SystemX

Inspiration

We have all experienced frustrating computer problems: frozen applications, slow performance, failed updates, or internet connections that suddenly stop working. Fixing these issues often requires searching online, understanding technical instructions, and navigating multiple Windows settings.

We wanted to build something simpler: an AI assistant that looks after your computer, notices problems before you do, and helps fix them without requiring technical knowledge.

That idea became SystemX, a personal AI-powered Windows troubleshooting assistant.

What it does

SystemX runs quietly in the Windows system tray, continuously monitoring your computer for problems.

Automatic detection: Identifies frozen applications, network failures, low disk space, Windows errors, and other system issues.
AI-powered diagnosis: Uses Grok or other supported AI models to investigate unfamiliar errors and recommend solutions.
Offline troubleshooting: Handles common problems through built-in repair playbooks, even without internet access.
Smart notifications: Alerts users immediately about serious issues while grouping less urgent problems into a daily summary.
Safe repairs: Explains exactly what a repair will do and requires user approval before making changes.
Verification and history: Checks whether each repair worked, records the actions taken, and supports undo when possible.

Users can also describe a problem in natural language and let SystemX investigate it.

Our goal is to make computer troubleshooting as simple as having a conversation.

How we built it

We developed SystemX primarily in Python, using Windows system APIs and psutil to monitor processes, hardware resources, networking, and system events.

For the interface, we used HTML, CSS, JavaScript, FastAPI, and pywebview, with pystray providing Windows system tray integration.

The AI component connects to xAI's Grok API, with support for additional providers including OpenAI, Claude, Gemini, Groq, and OpenRouter. AI models can request diagnostic checks but cannot execute arbitrary system commands.

We implemented local troubleshooting playbooks and a controlled repair system using Python and PowerShell. Sensitive operations require explicit approval, while an isolated administrator helper handles repairs that need elevated permissions.

We also use SQLite to maintain repair histories, Windows Credential Manager for API key storage, and pytest for automated testing.

Challenges we faced

The biggest challenge was balancing powerful AI troubleshooting with system safety.

Allowing an AI model to execute arbitrary commands could damage a user's computer. To address this, we designed a restricted toolbox where every repair must pass validation and receive user approval.

Another challenge was ensuring that SystemX remained useful without internet access. We separated common troubleshooting tasks into local playbooks, reserving AI calls for unfamiliar problems.

We also had to consider Windows administrator permissions, unreliable system signals, and preventing excessive notifications.

Accomplishments that we're proud of

We are especially proud of designing SystemX to be more than a chatbot that gives troubleshooting advice.

It combines proactive monitoring, real Windows diagnostic tools, AI reasoning, controlled repairs, and post-repair verification in one desktop application.

We also designed a repeatable testing environment using Windows virtual machines to evaluate both repair effectiveness and safety.

What we learned

Building SystemX taught us that integrating AI into operating-system tools requires much more than connecting to a language model.

We learned how to work with Windows system events, design permission-controlled repair workflows, integrate multiple AI providers, and separate AI decision-making from actual system execution.

Most importantly, we learned that a useful AI assistant should be able to take action while keeping the user informed and in control.

What's next for SystemX

We plan to expand the library of supported repairs, improve AI diagnosis for unfamiliar Windows errors, and evaluate the system across more real-world computer problems.

Our long-term vision is to make SystemX a reliable personal IT assistant that helps anyone maintain and troubleshoot their computer with minimal technical knowledge.