# Obsidian and Claude Code: The New Personal OS

**Podcast:** The Startup Ideas Podcast
**Published:** 2026-02-23

## Transcript

This is Obsidian.
And Obsidian is this little tool that people are using as their second brain.
But what's really cool about it is they're pairing it with Claude Code and they're getting crazy results out of it.
It's literally a game changer.
Now I've been slow to adopt obsidian because to me it's been a little daunting to look at.
So I had my friend Vin, and he clearly explains what Obsidian is, how to use it with Claude Code, how to set up these commands that really drive the most out of Claude and all the LLMs.
And it's an incredible episode, like a really game-changing episode, because I think that people who understand how to use obsidian and how to use Claude Code together, they're going to be able to live happier, healthier, and wealthier lives.
Why?
Because it gives incredible ideas to you on tap.
So I know that the people that stick around to the end of this episode, I think that for a lot of them, it's going to absolutely change how they use AI, and it's gonna be a super impactful way because you're gonna get better ideas at the right time, the right moment, and it's gonna make you happier, healthier, and wealthier.
Enjoy the episode.
The Startup Bag is podcast.
It's I've got my dear friend Vin, also known as Internet Vin on the podcast.
I literally begged him to come on.
I begged him.
I begged this man to come on and to teach us a very specific thing.
Vin, by the end of this podcast episode, what are people gonna learn?
I want you to have an understanding of how you can use Claude Code and Obsidian as a thinking partner.
I want you to have an understanding of how you can stop uh having to explain things to agents over and over again and just pass specific files in.
And I want you to understand how you can use obsidian and claude code to notice things about the way you think that you would not have noticed on your own without these tools.
All right, from your lips to God's ears.
Let's get into it.
Okay, so first is like what is Cloud Code.
So Cloud Code is this like agent that you can use um in a command line interface.
So it's just basically this tool you can use that you can control your computer and you can use it through natural language, right?
So I can say make a file or make a file on my desktop that says hello Greg in plain text, right?
And it's gonna go and do this.
That's really cool.
That's that's something that's that's new that wasn't possible before.
Before this, I had to go to the desktop, open some text editor, and then like create that file, right?
And now this file is on my desktop.
So I can say open the file.
There you go.
Hello, Greg.
That's crazy, right?
Um, what's interesting about this is if you have this agent that can like control and do things on your computer, that means that whatever you can describe to it, like it can start to do.
And so when you if you describe a project to it or you get into these long conversations uh with an agent, um, it can do more and more complex things.
The more information it has, the more complex things it can do.
But the problem is that if you have to, let's say, you know, like I write some super long description about a particular project, or I have like an hour conversation with this agent about a particular project.
It's like I don't want to have to create a new session to explain that all.
I don't want to have to explain that over and over and over again.
Um a lot of people are using like cloud uh or chat GPT on the web and it has things like memory, but you you can't like control, you don't know what's in that memory, right?
You don't know what it knows and what it doesn't know.
And so there needs to be some way of like you know, passing information into these agents that is easier and faster.
And the better information you can give it, and the faster the information you can give it, the more stuff if it can do for you, and the better uh the faster you can delegate to it.
Okay.
So now even if I let's say let's say I had like, you know, let's say I wrote like a big project description here, right?
Create a file that describes you know a project.
Um about uh to-do list app that is very minimally designed and reads from uh all of my calendar MI messages and um my sla and Slack and interprets it into a task list that of tasks that it thinks that I should do.
I don't know, some some idea, right?
So now this this is a file that could be on my desktop, and what I can do when I use cloud code is I can reference that file and pass it in whenever I want.
And why that's important is because it's the context, right?
Yes, the whole game is feeding the beast good context.
Yes, exactly.
And I don't want to have to do this over and over again.
And and and when I work on this uh over days, I I'm not gonna remember like what we talked about, right?
So I want some kind of file uh that I can like pass in.
Oh, sorry, Greg, one sec.
Yeah, and that's that's sort of the problem that a lot of people are facing with cloud code, is like they're using it and then they're saying, well, it's it's okay.
It's not like game changing.
And the issue is they don't have they're not they're not feeding the right context at the right time.
Yes, exactly.
And so here's so here's like a project description that it wrote, and obviously I can pass this in.
And this is like a this is like a general one that I just created, but you you can make these like very complex.
You can build them into like robust files, right?
Over time.
So we know that uh cloud code can create files and it can repeat and it can um uh uh read files, right?
So now I can say, let's say I created a new session.
So here's a new session, and now I can say, like, I want to work on this project.
And I go here, and it's gonna be um to do dash app.
Here it is.
Boom.
Now I didn't need to explain the file again.
I need I didn't need to explain the project again, right?
So it's gonna read this file and it's gonna start like you know, that saved me a lot of time.
Great project before diving in a few questions to scope the first session.
So that'll continue.
So now, what is obsidian?
Right?
Obsidian is this tool that it's kind of like an interface that sits on top of um uh a collection of markdown files, right?
So here, like this is reading a markdown file, how I use obsidian story development, right?
Um I have uh daily notes.
This is my daily notes.
This is a this is also um a markdown file.
I should do my own fundamental analysis into thing into how things stay pure when they grow and become more mainstream, right?
This is just a file that I have.
I have um, you know, like a file on Greg Eisenberg that I haven't put in.
Oh, that's we that's weird.
Yeah, that's pretty weird.
And so I I make files like notes of things that I'm learning from people and stuff like that too.
So I have different um files for everything, right?
