Desktop Automation is only available in the Profy Desktop client and supports macOS, Windows, and Linux. This capability is not available in the web version.
Platform Support
The underlying driver is powered by trycua/cua-driver-rs, communicating through a unified MCP protocol interface across all three platforms with consistent tool behavior.Core Capabilities
Mouse Control
Move the cursor, single-click, double-click, right-click, drag — Expert can precisely operate any visible element on the desktop.
Keyboard Input
Type text in any application, execute keyboard shortcuts, and switch input methods.
Screen Understanding
Expert can “see” your screen content, understand the interface layout and current state, and make operation decisions accordingly.
Cross-Application Coordination
Switch between multiple desktop applications — copy files from Finder to an email client, or extract data from a browser and paste it into a spreadsheet application.
Phantom Cursor
Profy Desktop uses a phantom cursor — the Expert operates in the background without moving your real mouse.- Cursor position protection: Your cursor position is saved before each action and restored afterwards
- Move action blocked: The
moveaction is hard-blocked — Expert can onlyclickto jump directly to targets - Collision Watch: If you move your mouse during Expert’s operation, the system auto-pauses to avoid conflicts
- Emergency Stop: Press
ESCat any time to immediately halt all desktop automation
How to Use
1
Open Profy Desktop
Make sure you’re using the Profy Desktop client (not the web version).
2
Describe the Task
Tell Expert in the conversation what you’d like to accomplish on the desktop. For example: “Open System Preferences, find the Display options, and change the resolution to default.”
3
Observe and Confirm
Expert will begin operating your desktop. You can watch the mouse movement and operation process in real time. If a step requires your confirmation, Expert will pause and ask.
4
Get Results
Once the task is complete, Expert will report the results in the conversation.
Typical Use Cases
Batch File Operations
Batch File Operations
Rename a batch of files, organize folders by rules, or batch-modify file properties — tell Expert the rules and it will execute them one by one.
Application Settings
Application Settings
Have Expert complete complex configuration changes in system settings or application preferences, saving you the time of navigating menus.
Data Transfer
Data Transfer
Extract data from one application, convert the format, and input it into another — for example, copying table data from a PDF reader to Excel.
Repetitive Task Automation
Repetitive Task Automation
Any mouse-and-keyboard operation sequence that you need to perform repeatedly can be delegated to Expert.
Learning Application Operations
Learning Application Operations
Unfamiliar with how a desktop application works? Have Expert demonstrate while explaining each step in the conversation.
Security
Image Pipeline
To reduce model token cost and speed up understanding, screenshots go through an optimization pipeline:- JPEG Compression: Screenshots are automatically converted from PNG to JPEG (max 1280px width, 80% quality), reducing file size by 60-80%
- Apple Vision OCR (macOS): Uses the native Vision Framework to extract text from screenshots without cloud API calls
- Post-action Sleep: A configurable 100ms delay after each action lets the UI settle before the next screenshot
App Approval
Desktop automation enforces per-app permission levels:
You can customize per-app permissions in Settings.
Recommendations
Settings
In the Profy Desktop settings panel, you can configure desktop automation behavior:- Any App: Whether to allow operation on all desktop apps
- Chrome Control: Whether to allow control of the user’s real Chrome browser
- Post-action Sleep: Wait time before screenshots (adjustable 50–500ms)
- OCR: Whether to enable text extraction from screenshots
- Collision Watch: Whether to auto-pause when user mouse movement is detected
Browser Automation vs. Desktop Automation
These two capabilities complement each other — use Browser Automation for web-related tasks and Desktop Automation for local application operations.

