The Definitive Guide to Setting Up Your AI Second Brain

There are now three brains that we have access to. We have the human brain. We have, let's draw our human here. We have the AI brain, so to speak. Let's do a little robot. We'll make it friendly. And then we have the shared second brain. And to me, this is really obsidian. Hopefully, that kind of looks like obsidian. All right.

So, I want to talk about these three things because there's a lot of talk about, you know, LLM wiki. Lots of people, including myself, use Obsidian, and you want to be able to connected to your agents like Claude Code or Codeex or whatever it is that you use. And so, how do all of these things fit together? Let's break it down and then we'll go into how to set it up in Obsidian. All right, let's go one at a time about what these brains are going to take care of. What are they actually for? And what are they powered by?

So, let's start with the human brain, which is for thinking. And I mean, you know, deep thinking, hard thinking, the type of thinking that actually has a bit of friction. I we're too eager to make all thinking frictionless. We're too eager to make our life frictionless and I think when it comes to using our cognitive powers and actually building our capacity to think it is the wrong choice and we shouldn't default to think that oh thinking should come easy, right? So the human brain is for thinking and it is powered by our experiences, our curiosity, it is powered by our taste, by our judgment, right? This is how we have diverse ideas, right? We could be looking at the same news. We could be looking at the same book, the same articles, but we will all have slightly different takeaways and how we're going to apply them because it is really embodied in how we go through the world and what we're interested in.

Now, the outcome we want to get out of using our biological brain to think is we want to improve our critical thinking, right? So we can understand the world in a more accurate and realistic and deep way. We want to improve, you know, our curiosity feel it. We want to improve our memory to build more connections. We want to uh think about thinking, right? To do this kind of higher order thinking, understanding how we think about things, how can we improve them, right? Build mental models about the world. So this is really what our human brain can do. All right.

Now let's move on to the AI brain and start to talk about what it takes care of because it's very different from what we just talked about here. I would argue that the AI brain is amazing at doing things. So instead of thinking on the human side, we are now talking about doing things. And this is powered by LLMs, right? Large language models that are increasingly improving every single day and it's getting better specifically at going through workflows, you know, coding, doing knowledge work. And so what we're seeing is the thinking part is not so great at most of us have seen, you know, it hallucinates. It can't really get to the real point if you know what's going on. But it's great at doing rope work. is great at going step by step, following a system of how to do things increasingly. So it is great at doing and I think just by making this distinction of thinking and doing it makes it easier to see okay where is AI complimentary and where do we not want AI to replace our thinking, right? We don't want to outsource the thinking that's core to our human brain, but we can outsource a lot of the doing that, you know, doesn't really bring any value and doesn't actually help us think through things like, oh, if I have to organize my calendar all the time, probably I can outsource to someone else.

The outcome here is increasing our productivity in leveraging our thinking to get things done, right? It's not just we're in our brain all the time theoretically thinking about all these intellectual things, but they actually need to translate into the real world. And AI is great at doing that. So this is, you know, human brain versus AI brain.

Now when people talk about second brains and I think specifically a shared second brain is where you have all the context in order to be able to power the two things that we just talked about. This is the context that's required to both think and do. Okay. So, we are using this context for thinking and doing. And I would argue that we want to set up our shared second brain so that it is a tool for thought. And what I mean by that is this thought, this tool should improve, protect and really leverage our cognitive ability. So what I mean by that is when we think with it with our human brain, what happens is because our short-term memory is very limited, as well as our long-term memory is very hard to retrieve the right information over time. What happens is if you put the things that you're working on into a second brain like Obsidian, but you can put it anywhere that it helps you go back to the things that you're working on, which helps you improve your cognitive ability. And I'll talk about a specific type of note-taking in cytocasten and why that actually rewires how you think more critically. So there are lots of benefits in having a some sort of second brain and I would say shared second brain so that it improves how you think just by using the tool.

