I used Windows for 25 years (MS-DOS, Win95 and Win98 too, but XP was the first OS I installed and maintained myself, so that’s where I count from). I knew all the shortcuts, all the workarounds, all the tricks to make it behave the way I wanted. Switching to Linux was never on the table. It was for server people and hobbyists, not for daily work.
Then I started using AI tools seriously, and within a few months I was running a tiling window manager on Arch Linux, configuring everything from a terminal. And I have no plans to go back.
This is the story of how AI made me switch operating systems, and why I think more people will follow.
The band-aids I didn’t know I was using
Before the switch, I was already trying to make Windows work the way I wanted. I installed PowerToys for better window management and virtual desktops. I used Windhawk to customize the taskbar into something dock-like. I ran Chris Titus’s WinUtil to disable the background services that nobody asked for but that ate resources anyway. I was pushing Windows to behave more like a focused, minimal workspace.
What I didn’t realize was that I was building workarounds for limitations that shouldn’t exist. Everything I was trying to add to Windows, features like proper workspace management, tiling windows, a customizable taskbar, these are just how Linux works by default. I was spending energy duct-taping a system instead of using one that was designed for this from the ground up.
The first step: a friendly Linux
I didn’t jump into the deep end right away. I started with Pop!_OS, a Linux distribution with a polished desktop that feels familiar if you’re coming from Windows or macOS. It was a good experience (even with the current beta Cosmic DE). I got to see how much better AI tools worked in a Linux environment, where the terminal is a first-class citizen and everything is transparent.
But I hit a ceiling quickly. A pre-packaged desktop environment is still someone else’s idea of how your computer should work. After working with AI in the terminal, I wanted more control. I wanted the ability to shape the whole system around how I actually work.
Building a computer from scratch (with an AI partner)
So I switched to CachyOS with Hyprland, a tiling window manager where there are no menus, no settings windows, no drag-and-drop configuration. Everything is controlled through text files and terminal commands.
A freshly installed system is a blank canvas. I thought it would be a fun experiment to paint on that canvas with AI, so the very first thing I set up after the base system was Claude Code, an AI coding assistant in the terminal, the same kind of tool I made the case for in episode 3 (The Renaissance of CLI). From there, I used it to install and configure everything else. I didn’t look up tutorials. I didn’t read through wiki pages. I described what I wanted, and the AI translated that into the right commands and configurations.
“Set up my drives to mount automatically.” Done. “Make the status bar show my CPU and memory usage with progress bars.” Done. “Configure keyboard shortcuts so Super+T cycles between terminal windows.” Done.
It’s a completely different way of setting up a computer. Instead of accepting what someone else designed for you, you build the system you actually want, brick by brick. The AI doesn’t care if you want something standard or something unusual. It just helps you build it.
When the system fits your brain
The result is a workspace where everything works the way I think. The shortcuts make sense to me. The information I need is always visible. The things I don’t care about are gone. There’s no notification center I didn’t ask for, no weather widget I’ll never look at, no “suggested apps” appearing in my start menu.
A small example. I use Todoist to manage my tasks. On Windows, that means a browser tab or an app window. On my Linux setup, I integrated it directly into the status bar. I can always see how many tasks I have left today. And because my AI assistant can talk to the Todoist API, tasks get added and checked off while I’m working, without me ever leaving my terminal. New task comes up while coding? I mention it, the AI adds it. Finish something? It gets checked off. I stay in the flow.
Or take my Logitech mouse. On Windows, you need Logi Options+, a 500 MB application that runs constantly in the background, just to configure extra buttons and the radial menu. On Linux, there’s no official Logitech software at all. So I recreated the radial menu with simple scripts. I even customized the haptic feedback. No bloatware, no background process, just a few lines of code doing exactly what I need.
After a while, I realized something. The limits aren’t the software anymore. The limits are my time and my imagination.
The numbers that sealed it
When my fully configured system was up and running, with the status bar, the workspace management, all my development tools, the terminal AI assistant, everything, it was using less than 2 GB of RAM.
My Windows 11 setup, after I had spent time disabling services and minimizing background processes, used 6 to 8 GB just sitting there (and this usage could increase with each update, so you might understand now why Microsoft recommends 32 GB of RAM for a “future-proof” PC build).
That’s not a small difference. That extra memory goes to AI models, development servers, Docker containers, the actual work I’m trying to do. The computer stopped wasting resources on itself and started spending them on me.
It’s not the Linux you remember
If you tried Linux years ago and it was a mess of driver issues and broken packages, I understand the hesitation. I had the same expectations. I thought I’d be spending half my time fixing things that just work on Windows.
That’s not how it is anymore. The Linux desktop in 2026 is genuinely stable. Software installation is clean and hardware support is broad. And when something does need fixing, the AI assistant that lives in my terminal can usually sort it out in seconds.
And honestly, I had already moved most of my software stack to open source years ago. GIMP and Krita instead of Photoshop. LibreOffice instead of Microsoft Office. Inkscape, Blender, OBS. So when I made the switch, there wasn’t much left on Windows that I would miss.
The only thing I still boot into Windows for is gaming. And honestly, even that gap is closing fast thanks to Proton and the Steam Deck ecosystem. But I actually like having the dual boot now. Linux is my workspace. Windows is the gaming console. The fact that I have to physically reboot to play games is the best procrastination barrier I’ve ever had.
And I’m not the only one looking
Looking around, I realize this isn’t just my story. Three things are pushing more people in the same direction right now.
Windows keeps getting heavier. Microsoft is putting AI features like Copilot and Recall into the OS whether you asked for them or not. Each update adds more background processes and more “suggestions” you didn’t request. The system that used to be familiar is becoming something else.
Windows 10 support officially ended on October 14, 2025. The Extended Security Updates program buys another year, until October 13, 2026. After that, the millions of PCs that can’t run Windows 11 because of the TPM and CPU requirements will be left without security patches. For those machines, the cheapest path forward is a Linux distribution. Microsoft has, ironically, become one of the strongest arguments for switching.
And as I mentioned, gaming on Linux finally works thanks to Valve. The biggest historical blocker is mostly gone.
None of these are why I switched. The AI workflow is what pulled me in. But they’re why I don’t think this is a niche move anymore.
I didn’t switch to Linux because I’m a hobbyist or because I wanted to prove something. I switched because AI changed what’s possible with a terminal, and once the terminal became easy, Linux became the obvious choice.
Twenty-five years of Windows gave me a system that someone else designed. A few months with Linux and an AI assistant gave me a system that feels like mine. Every shortcut, every integration, every pixel on my screen is there because I decided it should be.
I don’t think Linux needs everyone to switch. But if you’ve been curious, and if you’ve already started using AI tools in your daily work, the step across is smaller than it used to be. It might even be gone already.