FAQ

First, ensure your system meets the minimum requirements (Windows 10 or 11, 8GB RAM, stable internet connection). Try restarting the application. If the issue persists, check your firewall settings — ExpellusAI requires a connection to our servers to function. You can also try deleting the config file from your app data folder (accessible via User Icon > Key File Location) and restarting the app.

This can happen if your Windows display scaling is set to something other than 100%. Go to Windows Settings > Display and set the scale to 100% (recommended), then restart ExpellusAI. If you are using multiple monitors, ensure your primary display is the one you intend ExpellusAI to interact with. The app automatically handles DPI awareness, but non-standard scaling may affect accuracy.

Close unnecessary background applications to free up system resources. Ensure you have at least 8GB of RAM available. If you are using the CUDA version, make sure your NVIDIA GPU drivers are up to date. For the best experience, we recommend 16GB RAM and a dedicated NVIDIA GPU with 4GB+ VRAM. If performance is still an issue, consider switching to the CPU version of the installer, which has lower hardware requirements.

This is usually caused by insufficient system resources or a conflict with another application. Try closing other programs, restarting your computer, and running ExpellusAI again. If the issue persists, check the log files in your app data folder — open it via User Icon > Key File Location in the app, then share the relevant logs in the Community forum for help.

On first launch, Windows SmartScreen may display a warning — click "More info" and then "Run anyway" to proceed. After that, the installer may take 15–30 seconds to decompress files before the Welcome page appears, especially for the CUDA version (which is larger). This is normal. If it takes more than a few minutes, try re-downloading the installer in case the file was corrupted during download.

ExpellusAI currently supports Windows 10 and Windows 11 only. macOS and Linux are not supported at this time. We are evaluating cross-platform support for future releases.

The CUDA version uses your NVIDIA GPU for faster screen analysis, resulting in quicker task execution. It requires an NVIDIA GPU with 4GB+ VRAM and updated drivers. The CPU version works on any Windows machine without a dedicated GPU but may be slightly slower for AI processing. Both versions have identical features — the only difference is performance.

Privacy is our top priority. Screen analysis (element detection) happens locally on your machine. Screenshots are sent to our secure servers only for AI task processing and are never stored — they are discarded immediately after the AI generates a response. Your task prompts are sent to our servers for AI processing. We do not sell or share any user data with third parties. For full details, see our Privacy Policy.

When your token balance reaches zero, you will not be able to execute new AI tasks. You will see a message indicating insufficient tokens when you try to send a query. Your existing configurations and settings will remain intact. Token balance can be checked in Settings > Status. Contact our team for information about token replenishment options.

You can report bugs and request features through our Community board. Navigate to the Community page and create a new post under the appropriate category. Our team actively monitors and responds to community feedback.

ExpellusAI can interact with most Windows desktop applications that have a standard graphical user interface. This includes browsers, office suites, file managers, email clients, terminals, and more. However, some applications with non-standard rendering (like certain games or specialized CAD software) may have limited compatibility. The AI Browser feature also provides dedicated web automation capabilities.

Use the Flow Canvas to visually build your task. Drag an "AI Request" block onto the canvas, type your instruction in natural language (e.g., "Open Notepad and type Hello World"), and click execute. For multi-step tasks, you can chain multiple AI Request blocks together. The AI will interpret your instructions and perform the actions on your screen automatically. You can configure iteration limits and delays in the Configuration panel.