And the interesting thing that makes a vault, which is obsidian, what obsidian interacts with, this whole thing is called a vault.
What makes it different than a folder is that obsidian is not only interacting with just like a you know a folder of files, um, but what it does is it also allows you to make inter uh to connect relationships between files.
So I can say today I am on um a podcast with Greg Eisenberg.
Now this file is linked to that Greg Eisenberg file.
Super interesting, super interesting.
And so when people like like like people, uh there's a lot of people who really, really like using obsidian and tools like obsidian because of this ability to form interrelationships.
This is unique to just having a folder.
A folder on your computer cannot show these interrelationships.
And so it gets really interesting when you start to um keep making these like interrelationships over time, right?
And so what happens is here's a little visualization.
And so in that here, these are each one of these circles is a file.
And in and it's it's showing how it's like connected to all of these other files where I've written things about.
So here's like personal agent infrastructure, right?
And so I could look, you know, and I guess I should also add just to kind of a comment on this and what was difficult about doing this demo.
There is like so much personal information in here.
Cause this is like my personal thing.
So I don't even know like what's gonna show up on the screen here, right?
Um, but that's part of doing demos like this, which is which are kind of weird and interesting.
But you can see personal agent infrastructure, it links to like agentic AI.
There's like a link here to Telegram, there's a link here to like Toby, uh with the founder of Shopify.
There's a link to like presence log, ClaudeBot, you know, and then here's like I I have a podcast too called the other stuff, and like you'd see I'm obviously doing a lot of like thinking about that a lot, right?
And so I can also let's say if I go to Greg Eisenberg and I go to local graph.
So here's like all the times I've written about Greg Eisenberg, right?
Notes on time constraints, how I use obsidian, which is just kind of interesting.
Um, so if I'm listening to I love you too, Vin.
I love you too.
If I'm listening to a show and I'm picking up different patterns, I can I can reference that back to to Greg.
So that's really interesting.
But here's the thing.
The reason why people love obsidian is because of these interrelationships.
The idea that you could open a file and then you know, I just open this file, and then I'm like, oh, interesting.
I mentioned Greg Eisenberg.
I can click that and it goes to that file.
That's interesting, right?
It shows it works more to where uh it works more like the way your brain works.
Your brain connects these patterns all the time.
Here's yeah, yeah.
So I see why it's interesting.
But how does this get me better output?
Exactly, yeah.
So the next thing is Obsidian released this new tool called Obsidian CLI.
And what that allows you to do is it allows you to use Claude code and it can go and it can read all of the files in your obsidian vault, which is a folder of text files.
But with the obsidian CLI, it can give Cloud Code not only those files that it can read and access, but it can also give Cloud Code information about the interrelationships of those files.
So you can see, so Cloud Code can see that, oh, this file is connected to this file and this file and this file.
And that gets very interesting in terms of what Cloud Code can understand about you and what Cloud Code can uh understand about uh all of the relationships between the things that you're working on.
It can start to surface patterns about what you're thinking about that you are not seeing for yourself.
Some idea that you might have been writing about for a year in this vault, it could be a latent idea and it can just immediately say, like, hey, did you know that you've been writing about this same pattern in startups or in this particular project you're working on in every single note you're making across these different domains?
And and seeing that for the first time can be like a huge light bulb effect.
It can cause like huge progressions in your learning and your understanding and your point of view on the world, but also in what you're working on.
Um, so I've written out, I wanted to demonstrate how that actually works in terms of how I can pass information into an agent that would be impossible without obsidian and cloud code.
So here's some commands that I have that I use.
And I don't want you to be afraid of like all this stuff.
I know this can look intense, but um here's here's what um here's what I've got some commands.
And this is just terminal that I've created, and I'm running it in obsidian.
You don't need to use this.
You can just also do this in your own terminal session on in whatever tool you want.
But I put it in obsidian because I want to see it all together, and I wanted to show you the ways in which you can like integrate and customize this environment.
So here's a cool thing.
So context slash context.
Load full context about my life, work, and current state.
Reads context files, daily notes, and follows backlinks to build a complete picture.
So I'll just show you that right here.
So like let's say I open a new session in in uh in Cloud, just on my desktop, and now it's like I'm about to work on something.
But before I work on it, I can just type context demo.
Now it's gonna read a whole bunch of files about where I'm currently at, done.
Like I've already preloaded, I've already preloaded in all this context now.
So you can see it's gonna start reading all these files.
It's reading a readme, it's reading context about new, which is a media company that I'm working on.
It's reading about other stuff, it's reading my personal the other stuff is my show.
It's reading a personal workflow context.
And so I don't have to worry about it not knowing the key information that I want it to know.
I just did that one command and now it's gonna get all that information done.
So I can use slash today, which is a morning review.
It pulls calendar tasks, eye messages in the past week of daily notes in a pr into a prioritized plan for the day.
Why does this matter?
Well, okay, sure.
You can set up an agent and give it access to your calendar and your tasks and iMessages and things like that, but it's miss, it's that's that doesn't have all of the information about what you're thinking about and why.
If I'm writing daily notes about some particular technology or project or thing that I'm interested in, does my calendar reflect actively like does it match the subjects I'm actually writing about?
If an agent has that context, you can it can it can uh more effectively give you um information about what you should do or not do, or and it can more effectively make decisions on what should be in your calendar or not in your calendar.
Here's another one slash close day.
End of day processing, extracts action items, surfaces, vault connections, checks, confidence markers, needs to be updating.
So I have a bunch of hypotheses that I I think about and I give them a confidence rating.
