I am considering deepin as my future main Linux system, mainly because UOS AI and Claw Mode are moving toward something I believe is much more useful than a normal AI chatbot: an AI that can actually understand and manage the computer.
However, after watching several recent deepin 25/25.2 reviews, I noticed that most reviewers mainly test the desktop, App Store, GIMP, browsers, Steam and general usability. UOS AI itself is rarely tested in depth.
For me, the really interesting question is:
How far can UOS AI go in diagnosing, monitoring and optimizing the actual computer?
I recently had a hardware/sensor-related problem on my Windows PC. To diagnose it, I spent almost an entire day manually collecting screenshots, sensor readings, BIOS information, logs and other data and sending them to ChatGPT.
A deeply integrated system AI could potentially collect and analyze this information itself within minutes.
I would love to be able to tell UOS AI:
“Check my computer completely for hardware or system problems.”
UOS AI could then automatically inspect things such as:
CPU and GPU temperatures and clock behaviour
RAM usage and, where available, ECC/EDAC error information
GPU VRAM usage
SMART/NVMe/SSD health
kernel and system logs
driver status
PCIe problems
unusual background processes
system services
repeated hardware or software errors
It could then explain what it found and distinguish between:
Normal behaviour → something worth monitoring → a real problem requiring action.
Learning the individual computer
I think an AI operating system should also be able to observe the computer for an initial period and learn its normal behaviour.
For example, after a day or several days it could know:
normal idle temperatures
normal gaming temperatures
typical RAM/VRAM consumption
which applications are normally used together
which background processes can safely be reduced during gaming
normal CPU/GPU performance
unusual behaviour compared with the learned baseline
The AI should not constantly interfere with the system.
Ideally it would learn suitable profiles and then quietly apply them in the background.
Video playback: use a quieter and more efficient profile.
Local AI workloads: prioritize GPU memory and compute resources.
Safety is very important
For system-level AI actions, I would also like to see:
a clear proposed action plan before important changes
user approval for critical actions
automatic backups/snapshots
undo/rollback if an AI-generated change causes problems
a history showing what the AI changed and why
local processing whenever possible
transparent permissions for files, terminal access and system settings
MakuluLinux / Electra as an interesting example
Another Linux project that may provide useful inspiration is MakuluLinux AI-OS and its Electra AI system.
Their developers demonstrate AI agents performing multi-step system tasks, system troubleshooting, monitoring and proposed actions before execution.
I am not suggesting that deepin should copy MakuluLinux. I believe deepin has the opportunity to build this kind of functionality into a much broader and more mature Linux desktop platform.
I especially find their demonstrations of agent-based system tasks and AI-generated action plans interesting.
One more suggestion: please demonstrate UOS AI more deeply
I would also really like to see an official deepin video demonstrating things such as:
“Diagnose this PC.”
“Check why this GPU is running hot.”
“Find the cause of these system errors.”
“Check the graphics driver and Vulkan configuration.”
“Find out why this game is stuttering.”
“Optimize this computer for gaming, but show me what you plan to change first.”
These demonstrations would show much more clearly what makes UOS AI different from simply having an AI chat application on Linux.
I believe this kind of system-level, local-first AI could become one of deepin's strongest advantages and could move Linux beyond what Windows currently offers in AI-based system management.
中文 
Hello deepin Team and Community,
I am considering deepin as my future main Linux system, mainly because UOS AI and Claw Mode are moving toward something I believe is much more useful than a normal AI chatbot: an AI that can actually understand and manage the computer.
However, after watching several recent deepin 25/25.2 reviews, I noticed that most reviewers mainly test the desktop, App Store, GIMP, browsers, Steam and general usability. UOS AI itself is rarely tested in depth.
For me, the really interesting question is:
How far can UOS AI go in diagnosing, monitoring and optimizing the actual computer?
I recently had a hardware/sensor-related problem on my Windows PC. To diagnose it, I spent almost an entire day manually collecting screenshots, sensor readings, BIOS information, logs and other data and sending them to ChatGPT.
A deeply integrated system AI could potentially collect and analyze this information itself within minutes.
I would love to be able to tell UOS AI:
“Check my computer completely for hardware or system problems.”
UOS AI could then automatically inspect things such as:
It could then explain what it found and distinguish between:
Normal behaviour → something worth monitoring → a real problem requiring action.
Learning the individual computer
I think an AI operating system should also be able to observe the computer for an initial period and learn its normal behaviour.
For example, after a day or several days it could know:
The AI should not constantly interfere with the system.
Ideally it would learn suitable profiles and then quietly apply them in the background.
For example:
Gaming: prioritize GPU/CPU performance, monitor temperatures and VRAM, reduce unnecessary background load.
Video playback: use a quieter and more efficient profile.
Local AI workloads: prioritize GPU memory and compute resources.
Safety is very important
For system-level AI actions, I would also like to see:
MakuluLinux / Electra as an interesting example
Another Linux project that may provide useful inspiration is MakuluLinux AI-OS and its Electra AI system.
Their developers demonstrate AI agents performing multi-step system tasks, system troubleshooting, monitoring and proposed actions before execution.
I am not suggesting that deepin should copy MakuluLinux. I believe deepin has the opportunity to build this kind of functionality into a much broader and more mature Linux desktop platform.
MakuluLinux AI-OS / Electra development videos:
https://www.youtube.com/@makulu.online/videos
I especially find their demonstrations of agent-based system tasks and AI-generated action plans interesting.
One more suggestion: please demonstrate UOS AI more deeply
I would also really like to see an official deepin video demonstrating things such as:
“Diagnose this PC.”
“Check why this GPU is running hot.”
“Find the cause of these system errors.”
“Check the graphics driver and Vulkan configuration.”
“Find out why this game is stuttering.”
“Optimize this computer for gaming, but show me what you plan to change first.”
These demonstrations would show much more clearly what makes UOS AI different from simply having an AI chat application on Linux.
I believe this kind of system-level, local-first AI could become one of deepin's strongest advantages and could move Linux beyond what Windows currently offers in AI-based system management.
Thank you for considering the suggestion.
Best regards,
Silverman