how I use Obsidian + Claude to run my life

Summary

Andrej Karpathy calls these LLM knowledge bases. he uses his to learn. I’ve taken it further and built one that runs every serious system in my life: fitness, goals, business, podcast prep, all in one Obsidian vault, all running on one file called the index.

why Obsidian + Claude:

1. user interface. my workouts, my goals, my files, all editable from my phone. easier than living inside any single AI app.
2. AI agnostic. files are plain text on a disk I own. if Claude goes away tomorrow I can point Codex or anything else at the same vault. zero migration.

the index is the secret sauce. every chat starts there. it tells the AI what folders exist, what rules apply, and what frameworks to use. without the index, the AI forgets. with the index, your systems don’t break.

the INDEX framework:

• I, instructions. how the AI should behave inside the vault.
• N, notes. the source material.
• D, database. structured data the AI can read.
• E, engine. the workflows the AI should run.
• X, exportable. you own the files. apps come and go.

the principle underneath all of this: bet on your own systems. apps are temporary. files are forever. the CEO of Obsidian calls this “file over apps.” push it one step further to “systems over apps.”

start small. download Obsidian. create one note called “monthly goals.” write your monthly goal in plain English. connect that file to Claude. tell it: each day, ask me my daily three tasks, make sure each ladders up. that’s day one. you have the whole system in 30 days.

Transcript

the system that runs my entire life

Imagine if your fitness, your goals, your business, I mean every serious system in your life ran in one place on files you own with any AI you want. And you can do that by utilizing Obsidian and Claude, and you unlock something new and powerful. Andrew Karpathy calls these LLM knowledge bases. He uses his to learn, and when he dropped the idea, it blew up with over 20 million views. Now, I’ve taken it one step further and built out systems that run every area of my life, and it all runs on one singular file called the index.

And in this video, I’m going to show you exactly how to build one step by step, so your systems don’t break, your AI doesn’t forget, and you can find out if you’re under investing your time. And when you finally start using Obsidian and AI, you remove the massive limitation 99% of people aren’t even talking about yet. So, by the end of this video, you’ll have a system that works with any AI, it tracks anything in your life, your podcast notes, your fitness, your goals, and all of it for the long term.

Now, if you don’t know who I am, my name is Jerred Moon. I’ve hit the Inc. 500 two different times. I’ve run a total of five companies, and I’m optimizing them all with AI one week at a time, and showing you the systems that work. So, if you’re ready to build something cool, let’s dive straight in. Let’s first talk about why you’d want to use Claude and Obsidian, or really any AI in Obsidian.

why obsidian + claude (any ai)

And the first reason, and the biggest reason for me, is going to be user interface. So, this is a screenshot of a workout that I’ve created using AI, but it’s ultimately in my phone, right? And so, I can easily come in here, I can edit the files, I can add notes, all of those things from my phone, or from any computer that I have access to, which makes it a lot easier than working with files inside of Claude.

So, that’s reason number one. Reason number two, when you’re working directly with files from Obsidian vaults, it is AI agnostic. So, I can work with Claude today if Claude is the number one AI I want to be using, but if I want to switch to Codex, I can easily do that, and I’ll show you how. But, those are the two big reasons I really like to use Obsidian vaults because it makes it so much easier. But I think showing you exactly how this works might be the easiest next thing.

live demo: goals & daily planning

So, let’s dive right into that. I can say, “Hey, can you look at my example 1330 and tell me what I should be go doing today?” And go ahead and bid out build out an example daily file for what I should be executing today. So, it’s going to start working on that. And what it’s doing right now is I have this goals folder and it’s working on example goal goals that I have. But ultimately it has the example 1330.

This is all fictitious. The one goal is launch the sourdough starter online course and reach $25,000 in revenue. And the 1330 is just a goal framework that I’ve created that we use with entrepreneurs. I use for myself. It has all of the strategies and tactics is tactics I should be working on, the personal items, the business items, and then the weekly what I should be doing. So, it’s referencing this file anytime it’s doing anything, right? It’s going through and saying this is what it’s looking at.

Not only is it looking at the example 1330, it’s looking at the index. And the index is the most important part of any of this. It has all the rules, all the things that it should be looking at when it’s helping me build out a 1330 framework, goal-setting system, and telling me exactly what I should be doing today. Cuz I would never rely on AI to just be like, “Hey, what should I do today?” But when it has this checklist, which is over 4,000 words, it has 4,000 words of my system, my framework in here that it’s referencing on how to build these things.

And then it’s pulling the example 1330 that I have and telling me exactly I’m I’ve already told it everything that I want to be working on, the strategies and tactics. Then I come back here and then it tells me exactly what I should be doing. So, it pulled the example 1330 right here and it ultimately says, “Hey, here’s what you should be doing. You should be doing email 10 podcast hosts, blah blah.” And it gives me all of these tasks that I should be doing for my day.

And then the next thing that it’s going to do is it’s going to actually write that file. And we’ll see that file pop up over here in a second. But this is the bit most powerful part of it is the fact that it can do these things and now I can start to reference and work on these files inside of Obsidian. I can move away from Claude. Now I also told you that it’s AI agnostic, right?

