Posted by throwaway270925 8 hours ago
Based on the marketing it seems to run a sandboxed copilot instance that can impersonate the user to take actions, with their permission?
Something like "hey copilot install Putty"? and it does it?
I can relate to the reluctance to adopt AI features into the OS -- but I would also like to understand how they work and any utility they might provide.
I agree that a "good" implementation of agentic AI can have a lot of benefits, to casual users and power users both. But do I have any trust in Microsoft being the company to ship a "good" implementation? Hell no.
Windows has been getting more and more user hostile for years now, to casual users and power users both. If there's anyone at Microsoft who still cares about good UX, they sure don't have any decision-making power. And getting AI integration right is as much a UX issue as it is a foundation model issue or an integration hook issue.
From the MS support doc:
> "An agent workspace is a separate, contained space in Windows where you can grant agents access to your apps and files so they can complete tasks for you in the background while you continue to use your device. Each agent operates using its own account, distinct from your personal user account. This dedicated agent account establishes clear boundaries between agent activity and your own, enabling scoped authorization and runtime isolation. As a result, you can delegate tasks to agents while retaining full control, visibility into agent actions, and the ability to manage access at any time."
MS showed a little bit of something like it at Ignite yesterday, but for enterprise automations, the AI spun up a Windows 365 instance, did some stuff on the web, then disposed of it when it was done.
My concern is that the Windows Credential itself doesn’t have a ton of value (opening windows apps) but the browser cookie jar (e..g Edge or Chrome) , which the Credential unlocks, has tremendous value — and threats.
The core problem is lack of granularity in permissions. If you allow the agent to do browser activities as your user, you can’t control which cookie / scope it will take action on.
You might say “buy me chips” and it instead logs into your Fidelity account and buys $100k worth of stock.
Let’s see how they figure out the authorization model.
Started with Linux Mint then Debian/Ubuntu, tried some others too but ultimately just stuck with Ubuntu
Knowing nothing about how to configure it, I installed it with the graphical installer, booted into a tty, installed claude-code, checked the config files into git, and proceeded to vibe-code a basic sway (now niri) environment to see what it would feel like.
A month later, my NixOS environment is so much better than my heavily optimized macOS environment that I sheepishly use it inside a VM on macOS (UTM) or VNC to my desktop machine so I can use it from my bed.
LLMs really open the doors of desktop Linux since you can git clone all of your deps locally (your window manager, keepassxc, waybar, your apps, nixpkgs, home-manager, even the linux kernel, etc., etc.) and the LLM can dig into source code and web search to do things for you or debug issues. And NixOS adds a level of observability into what's going on since any changes show up in git-diffed config files.
If anyone is like me and used macOS because you used to use Linux but couldn't be bothered anymore when you'd run into a rough edge, you might find it fun to use NixOS + claude-code (or equivalent) running in ~/nix-config.
But: you will have issues.
I don't know if I would use the word approachable
IMHO, stuff is moving fast enough in the Linux gaming world that any distro built around taking its time to update things (i.e. Debian, Ubuntu, Mint) is liable to be a bad time. Anecdotally, I've found that redirecting new users interested in gaming away from those distros has dramatically improved their satisfaction.
oh no
I recently vibe coded a voice typing software (using Parakeet — your best bet is probably Handy though).
It works in my terminal. (I just changed my paste shortcut to Ctrl+V
I can now literally speak software into existence!
I made a thin wrapper around my llm() function I can pipe text into from Bash.
This allows me to make many other thin LLM wrappers, such as one that summarizes then contents of entire directories.
I have a thing called Jarvis inspired by a Twitter post, where I ask it to do anything in bash, and it just does that.
I wouldn't exactly say it's useful (I am unemployed) but I am kind of having my mind blown a little bit.
The future is already here, it's just not evenly distributed yet.
Thing is, we can talk faster then most of us can type.