This is an idea I'm working on, I feel very solid about it.
Here's another idea I'm working on, I'm not sure about it.
So these are like um daily operations things.
But this is what I use obsidian for the most, which is thinking tools.
I really, really, really like working with LLMs as a thinking partner.
That's my favorite way of using LLMs.
I know people like to use uh agents and LLMs to build things, but I really like using them to think alongside me and build when I feel like you know, I really have a novel way of viewing things.
So let's see here.
Um so ghost, here's here's a command I have.
It answers a question the way I would.
It builds a voice voice profile from the vault, writes in that voice, then evaluates the fidelity.
So I can just say, what do I think of AI?
And I'm going to show you this.
Challenge topic.
It pressure tests current beliefs using the vault's own history, finds contradictions, counter evidence, and shifts in thinking.
Why does that matter?
Well, if I want to make sure that I'm continually developing as a human being and as a as a as in my skills, I want to make sure that you know the POV I have isn't overly biased or limited.
So this can challenge me.
Emerge.
Surface ideas, the vault implies, but never states.
Conclusions from scattered premises, unnamed patterns, unarticulated directions.
This is super, super useful because a lot of times, you know, I can be stuck just surfacing ideas in a in in um in a lot of different ways, like for years.
And just having someone say a simple thing to me that just says, hey, this is just naming the idea.
Hey, did you know that you keep circling around this pattern?
Huge breakthroughs.
Uh slash drift.
It compares it compares my stated attention intentions against actual behavior over 30 to 60 days.
Surfaces what I am avoiding.
Ideas, people on this podcast, the listeners will probably like this one.
Deep 30-day vault scan with cross-domain pattern detection and graph analysis to generate generate ideas across all domains.
This gives me not just ideas on like things I should work on, like it gives me ideas for tools and things like this, but it also gives me ideas on like films I should watch, products I should buy, again, all influenced by like like things I'm writing about in my vault.
Um trace tracks how an idea has evolved over time across the vault.
So let's see some of this stuff.
The trace demo.
So I did this one already.
And the way this would work is I just like create a tab here, and I could just be like, Claude Trace.
And I had to create demo versions of all of these um commands because of how much personal information is in my vault, but still I don't even know, like I can't even control what is gonna show up on the screen.
And I I have a just a dumb question.
Like all those commands that we saw, is that commands that you created, or is that what obsidian created?
That those are commands that I created.
And um you can create them very easily by just asking Cloud Code to create a specific command.
Um we can include in the show notes and in the description, just like a link where people can click to get access to some of Vin's uh skills, if that's if that's yeah, for sure.
Totally.
Yeah.
So you could see here what I did was I just typed slash trace demo.
Demo is just because I'm showing this publicly, how I use obsidian.
And so what this does is it tracks my I uh the way I've developed and um the way I've navigated this idea through my vault.
So I have an example here.
It's really interesting.
So here I I did it and I and I and I let it run.
And what it's doing is as it says like I'll trace how your relationship with obsidian has evolved.
Let me start by building a vocabulary map and searching across a vault.
So it when you have obscene so now it starts going through the vault, reading all of these different files, and it can see all of the files that are connected using the obsidian CLI.
This is a lot, man.
This is this is something that I would never be able to do on my own.
And to read all of these files, to know what how they're interconnected with each other.
This is just not possible for me to do as a human being.
And then it spits this out.
I now have everything I need to construct this trace.
Here's the full evolution.
Trace, how I use obsidian.
First appeared January 11th, 2025.
Time span 13 months.
That's in relation to this vault vault specifically, meaning all the files in here.
It says pre-vaults, December 2024.
The essay, How I Take Notes in the West End of Toronto, dated December 1st, 2024, describes a complete system where obsidian plays no role at all.
The system is audio dumps via Mac Whisper, LLM dialogue loops, Canopia for spatial mapping, physical notebooks for cognitive friction, future land for tracking, arena for fragments.
The pipeline is capture process crystallization.
This is the baseline.
Obsidian is not in the picture.
Discovery and set skepticism, January to May 2025.
The first daily note in the vault, raw excitement mixed with uncertainty.
I could probably drop transcriptions in here too as a way of storing them.
The current theory is that in terms of note taking, bi-directional linking is not that useful, but I don't know.
Then it says the chosen tool.
Um originally I was back ringling, it says um here, how I use obsidian note, a pivotal realization about backlinking.
Originally, I was backlinking to general terms, podcasts or physical fitness or filmmaking.
I'm realizing it's not the most useful way to use obsidian.
The most important thing to do is to create notes for each of my patterns, theories, projects, or perspectives, and get them documented and out of my head and then link to those notes.
So it just like pulling from things I've written, and it's forming this historical history of this concept.
And I could do this with anything.
I can do this with like startups, I can do this with a particular project, with foods, my relationships, like a hobby, anything, right?
And then phase four, January 2026, a month of explosive building.
Everything still requires, and then like everything still requires me to actively prompt and manage each section session.
The next unlocked is figuring out how to get agents task to run automatically.
The friction is no longer with obsidian itself, but with the boundary between the vault and agent execution.
So you can see I'm really pushing myself, right?
And it's cool.
This is a this is a very useful thing for me to understand how my um use of this tool is evolving.
And it's just, I think it's just absurd that I can just be making notes and then about all of these different things through my life, like even as a parent, I I can I can like reflect on the different things I'm learning.
I just think this is insane that a computer can have this much information about me and surface these patterns.
I would not be able to do this on my own and this fast.
And what a great tool it is for me.