So if I opened up Codex right here, I could easily just mount a goal folder. So I have my goals right here and I can start asking the same questions inside of Codex like that I was asking Claude because I’ve already mounted it. So if I just say, “Hey, I want to add a project. I can use an existing folder.” I would select an Obsidian vault and then I can start using Codex or Claude to be able to go through these things.

So coming back here, it did everything that it needed to do. As you can see here, it has the example what I should be working on today. All right, I should be doing the podcast pitch. Again, these are all examples, not things I’m actually working on today. Gives me the monthly snapshot, this week, what I should be doing personal business based off of the goals I said I wanted to achieve. It gave me all of today’s anchor, any carryovers from yesterday, today what I’m I’m primarily working on and then give me the full plan for today breaking down every kind of minute of the day that I need to do.

Now I can take this one step further and say, “Okay, that’s great. Go ahead and add all the those things to my calendar.”

live demo: ai-generated workouts

But I don’t want to get too complicated from here. So that is how you tackle goals. So another example would be fitness. Hey, can you go ahead and look inside my fitness folder and pull my workout for today so I can take a look at it? And again, now we’re over here in this fitness section, right? It’s pulling the workouts that I have for today. It’s referencing an index file of my training profile, who I am, the equipment that I have, and then the programming methodology I’ve spent years and years creating on different energy systems that you should be hitting, different movement competency streams.

Again, complicated for most people, but for me, this is exactly what I want as someone who has been a strength and conditioning coach for a long time. And so it’s going through all of these things and it’s going to be looking at past workouts. It’s going to be looking at my workout today. And then when I come back here, it’s going to tell me this is my workout, right? And I can see that’s right. That was the workout that it had for me today.

But then I can even go one step further, and this is where it gets smart. It’s not just a file retrieving program. I could say, “Great. I already completed that workout this morning. Based off of everything else I’ve done in the index file, can you go ahead and create the next best workout for tomorrow?” And now it’s going to start referencing all of these rules that I have. It’s going to start looking at what I did yesterday, what I haven’t done, all of these things, and then it will start to populate a new workout.

Now, before while it’s working on that, I’m going to show you one last thing that you could use it for. Again, there’s so many examples. I’m going to go crazy with this as I develop more and more. But here is my my podcast. And what it has here is a summary of every single podcast, the tags that make sense based off of kind of the themes of the podcast. It does all this automatically, which I’ll have to show you in a separate video.

Gives a summary, the key takeaways, the full transcript. And then why everyone really loves Obsidian is this graph view right here. It’s not super helpful all the time. A lot of people act like it is, but ultimately it really isn’t all that helpful, but it can be if you’re looking for themes and ideas and connections. So this is my entire podcast, the index file that connects to every single folder. But then I can start to look up, “Okay, which episodes did I talk about leadership?” And it pulled all of these for me, tagged them automatically.

Okay, business operation, marketing.

live demo: podcast knowledge base

When did I talk about family, personal development? And so now I can start to see all these connections, not even as a tag, but also in between the connections. Then you can see this is a brand new vault for me. I’ve been using Obsidian for a long time, but it’s starting to build branches over here. Uh you know, these are probably, let’s see, the fitness ones, the goals ones. They’re all starting to connect, and I’m going to start to interweave all of these files so you can see more and more.

Okay, so now it went through everything. It talked about what I’ve done, the strength work that I’ve done, and it’s going to say, “Hey, tomorrow’s call, we’re going to be doing aerobic endurance.” And then it talked about why. And again, the most important part of any of this is not just It’s not a file retrieval program. It’s just not it’s not AI randomly creating goals out of thin air or rate randomly creating workouts. It all depends on the index file.

Now, before we start talking about how to build an index file, you can download a lot of the examples I’m going over in this video inside of the school community. So, if you want to learn more about the 1330, you can go to better strategy right here, get the 1330 template. You can come back here, go into time systems. This is the video I’m working on right now. You can get the index for goals, so you can add these things to your Obsidian vault and get to work.

You’ll see that the index files are probably the most important thing, right? These like I said, this one’s 4,302 words. And it gives the entire, “Hey, here’s exactly how you work through this.” It’s almost like a prompt, but you never have to repeat yourself because we know AI has a limited memory, and the longer you talk to it, the worse it gets. That is a proven problem with every AI platform right now. So, when you go through something like this, you have to create it.

And you can use AI to help you create an index file, but I’m going to go over kind of the framework of how you want to think about these things. So, index, I worked really hard on making sure that this was a nice acronym for you. Index, I N D E X. Let’s go over each one when you’re building this file, what the I needs. So, you want to write the constitution once. So, your master prompt, it tells AI, any AI system, how it works, the fields that are needed, formats, tags, templates, what’s allowed, how it you hand it to Claude or ChatGPT, and then it’s up to speed in 30 seconds.