The second thing is of course it will augment the AI brain even though, you know, the LLM has so much information. You've already noticed this that if you give it really quality information that doesn't exist on the internet or at least is a curated version of what you think will help the work that the AI is doing by curating that context. AI does so much better of a job than if it just had the information of the LLM. Right? So that again already improves the cognitive ability of your AI brain because it has the right context to work off of. So if you are wondering, am I setting up my second brain right? It is whether it becomes a tool for thought for you, that it improves the cognitive ability, whether it's you, whether it's your AI, or a combination of both, so that you get about better output at the end of the day. This way your share second brain can expand what's possible to do both by your biological brain as well as by your AI agents brain.

Okay, let's go into then what goes into the share second brain. And gosh, I am just excited even starting to talk about this. All right. So I mentioned first that we're going to split the thinking and doing. So let's write that down. Split thinking and doing because we mentioned right, thinking is for the human and doing is mainly for our AI. That's why when it comes to the information that we put in there, we also want to separate what is the ideas that we are working on, the ideas that we're creating versus what the AI is looking at and creating because number one, we don't want to mix or we don't want to mix human and AI output. And number two, we also don't want to distract or distract, did I spell that right? Distract our thinking from doing. These two are my golden rules.

Number one, not mixing human and AI output is going to be very obvious because I want to know which ideas are mine, which are the ones that I'm working on, which are the ones, you know, the notes that I've read through and I know are credible versus what is AI producing in the background because increasingly it's all AI agents are very eager to help you do stuff and it keeps on creating output that, you know, even if I haven't even reviewed, I don't want anyone one to be looking at it and thinking, okay, this is confirmed, this is checked, this is the things that's worthy of using. So I'm going to separate that. Luckily, it's very simple. I'll show you my method in a moment, but we want to separate the human versus AI um notes.

The second thing is not to distract your thinking from doing. I haven't really seen anyone really talk about this in the second brain space. I think most people see second brain as a pretty productive ive way of using um this kind of vault for example, you know, par the power method is very much about doing things, right? You have projects, you have things that you archive, you have, you know, all of these sections. So I don't use my second brain that way because I am specifically separating thinking from doing. So it in what I do, uh, I make sure I don't distract my thinking by thinking about my to-do list, right? The the thinking is specifically focused on, can I generate insight? You know, can I understand something, analyze something, create connections between notes so that I can understand what's going on and have some ideas of my own? That is the ultimate output for the thinking part of what I do, right? But for doing, this is really, you know, project management. I this really happens in a different. And this is really just, you know, work output output. And so, um, as my psychology professor from first year told us, if you're going to take one thing away from psychology, it's that every single space should have its own purpose. Do not mix multiple purposes into one space because you will get distracted and you won't do things very well. So I am taking his advice for my digital spaces. I'm doing the same, right? If I'm working on thinking about things, I'm going to give myself time. This is my sacred thinking garden. All I'm doing is engaging in remember like the hard thinking. It's not going to be easy to sit down with a big idea and trying to say, "Oh, do I have some insights about this?" Right? Like for example, what should humans do versus what should AI do? Like th this is something you have to sit down, you have to read, you have to research, you have to do a lot of things in order to have an answer of your own. Whereas doing is more like, okay, once I have the result, I can make a YouTube video about it. That is a very different type of thinking process that you go into because it's more like, okay, here's a workflow, how do I take it from step by step, and AI is really great at assisting in those kind of work um workflows. So we want to separate those two and I just want to highly emphasize thinking is a really difficult thing to do. So don't distract yourself with the cheaper doing of it, especially now that AI can just like very excitedly do stuff.

So how does this splitting actually look like in Obsidian? I'll show you mine first, which is more messy, and then I'll show you how you can get yours started if you are starting from scratch. So basically mine is everything in my folder which is called zero casten. I'll talk about that in a moment. Everything in my folder is for my thinking except for this one um folder which is called AI context, which is specifically for AI and it's got all of these different things that I'm working on, these projects that uh AI can use as context. So that's how I separate it. But if you are starting a vault, if we create it and you have literally nothing, I would just create two folders. This one for thinking and we'll do one more for doing. And the doing will be for AI and the thinking will be for you. And then, you know, you can put everything in there. Let's first talk about thinking and then we'll go into doing u and for how to set up our obsidian. So that it helps us think and it helps us do.