Like I can just like now write in here, and I'm, you know, as I'm thinking about things, and it gives me ideas, right?
About my life and the projects that I'm working on.
So I can say, you know, it's interesting the way that my relationship with obsidian has evolved over time.
It makes me think a lot about the way in which my relationship with computers has evolved over time.
Since I was a kid to now.
It's interesting how these things just happen and compound over time.
And we don't really realize it.
So it's just like a note, right?
It's an idea.
So that's an example of something.
So I think a part of getting good at obsidian sounds like reflect, you know, inserting, reflection into your everyday life.
Because a lot of people, you know, we're moving from meeting to meeting, we're busy, we're parents.
Um, you know, we grow up and we of course write things down in notebooks and stuff like that.
But I feel like as we get older, we actually write and reflect less and less.
Yes.
Uh, you know, how do you how have you been able to insert reflection into your life?
Yeah, I think um for me, it's really about there's there's two reasons that I think reflection is interesting, and making notes a lot are in is interesting.
One is that it's great to be able to look back on them.
Like for me, now obviously I can use an agent, but to for me to go back and see these notes and realizing that oh, like I'm a person that's continually changing, my skill is continually changing, projects are continually evolving.
It's it's just a mate, it's an amazing part of life to be able to reflect on how things are changing over time and how you are changing over time and how the world is changing over time.
But the other thing is that there's like a functional reason too.
A reason I like to make notes a lot is because that's how I generate ideas.
When I like get when I sit down on my computer and I write things down, that's where ideas come.
For example, this thing I just wrote here, it's just a quick note.
I'm just doing it in real time, I'm just making it up right now.
But um, by writing it out, I feel like I internalize it more.
And I like having good ideas.
I like progressing.
So because I like having good ideas and I like progressing, writing is how I do that.
And so I think to, you know, if you want to cultivate like a writing as a habit, I think first you have to uh connect it with the idea that this is how you progress and this is how you generate ideas, and this is how you uh have form original ways of thinking.
Um the other thing I'll say is that writing right now is a big way of how you delegate things to agents.
That's like a whole new that's a whole new aspect of it.
So if you can develop a writing habit, you have a lot more context that you can pass over to an agent, which then um dramatically I think increases the amount of like things you can delegate and the amount of things you can build.
I hope that was a good answer.
How does how does this relate to open claw?
Because if you think of open claw, it's essentially, you know, at the best case, an extension of you that could go and do things, you know, uh independently slash based on your guidance.
So, how can you use commands, obsidian and open claw and reflection harmoniously?
Yeah.
So I think if you look at like here's an example of one command that I do, it's just like a schedule command.
Um, and so what I asked this thing to do is I said, schedule, I said, can I take a meeting with Greg Eisenberg today at uh Feb 20 at 2 p.m.?
Right.
And what this does is of course it can look at my calendar and stuff like that, but it's also going to look through my daily notes, it's gonna look through what I care about, and then it's gonna give me some perspective.
So it says, your day is stacked, you're already recording on Greg's podcast this morning, followed by a team lunch outing and meeting with Peter and Vince.
Your Feb 17 notes show the Greg episode has been top of mind.
The vault has a dedicated Greg Eisenberg note.
No, not a two.
So the recommendation is no, not a two pm, but you might not need a separate meeting at all.
Yeah, that's val that's actually the correct answer.
How does that relate to open claw?
Well, open claw is this like autonomous agent that can go and do things if you uh set it up to do that, like it can do things without you having to prompt it all the time.
It can just go and make decisions and build things for you on your own.
So now what OpenClaw can do is in the same way I just did this command, open claw can do this on its own as well.
And it can go and read my vault, find connections, and then make decisions uh on behalf of me with like a deeper understanding of me.
And now instead of like managing an agent or talking to another human about working on something, I just focus on managing this vault.
This is like the new source.
I just continually try and make it so that this vault has all of the information needed so that I can delegate to an agent, and um the agent can just pull from this vault source and make decisions, and if it's not making the right decisions, I'm changing something on the vault.
I'm not necessarily working with the agent specifically.
Um kind of um that's my speculation on that subject.
I think it's very interesting.
Yeah, I think uh one thing that worries me a little bit about it is if obsidian is really your second brain, giving open claw access to your second brain is scary.
Yeah, scary.
And I would say that is the um fundamental um the the weird element of this technology, I would say.
And I have purposely given um obsidian, I mean, sorry, the Claude Code or any agent access to a lot of information.
I've purposely done that because my um relationship with this is I want to understand what these things are, and I want to understand what they're revealing about, you know, how our relationship with computers is changing.
But um, it's weird.
It it's like you have to really think about how much information you're sharing with these agents and whether that's the right decision or not the right decision.
And I think um it's gonna be very interesting to see how privacy as a concept evolves and changes and what we fight for or don't fight for, and like the in the future of our society and our world.
Even with every one of these commands, I had to create a new version of them, a demo version, so that it wouldn't reveal too much personal information while I'm on screen on this podcast.
And even then it's like a it's a toss-up.
You know, I could type the demo version, but who knows what's going to be shown on screen, you know.
What other commands do you want to show?
Um there's connect, which is allowed it allows me to take two domains and connect um them using the Vault's link graph.
So I can just say I did one here, and I just asked it to uh connect filmmaking and world building, and so it goes through and it reads all these different files, and then it can start to say, okay, let's let's connect these two concepts.
So notes and filmmaking, uh notes and filmmaking's neighborhood.
So it's like 35 film watch lists, my first meeting with Toby, notes in the world building neighborhood, um the world building essay, um, new as a media company.