It’s very easy for her to do.

the index file (the secret sauce)

So, that’s the first thing that you want in there is the repeatable prompt, right? So, I want you when I say create a new workout file, here’s what you should do. Reference these rules. Look at what I did yesterday, so on and so forth. So, you’re going to write the instructions. That’s the most important part of the index file.

index framework: i: instructions

Then you get to the N, which are notes.

n: notes

So, you want to have every workout, every meeting, every daily reflection, same structure every time. So, you open the daily note, the template’s there, fill in the blanks, and then you move on. So, So, I have the daily note, my workout, it’s going to be the same structure every single time. Then I can provide notes at the bottom of that workout. So, Claude or ChatGPT will be able to reference those notes and build on it in the future.

Next, we have the database, and this is really just a table of contents. So, all of my index files, after the prompt, all the rules, everything that it has, it has a long table of contents. So, if I show you one like the better podcast, if we go all the way down to the index file, you’ll see this one starts with a table of contents. So, it has a link to every single file that we have, right?

d: database

Everything that I’ve done. And this is really important, so it’s able to quickly go through these and, you know, find what’s important, what it might need to reference. If it needs to click on a previous episode, in this case for podcast or content prep, it can do those things and pull those files very easily. So, you want the table of contents, whether that’s for workouts, for goals. For goals, it’d be like a running daily log of everything that you worked on towards your goals.

For podcast, it’d be every podcast episode. For workouts, it’d be every workout that you’ve done.

e: engine

And it just runs this table of contents, and it just links to the notes. That’s ultimately all it’s doing. Pretty simple there. Then you go from database to the engine. So, you point Claude at the folder, which I’ve kind of shown you already, but I’ll show you inside of Claude. If I go to new task, that’s happening right here. So, you can click on here, and all that is an Obsidian vault that I’m working on. I’m working on the side of a folder.

So, don’t overcomplicate Obsidian. So, you can go to Obsidian at any point in time, you download it, it looks like this right here. I’m actually using the browser inside of Obsidian, but you can go to here, manage vaults, and then I can just create a new vault. It’s basically just a new folder. Don’t overcomplicate these things. You see, I have two. This one I created for the example I’m doing today. This is my one that I’ve been using for years.

It’s massive. I was a an user way before AI ever, you know, existed. I love Obsidian. I have a lot of information in there. So, you would create a new vault, and you can say where you want that to be. Browse a location on your desktop or downloads folder or documents wherever you want it and then you just point it there. So if you were into be being a chat GPT or if you’re in Claude, you say, “Hey, you connect it right here.” And then Codex kind of showed you that already.

You just add a folder and then boom, you’re good to go. Now you’re working within that framework already. So after the engine is set up, the X is for exportable. So I had to stretch a little bit here and pull the second letter in exportable, but this means that you own it. These are markdown files. So markdown files are the most simple files that exist on a computer and they can go anywhere. That is the best part of all of this.

Claude, Codex, they all work in markdown files, which is a beautiful thing and so does Obsidian. That’s why Obsidian is so easy to connect. Obsidian is not an Evernote. It’s not, you know, the Bear writing platform where you lease Ulysses or anything like that.

x: exportable (you own it)

What it does is it just keeps a very simple file structure so you can reference these files, you can create markdown files are very small and very easy. They’re just text files and so you can take this anywhere. So that’s my favorite part about this is I can easily work inside Claude right now, but if Claude starts to suck or maybe the company fails, Anthropic doesn’t do well, which doesn’t seem to be happening at all right now, I can move to Open AI or if other way around, I can move.

Maybe there’s a brand new one that comes out. All of my things are in here. I don’t have this big migration process if things change and that’s my favorite part of being able to have my own vault, my own brain inside of Obsidian. So the principle underneath all of this is always bet on your own systems cuz we know apps come and go, models come and go, the hot tool from 2022 is already dead and the hot tool from this morning will be dead by Christmas, I promise you.

The only thing that survives is plain text sitting on your hard drive running on a structure you built, your frameworks, your system. And the CEO of Obsidian has a phrase for this and he calls it file over apps cuz apps are temporary, but files are forever. And so if you want to push it one step further, systems over apps. That’s the ceiling.

files over apps, systems over apps

Don’t just own your notes. Own the living, breathing operating procedures that run your life. That’s the part that compounds. Now, if you want to start really small with this, you don’t want a 4,000 word index file, you don’t have to. But, you do want to start a simple index file today, and you can go with the smallest possible version. So, step one, download Obsidian and create a note. The note can just say monthly goals. Write your goals in plain English.

My monthly goal is to make $50,000. Connect that file to Claude.

start small: your first index in 30 days

Tell it, “Each day, ask me my daily three tasks and make sure each one ladders up to my monthly goal.” That’s day one. The eye of index is in motion. Once that loop is running, you add the rest. Templates, a database, the full constitution, one letter at a time. You have the whole system inside 30 days. But, right now, all you need to work on is getting something really simple set up with Obsidian and AI of your choice.

Now, this works great overall as a time saver, but if you want the best first system to start saving you time, you’re going to want to learn how to make Claude your executive admin assistant. And I did a full tutorial on that already, so you can click the video on the screen, and I will also link to it in the description, so you can keep building cool stuff.