All right. First of all, let's talk about thinking. So we are going to for the human thinking part, we are going to use a note takingaking method called zetle casten and zetto casten means note and casten means box and this is used to be a physical note takingaking uh method that was started in the 18th century and it was popularized and refined by the German sociologist I mean he's really a polymath academic who really refined the metal of the zetto casten um and he used I think it was like 90,000 plus notes in his notebooks in his zetto casten to write 70 plus books on all sorts of topics like politics, sociology, system thinking, you know, art, like he he was really amazing at creating all of this and 400 plus academic articles. And so this guy really showed how you can break down big ideas into atomic notes and connect them so that as you process them, you uh really have a good grasp of how to use the knowledge and how to create go from input to output.

Okay. So in a nutshell, zettocasten as a note-taking method helps you go from input to output. And in the middle of that, there's processing of the idea. All right. So I I did a whole video on this. So if you want to go into details, you can check it out. But I will give you the cliff notes version, which is after using it for four plus years. Here are the three key things that you have to have in your Obsidian.

Number one is you have to have atomic notes. This means that one note is one idea. So let's say, you know, you read an article or you listen to a podcast, there might be, you know, 10 ideas that jumped out at you. Each one of those ideas will be its own note so that you can express exactly what you learned inside that idea and be able to articulate that idea. So the cognitive ability that we're building is to identify and distill the key points. Right? As you can see already, you are processing the input that you're getting by figuring out which ones are the points and how to distill them into an atomic note. Uh, which is a very small, very succinct version of that idea.

Which brings me to the second one is if you heard 10 ideas in that podcast, in that book, in that article, right? Then you're having all these atomic nodes. So you want also to connect. Whoops. Connect your notes. Connected notes. So that's the second principle of Zasten, which is, you know, you have all of these little ideas floating around that look like post-its. Now you want to connect them like, oh, these two are together, these two are together, and this is how you build a web of knowledge, right? Again, you can see how this helps you go from input to output because not only are you making connections between new concepts, you're also making connections between old concepts, right? For example, let's say, you know, I'm reading the organized mind, right? There's I like a thousand ideas in here that are interesting. Well, because I think first in atomic units, every single time I'm reading each of the chapters, I'm thinking through, okay, oh, here's one idea that I identify. Okay, here's how I'm going to distill it and explain it as one idea, right? But then in each chapter, there will be, I don't know, 10 ideas. So then I would think about how do they connect to each other? How do they build in each other? Is it like a hierarchy? Is it like a lateral type of relationship? I'm starting to think about the connections between the notes, right? And let's say, you know, all of the concepts in his book is new to me, let's just say. Then I'm making connections between new ideas so that I can better understand them. I can better remember them, as well as I can connect them to old ideas. Right? Let's say as I was reading this, my mind was going crazy because, you know, it talks about how our mind structures information. It talks about how our mind naturally think and some of the counterintuitive ways that our mind thinks. And it just made me realize, you know, oh my gosh, like this is why framework thinking works because our mind naturally likes to create categories, right? Frameworks are great at doing that. Our mind is not great at keeping up you with, you know, a thousand different ideas, but it works really well if you have a meta structure like a scaffolding around your ideas, which is exactly what frameworks do. So now I have all of these connections between new ideas, between old ideas, and that is how I'm going to retain that information, right? As we've talked about earlier here about tool for thought, the biggest mistake I see with people building second brains is they're just trying to catalog everything that they know or they find interesting, and then you don't do anything to help augment how you actually think, right? But with zettocastin, it's very different because it trains number one, your mind to think in atomic notes. So every time you read, you know, I literally see how I think is very different. Instead of trying to absorb everything that's there, I'm looking for where's this atomic idea, where's this atomic idea, and I start to pick them out. I see them as separate things but related. And now I see how do they connect, not just to each other, but to the things that I know, which is of course how we learn, right? Is as Charlie Mer says, you know, we need to we need to hang the knowledge into a lattice work of knowledge, otherwise it's not usable. Sorry, this is my ugly version of the lattis work of knowledge and information. Right. So now I'm hooking all of these new things I'm learning onto what I already know, as well as onto each other. So it's a lot easier for me to work on these.