So these are different things I'm thinking about.
So bridge one, the interview portal and the constructed world in filmmaking.
If I notice something specific and asked a question about it, it would open a portal into a person's internal world, which is often a vast universe of concepts and beliefs and visions.
So world building essay.
I want my blog to show you what I value, what I believe, what I worry about, like a tomb from ancient Egypt.
I want my blog to be a place that you dig up and examine long after I'm gone.
These are things that I've written, and I can start to see how these ideas connect together.
Bridge to always on documentary equals continuous world building.
Always on documentary is a creative strategy where companies continuously narrativize their characters, pursuits, conflicts, and visions through documentary.
So these are like things that I'm writing about, and it's showing me the ways in which these are connected.
Um I think this can get very interesting depending on the kinds of things that you're willing to connect together.
You could get probably pretty crazy with it as well, depending on what you're writing about in your vault.
I could connect like shawarma and startups if I wanted to, for example, and see the kind of connections that are coming between these things.
Um again, really interesting because all of this is happening super quickly, and I don't need to explain any of this to an LLM.
I can just type something like slash connect filmmaking world building.
The the ex a lot of the examples you're using is personal reflection.
Yes.
How do you think about uh, you know, for example, note taking in meetings?
Like maybe you have like uh granola or Gemini notes taking, you know, uh taking notes or and and and sort of putting it into obsidian, or um and then by the way, when I say notes, then those could be meetings that you're not even in.
They could just be like, you know, Tommy met with Vince and they had this meeting and I want to put it in here.
Yes.
Um, how do you think about that?
So it's a really good question.
So I think you can use these vaults however you want to use them in terms of like um you could put any text you want in here.
If you want to put your granola meeting transcripts in here, you can put them in here, and you have to just make sure, you know, maybe you're doing something like this, right?
So you're just like meetings, and then you're like, okay, these are this is you know, project one.
And then every time you do a meeting, um, you take your granola notes and you just put them in here, right?
So you're like meeting Greg Eisenberg, Vin plus Vin, and then you know that file is created now, and you can just drop your trend, you can just drop your meeting notes in here, and now that's in the vault, and then you can pass that into the agent or the agent will discover it, right?
Especially if you start tagging, like, oh, I'm gonna tag this back to, you know, like my podcast or something.
There, now it's connected.
So now the the the agent it has like more context, and now it knows that's trend this transcript is related to this other file.
Great.
Um, I think that's up to you.
I think the way that you the amount of information that you put in here is up to how you want to use the vault and how you want to delegate to things to agents, and maybe you even want to create different vaults for different purposes.
For me, per for me, I use LLMs and agents as a way to increase my own level of understanding of subjects.
So I use it for a lot of reflection and things like that.
So I don't want an agent to write into the files.
Like I could easily get it to do that.
Like I could just say, like, even here, I've I've asked it to write a description of some commands that I can talk about today.
But I don't I don't I don't want it to make a file to do this because I want to control all the files in my obsidian vault.
Because I I always want it to pull from what I think about things, right?
Not what it thinks about things.
And if it starts making its own files in this vault, then I don't know, like is like when it's finding these patterns, is it finding patterns about things it's written or is it finding patterns about things I've written?
So I create a rule for myself which is like a strict separation between these things.
I only want it to write things on the side here, and then I will take that and and and and write what I think um should be included.
Right?
Some yeah, go ahead.
No, I was just gonna say, like, you know, I could see the power of just using it for your own reflections.
I can also see the power of you know, AI is really good at you know on the internet, finding information based on trends and stuff like that, distilling it in a way that you want, and having that being put into your world is also interesting.
Yes.
Totally.
I think that, and also like, let's say if you asked um the obsidian, if you asked the um Cloud Code to go through your obsidian file and and generate ideas, which um, you know, ideas for tools that you should build.
Well, then you can just like say, okay, cool.
If that's if I have an idea for a tool that I should build, just generate a description of that and then just build the tool.
So exactly.
Yeah.
I want to show this one so that it's like less on reflection.
Like, for example, like, you know, I'm uh I I built this thing called idea browser.com, and every single day we give this validated startup idea.
Like someone theoretically can go and you know, grab that information, put it in an obsidian vault, and then based on that, how basically, you know, help them build the actual thing, right?
Totally.
Yes.
So I want to show you this because I think it'll it'll really make it'll take it out of the realm of reflection and into the realm of building.
But the only issue is this takes a bit.
So that's what it's yeah.
So the other thing is that with all of these, like with these commands, another pattern that I'm noticing is that they take a bit because it's reading so many files.
And I would say that's a big difference between using obsidian and um like or using cloud code with access to this obsidian vault that I'm noticing is all of my requests are taking way longer.
And it's just because it's reading so much more.
So like look at this one.
So this is ideas demo.
So I'll run a comprehensive ideas generation.
Let me start by gathering vault structure and context in parallel.
And then if you look, it's really interesting to see what it's doing.
Right.
So it's like obsidian orphans, right?
So it's it's it's it's like I guess orphans are like files that are on their own, not connected to things, right?
So that's interesting that it knows that.
Obsidian dead ends, obsidian resolved, obsidian tag counts.
So it's just trying to it's trying to figure out like some connection between all these things.
Um, and then it says, okay, daily read.
So it's reading my daily notes.
Then it found this file called New Context, which is new as this media company I'm working on.
Then it's like read file, the other stuff context that that's podcasts.
The other thing I would say to you guys is I do manage um, I write uh I create context files for projects that are pretty extensive.
