How to Build a Second Brain With NotebookLM + Obsidian (Full Setup)

Every article you've read, and every idea you've had, is either lost in a random app or stuck in your memory where you'll forget it by next week. I built a system using Notebook LM and Obsidian that captures all of it and makes it

searchable with AI. So, everything I've researched this year is in one place I can actually ask questions about. So, in

this video I'm going to show you the full setup so you can build the same system today. The main problem with

system today. The main problem with every note-taking app is that they're built for input, not output. Capturing

information is easy. There are hundreds of apps that do that well. The problem

is getting that information back when you actually need it. You highlight a paragraph in an article, save a bookmark, take notes during a meeting, and screenshot a tweet. All of that information exists somewhere, but when you're working on a project 3 weeks

later and you need that one insight from that one article, you can't find it because it's buried in whatever app you happen to have open that day.

Traditional note apps are passive storage. You put information in and it

storage. You put information in and it sits there until you manually remember to look for it, which you usually don't.

That's the problem a second brain is supposed to solve. A second brain is an external system that captures what you learn and organizes it so your actual brain can focus on using information instead of trying to remember where you

put it. has been around for years, but

put it. has been around for years, but what's changed in 2026 is that AI makes a second brain more than just a fancy note-taking app. Because a system that

note-taking app. Because a system that actually works needs two capabilities: the ability to think about your information, meaning analyze it, compare sources, and answer questions about what you've collected, and the ability to remember it permanently in a format

that's linked, searchable, and yours to keep. No single tool does both. Notebook

keep. No single tool does both. Notebook

LM is the strongest AI research tool available right now, but it's not designed for permanent storage or knowledge linking. Obsidian is the

knowledge linking. Obsidian is the strongest knowledge management tool available right now, but it doesn't have built-in AI that can reason across your documents. [music] When you connect

documents. [music] When you connect them, Notebook LM becomes the part that processes your information, and Obsidian becomes the part that stores and connects it, and together they form a system that neither tool can deliver on its own. The first layer is the research

its own. The first layer is the research layer, [music] and that's where Notebook LM comes in.

Notebook LM is a free AI research tool built by Google, and its job in the system is to take raw information and turn it into processed insight before it ever touches your permanent knowledge base. You start by creating focused

base. You start by creating focused notebooks, one notebook per research topic, not one giant notebook for everything you're working on. Keeping

them separate means the AI's answers are always scoped to the topic you're working on. Inside each notebook, you

working on. Inside each notebook, you upload the sources that are relevant to that topic. These can be PDFs, Google

that topic. These can be PDFs, Google Docs, slides, YouTube videos, websites, or audio files. The free tier gives you up to 50 sources per notebook, 50 daily chats, and three audio overviews per day. The technique that changes the

day. The technique that changes the output quality is how you ask questions.

The default approach is to upload a document and ask summarize this, which gives you a generic summary you could have gotten from any AI tool. The

approach that actually produces useful insight is strategic questioning across multiple sources. You upload five

multiple sources. You upload five articles on the same topic and ask what do these sources disagree about? Or what

does source A claim that source B contradicts? Or which of these sources

contradicts? Or which of these sources provides the strongest evidence for X?

Notebook LM also has source toggling, which means you can turn specific sources on or off to control exactly what the AI draws from when answering.

If you want the AI to only reference your internal documents and ignore the industry articles, you toggle those off.

If you want it to compare two specific sources against each other, you toggle everything else off and keep just those two active. Audio overview is the

two active. Audio overview is the feature that saves the most time. You

click one button and Notebook LM generates a podcast-style conversation between two AI hosts who walk through your notebook, discuss the key findings, and highlight the patterns across your sources. I uploaded a stack of research

sources. I uploaded a stack of research papers on AI adoption trends, generated an audio overview, and listened to it during a run. By the time I got back, I had a clear mental model of the space without reading a single page, and I

knew exactly which findings I wanted to save permanently. That's the role

save permanently. That's the role Notebook LM plays. It doesn't store your knowledge long-term. It processes raw

knowledge long-term. It processes raw material into insight, and the insight is what moves into Obsidian. Notebook LM

handles the thinking, and the next layer handles the remembering. Obsidian is a free note-taking app that stores all your notes as plain text files in a folder on your computer. There's no

proprietary format and no cloud [music] dependency, which means your notes belong to you and you can open them in any text editor, even if Obsidian disappeared tomorrow. The reason

disappeared tomorrow. The reason Obsidian works better than Notion, Google Docs, or Apple Notes for a second [music] brain is the linking system.

Every note can connect to every other note using double bracket links. When

you type two square brackets and start typing a note title, Obsidian auto completes and creates a bi-directional link between the two [music] notes. Over

time, those links create a knowledge graph where ideas from different projects, different time periods, and different sources connect to each other automatically. You can click on any note

automatically. You can click on any note and instantly see every other note that references it, which means you never lose the context around an idea. For

organization, the system uses the PARA method, which stands for projects, areas, resources, and archives. Projects

are active work with deadlines. A

product launch, a hiring plan, and a client deliverable. Areas are ongoing

client deliverable. Areas are ongoing responsibilities without a deadline.