Which brings me to the third principle that in zettocast so important is that the processing of the information is the most important action. It is not documenting things, right? It is not, okay, I'm going to take take, you know, a thousand notes on this alone and call it a day. It is we have to process the information there, which is why he says, you know, you have to rephrase each of the idea in your own words, in your own words, right? You have to connect them, which is also a form of processing. You have to break them down. And I would argue that, and I'll show you in a moment, is every single note, you should spend some time on the note title in order to actually articulate what you think the idea is because it helps you come back to it and build on it and to build on ideas over time. So you will see that by processing notes, which is really notes stands for knowledge, right? By processing them, you'll see how you actually improve your cognitive abilities.

What all of this helps you do is it improves your memory because not only are you working on the idea and through multiple iterations making it atomic like and then connecting it to other people and other people other notes, as well as processing it, rephrasing it in your own ways, but also working on it over time so that you're keeping on coming back to these things improves your memory, right? It improves your critical thinking because now you are searching for how does it link to others? How does it affect each other? You can also have lots of tools that help you do this critical thinking. I talked about the idea compass in this video here. You can check it out later. It helps improve your creativity. And I don't mean the artistic kind. I mean the kind of you can generate your own ideas. You can generate your own frameworks by understanding how things work, right? Like after I read the organized mind, I can create new ideas based on what I know about framework thinking, what I know about system thinking, what I know about, you know, creating second brains that this book also helps us think through. It helps with creativity and of course it helps with whoops metcognition, thinking about thinking, because you will notice as you're processing notes where do you get stuck, right? Where do you feel the friction? What's easy to you? And as you think about thinking, you can make this system more of your own. Like for example, I started zeta casten the classic way, which is you have you have three three types of notes, notes, you have fleeting notes, you have literature notes, and then you have permanent notes, right? And the idea is you process the fleeting and literature notes into permanent notes, and then permanent notes connect to each other. But as I'm reading and using the system, one, not in physical form, but in digital form, and two, with my own, you know, cognitive quirks. What I decided to do was both of these type of notes become just raw notes that I put into my vault and then I just process directly onto them so that they naturally become permanent notes as I work on them. And that is a process that works for me. So you can really start to use improve all of these cognitive abilities and really create a system that works for you because thinking is uh quite unique to every single one of us and we all have our superpowers when it comes to thinking. So you want to be able to go through this process and then figure out something that really works for you.

What you'll notice here is this is very much a bottom-up approach. And you'll see in a moment when we talk about AI and the doing section, that's very much a top-down approach. The amazing thing about bottom-up approaches is as your ideas go forward, you'll see they start to pull each other and there starts to be centers of gravities and you'll notice that those are areas of ideas that you're working on, that you you're curious about, that you spend a lot of time digging deeper into. So it starts to reveal patterns about how you think. This is really great already, but on its own. But later when we combine it with AI, and this is why I say you want to have a shared second brain because AI is great at then noticing those patterns, whereas humans, you know, take a lot more time looking at these notes and trying to figure out what is the actual pattern. So that's why uh I choose to share my ideas with my AI so I can get the most out of both my brain as well as the AI's brain. If you have other, you know, um, strong opinions about how to use AI, you don't have to use it. You can just use it, you know, just between you and Obsidian, which is completely local. It doesn't connect to anything, uh, in the cloud unless you choose to do so. So, you can have the same sort of effects uh, with your notes. So, this is really bottom-up thinking.