I'll show you.
I I was I didn't know if I was gonna show this because it's very personal.
But like for the other stuff, look at this.
So other stuff working context, what shifted recently, front loading profile by traveling to San Francisco, New York City to record guests.
This is super personal stuff.
But what is the other stuff?
The format, core beliefs of the show, research is the foundation, solid.
The best conversations feel like discovery, solid.
You know, here's the team that's working on it.
And so what happens is this again, very personal, but this is um context that it just pulled in.
So now it knows who's working on my podcast.
What are the what are the recent hypotheses, hypotheses I'm exploring.
Um, and it just got that information.
That's just like one of the things it did.
Personal workflow context, super personal file, but it shows like um, you know, like uh what my daily schedules is like, you know, things that I have to do, like in my personal life.
So it's like pulling that how I like to work, how I don't like to work, personal agent infrastructure is another thing.
Let's see what happens if I pull this up.
This is a project in which I want to take a step towards increasing my personal infrastructure workflow, delegation, however you want to describe it with agents, understanding what it means to delegate to agents more and more.
Implementation approach.
So this is like what I'm writing about the file about how I'm thinking about using agents personally.
And again, that's an example of one of the files it's reading.
That's just one of them.
So you saw like the other stuff, the personal workflow, and it's factoring that all in to this task I've asked it to do, which is generate ideas for me, gathering data from your daily notes, calendar, and vault structure.
This takes a moment since it's pulling from multiple sources.
Again, one of the um things with this is that it's just going through a lot of information, man.
A lot of information.
So it takes longer.
You know, it's already been going for five minutes.
Right.
And so that's that's something I'm noticing.
But for me, that's what I want.
I want that.
I want, I want a response from LLMs that is very, very contextual uh to the things that I'm writing about.
And I think a lot, and I think that's how me and an agent can work best together, where I just focus on continually um noting that my the the where I'm currently at in terms of the projects I'm working on and what my understanding is and what I find interesting.
I want to maintain that and make it as current and as deep as possible.
So whenever I'm talking to an agent, it has the best representation at all times of who I am in that moment when I ask the agent for something.
Yeah, that's the goal, right?
You that's the question that we all should be asking of ourselves, which is does the agent have the most up-to-date information on the projects, on my preferences and my dreams and my hopes and my goals.
Um it's only as good as the up-to-date version of that, correct?
Yes, 100%.
The quality, the the quality of information that the agent has entirely determines what it can do for you.
Right.
If it doesn't know a lot about you, it's not going to be able to do a lot for you.
But if it knows a lot, then it can it can it can do things for you that I think like even some of your in some gate, it's kind of weird to say it, but I mean like that you don't even know about yourself in ways.
I mean, it makes sense, right?
Because ultimately what this is doing, like to distill it to its core, it's connecting the dots.
Yes.
Obsidian and obsidian and Claude code here are are connecting the dots.
Now, it's actually quite difficult as a as a business owner or just as a personal in our personal lives to connect the dots.
Like why do people in a lot of ways go to you know, coaches, therapists?
Yes.
Um, you know, if you go to a therapist and you know, you have someone who is you're doing most of the talking, right?
Think about it.
You're doing a lot of the reflection, and the therapist and coach is sort of guiding you.
So that's that's what this is doing in a lot of ways.
And I'm not saying, by the way, that don't go to your therapist, you know.
Just but I'm but my point is it helps it helps you helps you uncover what are their dots and how you can connect them.
Yeah, absolutely.
And for me, yeah, it's just it's just really exciting.
Um, and yeah, it's just it's just a crazy time with computers.
So let's look at this.
This thing finished.
So this is an idea generation report, vault relationship exploration.
This is this is pretty extensive, right?
Like to get an idea report.
Like, I I think this is this is really gonna show how we can move you move from reflection to something actionable.
So structural highlights.
So again, this is just obsidian stuff, right?
Orphans worth noting.
There's some defense technology stuff here, just a theme that's uh growing in Canada, massive intellectual investment sitting in isolation, also orphan agentic software.
So orphan just means these are files I haven't really linked, random notes I just wrote once or something.
Umresolved links that reveal latent interests, uh hidden relationships.
Again, all reflection stuff.
Fine.
What's working?
Obsidian Cloud Code is a combined system, is working for me.
This is producing genuine breakthroughs in thinking and output.
Day per domain structure when enforced.
This is basically I started uh splitting my schedule where each day has a specific focus.
This is cool, and this is very true.
The Greg Eisenberg episode as a forcing function.
It's compressing months of thinking about obsidian and agents into a clear thesis with demos.
Very true.
It you know, coming on the show and doing this forced me to um synthesize everything I knew and present it.
But here's where we're going to the actionable stuff.
Tools to build.
The slash graduate slash command, daily note idea extractor.
Based on daily notes, are full of idea.
Sorry, daily notes are full of idea tags and interesting thinking that never gets developed.
The Vault has nine idea tags but hundreds of undiscovered insights.
Build a command that scans recent daily notes, identifies ideas, tagged or not, and prompts you to decide create a standalone note, add to an existing file, or dismiss.
This turns a daily note stream into a structured idea pipeline.
Obsidian Vault for New.
It says, I just have to manage and set up a central obsidian vault for new.
What that means is in the same way that I'm creating this vault, and it has all my ideas and my patterns and everything like that.
Um, why would I not create one for my team?
Where they like as a team, we can go and ask this vault questions and we can all uh contribute to it.
Here we go.
Tools to start using.
What is this?
Types for all external documents.
Uh interesting.