Voice + Programming is slow because of all the special symbols. But voice + vibe coding? The ability to tell your LLM to do tasks, while you focus on other parts of the code, without the need to switch tabs/windows.
What about "change the color green on this element (html page), where my mouse is pointing"... Annoying with keyboard if you need to switch windows, very possible with voice.
And LLMs are very forgiving for mistakes, unlike if you want to voice program where every symbol needs to be accurate.
People do not realize that programming as we know it, is going to change.
I saw yesterday that I had been approaching software incorrectly. It feels futuristic because it's so fast, but it's still linear. One guy making one thing at a time (with some help from the computer).
But software can now be made so rapidly, that the bottleneck is actually curation. You can now generate a hundred ideas for software and a prototype for each one in the time it takes to make some coffee.
Going through all of it is the part that doesn't scale, it's bottlenecked by the individual. That's the reward function, right? Taste, discernment.
At this point software can grow itself, it can mutate, and it can combine with other software. I think building is entirely the wrong metaphor now.
I think a better metaphor would be a genetic algorithm. You try a bunch of stuff and see what works. Then you combine the best parts.
The general rule here is that you use it for what it's good for it's actually really good.
The "typing into my terminal" is mostly for interacting with Claude Code. I wish that part worked on my phone.
Although I do use the voice typing tons for text chat, ironically.
1. Linux decades ago was not "new user friendly"
2. Wine and PLayOnLinux was all we had with endless problem, and heavy dependency on Windows files like DirectX and libraries
3. Windows dominated the gaming market
4. 3D GPU driver was non-existent
The single reason why gaming on Linux now is better than Windows, has one name: Valve
SteamDeck/SteamOS changed everything, the whole Wine process is managed by the OS and no longer by the user. You may need to change the Proton version, that is all. That also pushed GPU drivers to be better supported on Linux.
Valve single handled what gaming on Linux has become. I run Mint Cinnamon Linux, and even tho it is not "SteamOS", I can play Steam games just fine.
Microsoft terrible takes and AI, is also pushing gamers over to Linux, better FPS on Linux than Windows. The only restriction is kernel anti-cheat software that only runs on Windows, but many games do not use that and the ones that do use it like COD(dead game), BF, etc, isn't everybody cup of tea.
If it wasn't for Valve, Linux gaming would still be as dead as it has always been.
To make it more perfect, users that use their computer for browsing, writing docs (LibreOffice), etc, can be done on Linux for free.
You as a computer user in 2025, you have little to no excuse to try Linux, but try something good like Mint Cinnamon Linux that is extremely new user friendly, good for browsing, good for development work, solid for gaming, video editing is chef kiss, etc, etc. Avoid Ubuntu (they are going proprietary).
Doesn't COD have like over 100 million monthly active users?
I've been utterly astounded by Proton in the last year. Nearly every game I have run has run just about perfectly, often better than on Windows, and I'm able to play them with an Xbox One pad no less.
Valve absolutely deserves a lot of credit, but I do think that the constant effort from the Wine people should get a lot of credit as well. Wine has had constant progress for three decades, with every release getting a little better. I haven't worked on it, but I suspect 90+% of the work with Wine is figuring out all the weird edge cases that have popped up on Windows throughout the years, which is often slow, tedious, thankless work. Valve did a lot of work but there's a reason they opted to improve Wine instead of writing Proton from scratch.
Steam Proton makes the whole process painless, you only select which Proton version to run, and that info can be obtained from ProtonDB if you encountered any issue, it is beautiful.
As for Linux, even emulators works like never before. I could never get PS4 emulator to work on Windows, I got PS4, X360, GameCube, and a bunch of other emulators running on Linux like I couldn't believe it.
You can do the same from within SteamOS itself, you just install an app, select the emulator and you ready which is far easier than me doing this from from Linux.
I just want to give credit where credit is due, because a lot of this wouldn’t be possible without the hard work of the Wine people. “Shoulders of giants” and whatnot.