Now, let's get specific on how to set up your zettocasten note. And specifically, I want to focus on the anatomy of the note because this is where all of your thinking is going to happen. So, I'll show you in Obsidian how to set that up. But before we do that, I just want to mention, you know, what is Obsidian in the end? Is basically just a reader. A reader of markdown files. That's it. Okay. Markdown files. You probably see have seen this MD files. You know, uh, AI loves to read MD files because they are really lightweight. and it's just text. And so, you know, if you send a PDF to a AI, it doesn't just read the PDF. It has to actually process everything in the PDF to understand what's actually in there and then is able to read it. Whereas with MD files, it can just write read directly the text because there's nothing fancy or fluffy in MD file. So obsidian, my very rough drawing of obsidian, uh, is just a reader for these MD files. You can think of, you know, preview is a reader for images. Obsidian is just a reader of markdown files. And so what happens in Obsidian is you have to first choose a folder for Obsidian to know which folder to read. And then within that folder, you just have a bunch of MD files. MD files. All of these MD files. And then it just shows up on the side here like, oh, this is file number one, file number two. And when you click on it, you can actually read what's inside.

Now, why that h that matters is because MD files are really ugly by default. Like you see this, this looks no different than Google like a Google form. See the MD file. It actually looks like this. Okay. It plain text version. It has these little symbols to do um, you know, italics. It uses these different symbols for links and stuff like that. So it's actually a bit hard to read for humans. And that's why Obsidian just makes it easier for you to read. Same as a Google doc. So that uh, you can use the same um, shortcuts as you would use on a Google doc. So for example, if you want to bold something, you just select it and do command B for bolding or control B if you're on a PC, right? If you want to italicize it, you you command I and it becomes italicized or, you know, on PC, you control and I. So that's basically why people use Obsidian. It's just easier to manipulate. And of course, because it works on a folder with a bunch of files in it, it is stored locally. You're not using Notion's cloud. You're not using Google's cloud. You're not using anyone's cloud. You're just pointing Obsidian to open up a folder on your computer and you're looking at the files inside. That's why Obsidian is awesome. So the technology itself is not the groundbreaking thing. Is how you think and do stuff inside the vault that makes it amazing. Which is why I want to talk about the anatomy of the note and specifically how to take it the zettocast way so that you get all of this benefit to your cognitive abilities.

So let's do this. Um, as you'll see, these are folders and these are notes. So you can drag your note inside a folder and then you'll see that, okay, it's inside this folder. You can drag it here and then it becomes a note inside that folder. So, it's all very easy and intuitive to use. But the main thing to do here is to create notes. You can create notes of several different ways. You can at the press here to open a new tab and then click create note. You can also press this to create a new note or you can just do command N to create a new note. So very similar shortcuts as most of the word processing documents.

There are a few things that's very important inside a note and I'll talk about four things specifically. Right? So this is the MD file. First of all is the title and as I've mentioned before, you want this to distill your atomic idea. So first, we want to articulate the atomic idea inside the note. So this one will be called anatomy of a zetto casten note. All right. So it should distill what it is that we're looking for. Let me give you an example in my vault. Uh, for example, this one, philosophers are the OG rebels. So without reading any of the body text, you understand. Okay. This one is about how philosophers are, you know, rebels. Great.

Then of course, we come down to the body. So this is where you type the body of the note, right? So as you'll see, it's different. Automatically the title is using a header and the body is using paragraph form. So the body of the note is where we're going to record down uh details. So in my example, I had a quote from the book the stone reader and then I rephrased what I think is the point, which is, you know, the philosopher's job is to question authority was given, right? So they're rebelling and they're thinking for themselves. That's basically the key idea for me. All right. So at least there is a body text form. But you'll notice that I also have these weird links under here and this is where we get to the power of the connected notes. Again, let me show you here. So, what you can do is connect notes by using double square bracket. Once you click two double square brackets, you will see a list of all your notes here. As you can see here, because we don't have many notes in here, that's why uh we only have two that show up. I'll just use welcome as the other note. And what you'll see is now my anatomy of a zettocast note is connected to welcome. When I click on it, it will open and I can go back by uh just pressing this. So it's all very intuitive uh navigation.

Okay. What this does and specifically why I want to link nodes are twofold. One, I want to connect source notes so that I don't forget where the ideas come from. And if you have been using Obsidian for a while, you probably have experienced this at one point. When I started, I wasn't very disciplined in documenting all my sources. And then I got so frustrated because I would come back to a amazing note or a month later, five months later, and I have no idea where it come from. It doesn't even look like my writing. I mean, I know I wrote it, but I just don't know where it comes from. And there is no connect notes anywhere. So, it's just a floating orphan and I can never find the source again. So, make sure you connect your sources, okay? By just simply doing a double bracket and connecting to the note.