A time blocking act that enforces day, a time blocking app that enforces day per domain.
Meaning, since I'm trying to focus on one thing each day, one aspect of my life, it's saying, why not create a time blocking app that forces you to do that?
Interesting.
Systems to implement.
One sentence in obsidian, agent handles the rest.
This is literally the demo three version of the Greg Eisenper prep Eisenberg prep.
You're already imagining it.
The next step is to actually build it.
Start small, right?
Schedule a call with person about topic this week in a daily note and have Otis or CloudBot or OpenClaw pick it up and handle it.
So it's saying maybe you can delegate right from the note itself is how I'm interpreting that.
Super interesting.
Like maybe that's even like a new UX pattern.
I don't even know, right?
That you could build into like these different tools.
Umjects to investigate.
Christopher Alexander's pattern language applied to digital spaces.
Interesting.
Um, Black Mountain College as a model for the stadium.
Stadium is a physical space we have in Toronto.
Authorless media as a concept.
Uh, how Shenzhen's hardware ecosystem actually works.
Things to write and publish.
That'd be useful.
Context architecture essay, the computer as a place.
Software will become fashion.
What Toronto Theory actually is.
Editorial thinking zine.
Conversations to have.
This is interesting.
These are real people.
Aaron, stadium workshop host about becoming an anchor of technical programming.
This is a space we have in Toronto.
Tarun, again another person.
Program about making program the flagship series.
Steph Ango, Obsidian CEO, about the vault as a place.
Um this is like yeah, this is crazy.
It's it's suggesting people I should meet, right?
Top five high impact do now, build the graduate command or do a manual weekly idea review.
Um this is crazy, dude.
This is actually crazy.
Like, and the fact that it's in plain text and it's just not beautiful, it's not, there's no images.
It doesn't make it easy to read, but I kind of like it because it's it's like dressed down, you know what I mean?
Yeah, I mean, I like that aesthetic because I'm a nerd, but you could just say, you know, you could you could just say, obviously, you could just say, can you turn this into a beautiful readable HTML file that is on my desktop?
This is hard to read.
And it's gonna do that.
Right.
So I mean, if you don't that's says if you don't like that, just do it however you want to do it.
You know what I mean?
I I like it like this in this kind of like this aesthetic.
Um, but yeah, that's how you can move out of reflection.
And of course, you know, of course, we could also say, like, here, you know, if we don't, if we don't want to do this, uh, we could also say like it recommended that slash graduate command.
So I could just say build the slash graduate command.
Mm-hmm.
Right, which is interesting.
And is that how you started building a lot of your commands?
Like it it sort of it started suggesting it and you're like, just go build it.
Well, I started off actually like building them myself.
Like I'm like, oh, uh, trying to think about commands myself.
But then yeah, I said I started asking the agent, like, well, wait a second, what commands do you think would be interesting?
And and and just and this might be useful.
Another thing that I'd like to do is I like to move to like higher levels of abstraction when I'm using an an L LLM.
What I mean by that is I could say like, oh, make a command that um tells me what I should focus on each day.
And that's that's like um a command that I thought of.
But uh when I when I another thing you can do is you can step back and I can say, based on my obsidian vault and what you know about me, form an understanding of what you think my level of understanding, like where you think my skill level is in terms of of a person and the projects that I'm working on, and based on that, suggest the kinds of commands I should use that would take me from the level I'm at to a higher level, right?
Get like, you know, get it to suggest the commands for me instead of me suggesting the commands, and I can pick between them.
So look at this.
This is this is it was this was the agent's idea based on what it read in my vault, based on the notes I'm taking, right?
So let's see what this is.
Daily note idea extractor.
Ideas, insights, and original thinking accumulate in detail in daily notes, but rarely graduate into standalone notes where it can where they can compound through backlinks.
This command scans recent daily notes, surfaces the best candidates, and helps decide what to promote into like an idea or something, right?
Sure.
So this is how it's gonna work.
It scans all the recent daily notes, it cross-references with the existing vault, it presents candidates, uh it graduates selected ideas.
If uh if creating a new standalone note, create the note in the vault route, write the note as a mini essay or working document that captures the core claim or question, context from the daily note where it originated, connections to other vaults, notes as backlinks.
Now, all this stuff, like it captures the core claimer question.
You might be looking at this and think, like, okay, this is just the text that the agent generated, and it is, but also it hits differently for me because I know like I'm writing a lot about these things.
I know, I know like even like the mini essay thing, these are words that mean specific things to me, which is just so it's just so crazy.
It's very contextual.
I know what it's talking about because I spend a lot of time in this tool and I spend a lot of time writing.
So yeah, create it, and it's gonna create that command.
Right?
And it's like that's that's nuts because I'm just going about making notes, and I have this parallel agent that is looking at my notes and giving me ideas on how I can prove my workflow, improve my life, and then and then not only can it just suggest it, it could just build the thing and it's done.
And we have it right here, slash graduate.
I could just hit it and it's gonna run.
That's crazy.
If if I'm open AI or anthropic, I'm buying obsidian, right?
Because it's the missing link.
Yeah, it's nuts.
It's the missing link.
Why the fact that there are people like you've sold me on this, by the way.
I've I've downloaded obsidian.
I think it's a free tool, right?
Yes, it is, yeah.
It's open sourced.
I've downloaded it, but I have not created my vault because I wanted you, Vin to take me through.
I knew that this was gonna be great.
I knew that I would go through this.
This actually exceeded my expectations.
Like, this is the the fact, like it makes no sense.
It makes no sense.