Team management, personal finance, professional development, these stay active as long as the responsibility exists. Resources are topics you're

exists. Resources are topics you're interested in but not actively working on. Industry trends, frameworks you've

on. Industry trends, frameworks you've learned, tools you're evaluating. This

is your reference library. Archives are

completed or inactive items. Finished projects, outdated research, anything you're done with but might want to reference later. Each PARA category is a

reference later. Each PARA category is a folder in Obsidian. Every note goes in the folder that matches its current purpose, and notes move between folders as their status changes. The structure

takes about 5 minutes to create, and it scales to thousands of notes without becoming unmanageable because the PARA categories keep everything sorted by what it means to you, not by when you created it or which app it came from.

[music] The PARA folders keep everything sorted, but the links between notes are where the value compounds because every new note you add makes every connected note more useful. And the workflow that feeds those notes is simpler than it

sounds. The practical workflow looks

sounds. The practical workflow looks like this, and once you've done it a few times, it becomes second nature. You

start in NotebookLM, upload the sources for whatever you're currently researching, ask strategic questions, and let the AI surface the insights that matter. When NotebookLM gives you

matter. When NotebookLM gives you something valuable, you don't copy-paste the AI's output into Obsidian. You write

the insight in your own words as a concise note because a note written in your own framing is something you'll actually understand and use 6 months from now. An AI-generated paragraph you

from now. An AI-generated paragraph you pasted is something you'll skim past and forget. You create the note in Obsidian

forget. You create the note in Obsidian with a clear title that describes the insight, not the source. Retention drops

when onboarding exceeds three steps is a useful title. Notes from McKinsey

useful title. Notes from McKinsey article is not because in three months you won't remember what that article said or why you saved it. Then you link it. This is the step that turns a

it. This is the step that turns a collection of notes into a second brain.

You connect the new note to every existing note in your vault that it relates to. The retention insight might

relates to. The retention insight might link to your product strategy project, your onboarding research resource, and your customer experience area. Those

links mean that when you open any of those notes in the future, the retention insight surfaces automatically. I was

preparing a presentation on how AI is changing hiring practices. I created a Notebook LM notebook, uploaded six recent reports and articles on the topic, and asked three strategic questions. What's the biggest

questions. What's the biggest disagreement across these sources? What

trend do all six sources agree on? And

what's the strongest data point in any of these sources? Notebook LM came back with cited answers for each one. I

pulled three insights from the responses, wrote each one as a note in my own words, placed them in my presentation project folder in Obsidian, and linked each one to my existing notes on AI workforce trends and talent strategy. That

strategy. That I built from those notes was stronger than anything I could have assembled by reading all six reports manually and trying to cross-reference them in my head. Every session like that adds a few more connected notes to your

vault, and the value of those connections is something you don't feel on day one but becomes obvious after a few weeks. After one week of using this

few weeks. After one week of using this system, you'll have somewhere between 10 and 20 linked notes in your vault. At

that point, the connections are sparse and the system feels like extra work compared to just writing notes in whatever app you had before. After one

month, you'll have 50 to 100 linked notes. And that's when the system starts

notes. And that's when the system starts working for you instead of you working for it. You'll open a project note and

for it. You'll open a project note and see links to insights you captured three weeks ago from a completely different context that are suddenly relevant.

Connections you didn't plan start appearing because the linking structure surfaces them automatically. After three

months, your vault is a searchable knowledge base that reflects how you think about your work. Every research

session, every meeting takeaway, every strategic insight is in one place, linked to everything it relates to, and retrievable in seconds. There's one

trick that ties both tools together in a way most workflows miss. You can export your Obsidian notes as markdown files and upload them back into Notebook LM as sources. That means you can create a

sources. That means you can create a notebook from your own knowledge base and ask the AI questions about everything you've ever written down. I

did this with 3 months of notes on AI trends and ask Notebook LM, "Based on everything I've collected, what's the biggest pattern I haven't explicitly connected yet?" It flagged a

connected yet?" It flagged a relationship between two trends I had noted separately but never linked. And

that connection became the thesis of a presentation I gave the following week.

That's the loop. Notebook LM processes new information into insight. Obsidian

stores and links that insight permanently. And when you feed your

permanently. And when you feed your Obsidian notes back into Notebook LM, the AI can reason across your entire knowledge base and surface patterns you missed. The system gets smarter every

missed. The system gets smarter every day you use it and it never forgets. And

the one thing that makes this system even stronger is filling it with high-quality knowledge from the start. I

put together a breakdown of 13 [music] free AI courses that cover everything from research to automation and you can watch that right here. Thank you for watching and I'll see you in the next one.