The second thing you want to do is to connect to related notes because again, when we talked about here, the whole point of using zetto castin is so that you can connect to different notes, right? So how you do that is by like what I did here, by doing that. So what I like to do is just create a related section so that I see, okay, here's the body of text that contributes to the main idea, and here are the related ideas that are here. So I did three here. There was the philosophy master class um about pursuing philosophy is a serious way to be a non-conformist. I have a question here. What is a philosopher and what is the purpose of art and philosophy? These are the notes I already had in my vault. So, I'm connecting to them. And what I can do is click into these and it will give me the source notes of the other note. And this way I can go down rabbit holes within my thinking garden. All right. So, that's connecting your notes. You have the body, and then you have your connected source. And I will draw the square brackets here as well as the related notes.

Now, of course, you know, you I make them very easy to read by having these the main body section and then having these kind of as separate sections of the note. But you could be someone who loves to write in prose and you work in the different uh parts of the note. That's totally awesome as well. You know, your note is really your specific way of thinking about something. So do something that's natural for you. But you do want to have some links to other notes. My rule for myself is wherever possible, if I add a new note, I need to add at least one connection so it's not floating somewhere else and I can retrace it back somehow. Uh, but that doesn't always happen. If I'm in a hurry or something like that, but really at least having one note connected to a new note is the best way to grow the web, the lattice work of your ideas.

So once you have all of these, you've done naturally a lot of processing of your ideas, right? You've you've rephrased the idea. You've given it a title that distill what it actually is. You've added the source so you will be able to retrace your footsteps and you have related notes so that it builds onto the note by itself. So this is all you need inside a thinking part, the thinking part of your obsidian. There are lots of other things that you can do. You can add tags if you want by using the hashtag key. You can create them. I don't know, you can say processed or, you know, interesting. And you can tag your different notes so that you can find them. You can create metadata inside of your notes. So you do it by doing three dash and you'll see this properties come up. Oh, let me do that again. And what you can do is you can add your tags here. You can add all sorts of properties. You can have date created, let's say. And you can choose what kind of property it is. So it could be date and time. So you can actually add which day it is. You can you can do all sorts of things here. It could be an alias for the title. There's lots of things that that you can, you know, get crazy with, but just to get started, you all you need is just the title, the body, and some connected notes. That's all you need. And if you want more, you can check out this video here. I've walked through how uh I set up other elements of my note. Uh, again, don't spend too much time making it all intense. The goal, the goal of the null tier is to create your own insights. You're just thinking and creating some new ideas and integrating them into your own lattice work of ideas. We'll talk about doing next with AI and how we set up notes for our AI.

Now, let's talk about the doing part of our Obsidian vault for our AI agent. And this is completely different logic. The things that we're putting in is completely different. We don't need to follow Zettocasten. Rather, we want to have five types of information that we put in here. So, one, we have our doing folder. And this unlike our bottom-up version of our thinking, this is really top-down. This is top-down. It's well organized. It's hierarchical so that the agent can better understand what information is there. So we have our main folder doing, like so for example, in my Obsidian vault is called AI context. Doesn't really matter what you call it. There should be one folder that exists that you can point your AI to and say, this is everything that you need to conduct your workflows for everything that's related to what I'm doing. All right.

And then you want to have five different uh sections. Hopefully, we have enough space. Okay. The first one is a about me file. It will be in markdown so that your agent can read it. Everything here, wherever possible, you want to make it into markdown files so that it's more token efficient. But uh, the idea is your about me file should tell your AI agent, give your AI agent context about who you are. So all of the thing here, the main goal is to give context to your AI. So the most important thing, of course, is you, right? Who you are? What are your preferences? What are your standards? What do you not like it to do? For example, I put in, don't use m dashes. You know, the the long dash that AI loves to use. Also, you want to tell it your goals, your visions, the what is it working towards. I did a full video here talking about the about me file with a template on what you can share in order for your AI to get up to speed. So you can find that video here and download the template down below. Uh, I'll put also put a QR code somewhere on the screen so you can get that.

