary of everything in your notebook. You can use this before meetings or to send your manager when they ask you, "What's the status of this one thing?" You can also create an FAQ document, which generates the top questions someone [music] would ask about your topic with answers pulled from your sources. Now, this is great for documentation or onboarding of new team members. Now, all All this is customizable. You can [music] just create your own report or have it be whatever that you want it to be. All you have to do is add multiple data sources and it'll simply give you that [music] specific kind of report for your added sources. Now, most people hit study guide and stop, but if you run all five, they each reveal a different shape of the same information and often the one that you didn't think you needed that is actually more useful. Now, the next hack is [music] prompting a podcast. You can actually prompt podcast and generate an audio overview from your sources and turn into a two-person AI conversation.

[00:04:27]Most people hit generate and get a generic overview of everything. But, here's how you get something that's actually so much more useful. Before you hit generate, click >> [music] >> the edit icon. You can get a customization panel and you can set things like how deep you want the conversation to be, do you want it to be explorative, do you want it to have opposing perspective, or do you want it to have just a quick summary? Now, instead of a 20-minute overview of your entire notebook, you get a focused 8-minute conversation on exactly what you asked for, framed for your specific context. I personally really like this when I have to do a lot of research on a topic, but honestly feel lazy. I simply create this audio podcast [music] and I listen to it when I'm out on a walk or simply doing some chores. It's a great way to passively learn a new topic in [music] honestly a pretty engaging way.

[00:05:13]The next hack is that you can interrupt your podcasts and ask specific questions. Now, this one genuinely feels like a superpower. While the audio overview is playing, you can hit the interactive mode button and a [music] text input box appears. The AI pauses the show, answers your questions using your source documents, and [music] then picks up exactly where it left off. It's like being able to raise your hand in the middle of a lecture and have the professor answer you in real [music] time. Except the professor has read everything in your notebook. Now, the next hack is query across 20 documents at once. [music] The best part for me about Notebook LM is that the chat doesn't query one document at a time. It queries all of your sources >> [music] >> simultaneously and it can find contradictions between them. You can input 15 different such research papers about a topic and then ask it to find the one that has a concept or idea that's unique and different from everything else. [music] It makes knowledge synthesis so much easier.

[00:06:09]Think of it like a Google or a Wikipedia for your own specific knowledge base.

[00:06:14]The [music] next hack is asking it what's missing. Now, this is a query almost nobody thinks to run [music] and it's one that is probably one of the most valuable ones when it comes to your entire knowledge [music] base. It's doing a gap analysis of your own sources. Now, most people or tools tell you what you already have and know, but Notebook LM can tell you what you are missing, which is infinitely more useful when you're preparing for something that's super specific. What you do is you ask Notebook LM to use all of your existing sources and tell you what you're missing to achieve your specific goal. Notebook LM is connected to Google that can actually search the internet in real time. This lets it tell you what you are missing on. Now, you can add sources on those missing [music] topics, you can run the query again and repeat this until the gaps close [music] and you effectively don't not know anything. Now, the next hack is information alchemy. It's as cool as it sounds. This is one of the most powerful workflows in Notebook LM and almost nobody really uses it. The context. Notebook LM has a limit of 50 sources per notebook. When you are doing serious research, you'll hit it fairly quickly. Now, here's how you work around it. Step one, with all of your sources loaded, click briefing document. This briefing doc is essentially a distillation of everything that you have in those 30 [music] sources written by Notebook LM itself. All of the key ideas synthesized. Step two, save this output as a note. Step three, in the sources panel, uncheck all of your 30 sources.

[00:07:45]Step [music] four, add the saved briefing note as a new source. Now, instead of the 30 documents that you previously had, you just replaced it with one single synthesized note. You can now add 49 completely fresh new sources, effectively compressing the previous sources you had into [music] one single dense source, and layering 49 more on top. You are effectively getting the analytical power of 80 sources within a 50 source cap. I call this information alchemy. You're transmuting raw material into concentrated knowledge, and then loading the next layer. The next hack is saving AI responses [music] and combining them. Now, when Notebook LM gives you something useful, don't just read it and move on. Every answer Notebook LM gives you can be saved as a note, and notes [music] stack up across session. Now, you can export these notes to documents or even a Google Sheet, and over time you can keep [music] on researching and learning about any topic. Now, you basically have drafts on a report or a research paper that you need to write without having to prompt your way to do so. The next hack is comparing two versions of a document.

[00:08:49]This one saves hours for anyone who works with specs, papers, or documentation that evolve over time.

[00:08:55]Now, here I've got two different versions of the same technical spec sheet, and if I were to find out what [music] is the actual difference, it would take me hours to manually read through all of them. Instead, [music] I add both of them as sources, and I simply ask my notebook what is the difference. And it summarizes it for me without me having to look through any of them. Now, [music] hack number 12, ask about diagrams and images in your PDFs.

[00:09:19]Most people assume that Notebook LM can only read text. [music] It can't read images, but in fact, it can. Notebook LM can read diagrams, ER schemas, [music] flowcharts, and architecture images inside your PDFs. If you upload a research paper that explains its methodology with figures, [music] you can ask about the figure in plain English. No more staring at a confusing diagram for 10 minutes trying to reverse engineer it. The next hack is using it as a code base navigator if you are a team lead. Now, this hack turns your existing documentation into an interactive onboarding assistant. New engineers stop asking you the same questions week over [music] week. Now, I have loaded a project's readme, API docs, and architecture notes into my sources. Now, watch. NotebookLM just generated an onboarding guide for the project's own documentation. This honestly saves so much [music] time.

[00:10:07]Now, the next hack is sharing a notebook as a living knowledge base. NotebookLM lets you share an entire notebook with anyone via a single link. But, the part most people don't know is when four or more people are actively using a shared notebook, you get interaction analytics.

[00:10:24]>> [music] >> You can see in real time which sources people are visiting the most, what questions people are asking, and which parts of your knowledge base can't answer them yet. Now, this is particularly useful for engineering [music] teams. You share a notebook with all of your onboarding docs, architectural decision records, and run [music] books, and new engineers get answers instantly instead of pinging everyone on Slack. If you [music] are a TA or professor, you share a notebook with course readings and let students query it directly. And instead of a static wiki that nobody really reads, you now have a living knowledge base that answers every single question and shows you what people need and tells you what's missing. Now, putting it all together. Let me show you how these combine in a real workflow. I'm preparing for a distributed systems interview, and here's the full loop.

[00:11:12]There are YouTube URLs to MIT Open Courseware lectures on distributed systems. There are actual EPUB files of textbooks that I've been referring to.

[00:11:20]There are three research papers on Raft, Cassandra, and distributed transactions, and my own [music] notes from a previous course exported as a PDF. All of these sources are loaded onto my notebook. I generate a study guide first to get the full picture. Then I run on what's missing query. I realize that there are two gaps >> [music] >> which I did not know about. I add two YouTube lectures on those topics and then I set up a focused audio overview asking it to prepare me for a senior engineer system design interview. I listen to it on my commute and when something doesn't click, I use the interactive mode to ask a follow-up without stopping the podcast. That is the full loop. You ingest [music] broadly, you find your gaps, you fill them, and then you synthesize in whatever format fits how you learn. A written guide or a podcast, a quiz, or even a first draft. Notebook LM isn't a note-taking app. It's a research operating system. Once you start using it like one, everything else feels slow.

[00:12:20]So hopefully after watching that video, using Notebook LM has become incredibly easy for you. If you like this video, please subscribe to our channel and you should watch this video on cloud code next.