The 3-File AI System That Works With ANY MODEL

If you're wiring your entire life into

AI tools that didn't exist six months

ago, systems built on someone else's

infrastructure,

then what happens when that tool

changes? When the API breaks, the

company pivots, or your favorite model

gets deprecated? Will you reach for your

ideas only to realize they were never

actually yours? This video is about

making sure that never happens to you.

I'm going to share what I'm calling my

AI OS, an AI system that you actually

own. Just a handful of plain text files.

You're not locked into Claude, OpenAI,

or whatever everyone's building their

lives around this week. Because when the

next shiny AI paradigm shift happens,

you want your system to still be

standing.

I'm Nick Milo, and I'm here to help you

build an idea verse, a home for your

best thinking. Let's get started. It's

no secret that I'm a huge fan of

Obsidian. It's the world's current best

app to manage your thinking, your

projects, and your life. But what really

sets it apart isn't its features. It's

that you aren't locked in. Your notes

are literally just text files in folders

sitting on your computer. Which means

you don't need Obsidian to access your

knowledge. You could open your idea

verse in any app that can read plain

text and edit them immediately. So, if

Obsidian goes away tomorrow, you're

completely fine. Obsidian CEO calls this

philosophy file over app. But, as we

enter the world of AI, it's not just

your notes app that we're talking about

anymore. We have to take it one layer

deeper. We need not just file over app,

we need file over AI. If you spend

hundreds of hours setting up your

personal preferences and structuring

your files to be perfectly read by

Claude or Codex or whatever's hot right

now, only to have that tool go offline,

you've lost access to your entire

workflow. That's why we want file over

AI. You want your AI processes owned by

you, built into your core files and

folders that carry the the of your AI

operating system, and not locked away in

any specific tool. Your core files and

folders should move with you and your

ideaverse. Here's exactly what a system

like that looks like. This is the my AI

OS system. It works in any app with any

AI model.

So, layer one is your ideaverse. This is

your own universe of knowledge that

you've been building either for years or

you can start building today. This is

all your notes, your ideas, your

connections in simple files that you can

take into any app anywhere. Now, I break

my ideaverse into three big sections.

The Atlas for my knowledge, calendar for

notes related to time, and then efforts

for projects I'm working on. ACE for

short.

Your ideaverse is the center of your AI

OS. The experiences and thinking that

aren't available to anybody else but

you.

And the more you put into it, the

smarter any AI becomes when it starts

working with you. Layer two, your maps.

This is the connective tissue. A few

core files that sit between your

ideaverse and any AI tool. It makes your

knowledge able to be navigated by AI. I

call the files in layer two, your maps

and manuals. Think indexes for both you

and any AI working alongside you. And

there are just two essential files that

matter most, which we'll talk about in

just a minute. Layer three, your tools.

These are the actual applications you

use. Now, it could be Obsidian to

navigate your notes, and then it could

be Claude co-work to work on the files

themselves. Notebook LM, open Claude,

Claude code, Gemini. The tools in your

toolkit will change, and that's because

this is the most replaceable layer.

Models update, apps get discontinued or

updated, new tools break out, and that's

fine because your ideaverse and maps

stay central. So, no matter which app

rotates in or out, you'll be ready for

it. These are the three layers of your

AI OS, and for this system to work, we

need to focus on the middle layer where

the core files that connect your idea

verse to the outside tools lives. Think

about it. Point an AI at a vault, folder

with tons of notes, 20,000

interconnected notes, and it's going to

struggle. And it takes a lot of tokens.

It takes a lot of time. It takes a lot

of resources. But, when you give it a

map, it can navigate. It can figure out

which files are actually relevant,

isolate those, skip the rest without

having to scan everything. Fast, rapid

context building. That's what effective

AI needs. Maps of content are something

I've been refining for years to make

sense of and navigate my most important

thinking. But, today maps are even more

powerful than ever, and that's because

AI needs the exact same structure to

navigate large amounts of context that

we need. Your maps show the information

highways which you or AI can navigate.

As long as you build your maps in a way

that gives sufficient context, you could

hand them to an external tool, and it

should be able to work in your idea

verse immediately, especially if you use

a folder system like Ace that gives each

section of your idea verse a clear guide

to what's inside and how to engage with

it as well. So, now let's talk about the

two core files that every AI OS needs so

that you can connect your idea verse to

any tool you want to work with now and

deep into the future. Before anything

else, set up your basic folders. If

you're in Obsidian, you have a vault

folder, and then Ace will be your three

core folders underneath. Check out this

video to set up Ace in the best possible

way and explain what these bonus

folders, the plus folder and the X

folder, are for. But, for now, just

create a folder for each to start with.

Once Ace is set up, we can create our AI

OS folder, and this is where all of our

AI related maps will actually live,

except for one. The first map file to

create is your me.md, a map for your

portable identity. If you've used Claude

Co-work or Code, you might know about a

Claude MD, a master file Claude reads

every time you start a session. But,

that's Claude specific. If you switch to

a different AI tool next month, that

file doesn't go with you. Instead,

you're going to create your own me.md,

just a markdown file. Just make a new

note in Obsidian, and you can just call

it me. If we go to the file explorer or

the finder, we'll see that it says

me.md. Once you make it, just start by

adding around 50 to 80 lines that tell

AI, "Here's who I am. Here's how I

think. Here's how I want you to work

with me." That's why it's called me.md,

because it's for you. It's your file,

and it's portable. It's the guiding

light file that you want your AI tools

to read anytime you start a work

session. Here's what mine looks like

right now. At the top, I make it clear.

This is your briefing on who I am, how I

think, and how to interact with me.

There are a couple quick links for me to

navigate through if I would like, so I

can just jump around to these other ones

that we'll look at shortly, but AI works

best with file paths. So, I'm giving it

very clear file paths to find the other

most important maps. So, I have a me

section, and then my working

preferences. So, in the me section, I

have a high-level summary statement of

what I'm about, some first principles,

some linking your thinking mantras that

I live by, a hypothesis on meaning,

values and virtues, how I think. You can

see some sections I'm still filling out,

but sections that I have in here,

key frameworks and intellectual tools,

just a lot of stuff that's going to

inform AI on how to work with me. I

quite like this intellectual lineage

with some thinkers that have really