The second thing you want to have inside your doing folder for AI are frameworks. It wouldn't be my video without talking about frameworks. And honestly, this is underestimated. More people need to be thinking about this because what is the number one problem with AI is AI slop, right? Is that it's using very generic and average information inside LLMs, right? We talked about it way back here that what's powering your AI brain is the LLM, right? And it is has ingested the information of the world, which is awesome, but because it goes for the statistically likely next thing, which means even Though there's wonderful information out there in the internet, because it is not common, it is not the average, the most likely thing that AI is going to use, it doesn't actually get taken into consideration when it's generating stuff for you. That's why we want to create context and specifically we want to distill distill things that you already know you know how to do, like your SOPs, your frameworks for doing task X, right? You want to create frameworks for how you review things, your standards, right? What it means to be good, what it means to be bad, like what all of these can be distilled down to frameworks, which then uh, if you are using AI agents, you know, you can turn these into skills, these slash commands inside AI agents. If you don't know what I'm talking about, you can check out this video here. But even if you don't use uh, AI AI agents extensively. You need to at least be able to articulate, if I don't like the average crap that AI puts out, what do I like, right? You can't just complain that and say this is BS, but then not be able to bring something better, right? And the frameworks here do that to say, okay, instead of just using average information out there, I want you to look for, let's say, uh, what's a good one here? Oh, let's talk about how do you prioritize task, right? For example, prioritize task. You know, AI just do whatever is the last thing that you ask it to do. But you wanted to think through a bunch of tasks. Let's say you have a 10 list of 10 tasks. How do you think about it? Very classic example, Eisenhower matrix, right? We talk about urgency and we talk about importance of the task. So if it's urgent, urgent and important, you want to do it. If it's urgent and not important, you want to delegate it later. If it's not urgent and not important, you don't do it. And if it's not urgent but important, you want to do it next, right? And you don't want to do it first. Something like this, right? That is a framework. Once you tell it to your agent, now it knows how to prioritize it. So, it won't default to the average that it uses, right? Say same thing. And and you don't have to use other people's frameworks. You can have your own. For example, I don't like the Eisenhower matrix because it the urgency is usually decided by someone else. So, what I like to use is, you know, I have my own framework. It's still a matrix. It's based on the idea of the Eisenhower matrix, but instead of that, I think about the amount of effort that it needs and whether I can delegate or not. Whoops. Delegate, right? If it's high effort and I can delegate it, then I will delegate it to someone else. If it's high effort and not able to be delegated, then I will do it. If it's low effort and I cannot delegate it, you know, maybe I do it, maybe I don't do it. And then if it's low effort and I can delegate it, definitely I delegate it and get someone else to do it. Like that could be another way that you use, right? The the type of frameworks that you use will reflect your taste, your judgment. So you know whenever people talk about, oh, like people with taste and judgment will will be the ones that create the best output. Where does it go? It's not just fluffy things. You have to be able to distill them down into frameworks so that your agent can use them. You can use them, right? You get scale out of your thinking and this framework comes directly from this thinking work. So let me maybe write it down here, right? Is this thinking is being used here in the doing. It's all connected. That's why I recommend sharing your second brain between, you know, your human thinking as well as the AI doing because those two are connected, right? And so you can share the frameworks this way.

All right. The third thing we want to do as we talked about here is you don't want to just tell, you also want to show. Which is why we want examples. And what examples demonstrate is is show the AI what's good. Because AI is a great student, right? Is like this smart PhD student who's going to be able to understand your frameworks and be able to apply it to his work, but also it is, let me draw that here, right? It is very smart, but it is also uh, what is a good example here? It is also a good apprentice, maybe a ear and eyes. Okay, this is now looking very creepy. I don't know how to draw an apprentice. Anyway, um, apprentice, uh, AI apprentice. All right. So, it learns really well by seeing examples. And here you want both good and bad, which connects right back to your frameworks, right? You have some sort of criteria that you use to say, this is good, this is bad, this is what I want, this is what I don't want. And then if you just show it, even just one example of good, one example of bad, bad, the AI is able to understand what that actually means because it has more context around what good actually means, what do you mean by these words that you use might that might be vague. Okay. So having examples are really important.