The the fact like that if you are if you are using if you are serious about using LLMs to take your ideas and put in and get the most out of them, if you're serious about building, you know, what people are calling a personal OS, uh, and you are not using a centralized note-taking tool like this, yeah, that uses markdown as the foundation, then you are not using LLMs properly.
Yeah, or at least not at the limit.
Yeah, yeah, exactly.
You're not getting the most out of it.
Yeah.
You're not getting the most out of it.
So I think what's what's difficult about this is that it require it does require a lot of time and to actually set it up properly.
It requires uh like, yeah, I mean, it takes a lot of time and it and the UI is in an is is so daunting in the sense that it's a blank canvas, and it's not like hey, you should like write your preferences over here, or you know, you kind of just have to you have to come up with these ideas yourselves.
Yes, but that's still so amazing, right?
Because I mean, even when we work with other humans, we have to find a way to explain things to them, and I just think it's so cool that now we can work with these agents and we still have to explain things to them, but we only need to explain them once because once we get it down on into a file, we can always reference that file, that explanation of a project or a preference or anything, and it's always there, and you can pass it in.
Yeah, a file is like essentially perfect, a perfect memory, yes.
Right?
Human beings have memories, like we recall things.
Yes.
But there's tons of studies that show that what we remember, in fact, is completely different than reality.
For example, when we went and got our that haircut in Mississauga, I could have thought that I had the best haircut.
You know, that's what my memory remember remembers.
It was a great haircut.
But who knows?
It could have been the worst haircut I had ever gotten.
Now obsidian or whatever tool you end up using, like rem you know, if I had written like it the memory, the file, the markdown file is perfect.
So that when I link that or I recall it, it is going to give me a perfect uh date, you know, data point.
And the other thing about you know, these files is that you hope, well, they're not biased, basically, they're as biased as the human being is in terms of writing the reflections.
Yes.
At that moment in time.
Yes.
It's crazy, man.
And yeah, it's just crazy.
And there's all there's all of these different aspects to it.
There's the privacy of it and what that means.
There's the power of it, the fact that now you can just work with these computers in natural language and just delegate to them.
Um, there's the fact that there's people like me that are that are using these tools and and trying to figure out how to delegate stuff to agents in this way.
There's people that are like me that are even more hardcore in in different ways and and pushing them.
And I just think it's such a crazy time to be alive because I think we are potentially watching a fundamental shift in uh the human relationship to computers, and it's just I just I'm just really happy to be alive while this is happening, and I'm curious like how is this all gonna unravel?
Well, what's what's cool about this is 99.99% of people are not going to spend the time to actually set up something like this and make it a part of their daily lives, and the alpha, so to speak, is in terms of leading a more productive, happier, healthier, better, more money-making career is in in using something like this with an LLM.
I think so.
I'm not saying download obsidian today, and I have no affiliation or whatever with them, but I'm saying like pick a twit sounds like what we should all be doing, and I'm I'm talking, I'm giving myself this this advice.
It's like there's no excuse anymore for me not to be writing down and reflecting into markdown files, yes, in a world that where LLMs use markdown files as the oxygen, yes.
Like people think tokens are the oxygen, yes, but they're not, yes, the markdown fires are the memories.
Like, like think about what a human being is.
Yes.
You know.
Is a human being the energy of a human being, or is it the memories of it, you know, what we recall, you know?
Yes.
I mean, that's like a philosophical question.
And maybe it's a bit uh, you know, uh a bit of both, but it's you know, I think that there's something really, really fascinating about MD files as an underrated about them in order to have a true computer experience in in today's day and age.
Yeah, there's definitely something going on here, some fundamental shift.
Yeah, it's awesome.
Yeah.
And like I have bad work, my work, I you know, I'm learning in real time, right?
Like, and I I don't have the right vocabulary to even explain this.
Yes.
And neither do I, man.
Neither do I.
I I'm trying to I'm trying to figure it out in real time.
That's that's why I think like I know I show something, and and for me, I I'll do something or I'll see something.
And my friends are like, they kind of laugh because I'll just be sitting at my computer just tripping out.
And I think it's because I really like computers, and I and I cannot believe that this is possible.
I cannot believe that I can just be making notes on my computer like I have been since I was a kid, and then all of a sudden, this agent can scan through it and build things because of it, and and like connect patterns that I could never see.
It's nuts, man.
It's nuts.
And at the root of it, you're right, is just a collection of interrelated markdown files.
Yeah.
Cool, man.
I appreciate you.
I don't know if you can see my mind, but my mind is blown right now.
Thank God, man.
Yeah.
Yeah.
I wanted to do right by you.
I also just like I say this every time, man, but I'm just gonna keep always saying it to you all the time.
I really, really, really, really appreciate everything that you do.
I think your pattern recognition and your pattern matching is like like really underrated.
I think there's a lot of things that you do that I don't think it's like difficult to see if you're not really paying attention.
I just want to say like thank you for uh for everything that you do.
You're always putting on like new voices on your show.
I see it, I really appreciate it.
And it's just been it's just an honor to know you, and yeah, just thanks for the opportunity.
Thanks for everything, man.
I appreciate you, Vin.
You're a legend.
I'll include links for where to follow criminally underfollowed internet Vin on X on his YouTube show, uh podcast in the show notes and description.
You can go and check him out there.
Uh people, please play with some of these tools uh and let me know what you think.
Let Vin know what you think.
Please.
And uh Vin, I will beg you to come back on the show another time.
And I hope you come back on again.
For sure, man.
Thank you.
Thank you.