Then let's go to the fourth thing, which is uh, in general, you want to be able to give it a knowledge base which contains everything that it needs to know in order to conduct its work. Right? Just imagine your AI as an employee. When they first log in, they get a computer and they have access to these different folders. Maybe it's of past projects, maybe is of, you know, standards, branding, whatever it is, whatever it is that you onboard a human with uh, for the same type of work, you want to onboard your uh, AI agent with. So that's why you want a knowledge base so it acts as a source of truth. This way also you don't have to every single time tell it uh, where things are because you're also going to create one more thing, which is a knowledge map, which describes in all of this mess, right, where are things and it's some sort of map that maps out hierarchies, these, and so it understands, oh, whenever I'm looking for something, I can come back to this knowledge map, which then links me back to the right place inside the knowledge base, and inside the knowledge base, it will have both the frameworks as well as the examples, so I know exactly how to do my work, and I'm going to be able to tie it back to, right, the goal, the preferences of you, so that the work that it does is specific, tailored to you. And this resolves the problem of, oh, AI is just giving me crap. It is not doing what I want it to do. And you'll notice once you get this set up, it becomes a loop. And what I mean by that is AI will create will actually do the thing, right? It follows everything in here, uses the relevant materials to create some sort of output. And if that output is good or bad, you can feed it into this, right? You can say, "Oh, these are the example. Oh, this was a bad example." Put it in the bad example folder. And now you have another good demonstration of a bad example. Or if it's a good one, you can put it in there, right? The output it creates also flows into your knowledge base. Let's say it created, I don't know, a report on the latest industry trends. Well, now it can live in your knowledge base, which grows your knowledge base, and then it gets indexed into your knowledge map. And so in over time, you will create a a context for your AI to be able to do more work.

So if you've heard of Kaparthy's LLM wiki, he's kind of talking about this situation. He does it slightly differently, and I will just show you quickly. Uh, he he basically will get the AI to build and maintain its own wiki. And um, you know, he talks about you can use it for personal, you can use it for research, you can use it for business. But I think a better way of thinking about what he's doing here, specifically because all of the thinking and processing is taken care of by AI, I think is better suited for doing things. It's not really like, yes, you can read a book, but if you're not processing the notes and your AI is processing the notes, you're not actually getting true takeaways from the book, right? Like your AI is busy organizing everything, but you don't remember what's happening or you haven't even read what it's outputed. So I I don't think it's great for thinking tasks, but I think it's amazing for doing tasks. And what he does is he put raw raw sources, which you can, you know, link it to your knowledge base. He uses the wiki form, which is the knowledge map, and he uses schema, which is a document that tells them how the wiki is structured, which is more like an a combination of about me as well as the knowledge map, so that the agent knows what's going on. But as people who have actually played with the uh, with his way of setting things up, it's a very specific way of using AI and a very specific segment of how you want to use your second brain. So I do not suggest that you use the LLM wiki as your share second brain because it's missing so many different parts of the type of processing that you want to do, the type of thinking you want to do, as well as, you know, addressing the biggest problem with AI output, uh, which is it is not tailored, right? And just having AI processing a bunch of your notes is not going to give you the same type of results as if you have specific frameworks, specific about me, right? Specific examples that are filed. So that is uh, the doing part.

All right. This video is getting super long. So I am going to stop here. There there are so many exciting things. I've put everything into a Google Docs and with screenshots of this so that you can better reference this in the future and also customize it to your own type of thinking. If you want to find out more about the details of which buttons to click to set up Obsidian properly, check out this video here. If you want to look for Zettocasten, check out this video here. And let me know of any questions. We got to do some deep dives into this uh on the channel so that in the age of AI, we can not replace our thinking, not outsource our thinking, but work together with AI so it does what it does best and we do what we do best. And what we enjoy. So yeah, let me know your questions and I'll see you in the next video.