RPA Self-healing with Copilot
Designed a self-healing system for Power Automate Desktop that automatically repairs broken UI selectors during runtime using Copilot, enabling reliable automation even when application interfaces change
Overview
The Self-healing with Copilot feature in Power Automate for desktop is designed to enhance the reliability of UI and web automation by automatically locating and repairing UI elements during runtime. This feature leverages Copilot to identify and fix invalid selectors, ensuring that automation processes can continue smoothly even when UI elements change. It works in both attended and unattended modes, providing old selectors and UI hierarchy to Copilot for accurate repairs. Additionally, users can save the repaired selectors for future use, ensuring ongoing automation success.
Key capabilities include:
- Automatic detection of invalid UI selectors during runtime
- Real-time repair of broken selectors using Copilot
- Support for both attended and unattended automation
- Intelligent analysis of UI hierarchy for accurate repairs
- Ability to save and reuse repaired selectors
Key Responsibilities
- Led the design and implementation of the AI-powered self-healing system for Power Automate Desktop
- Created an intuitive interface for reviewing and applying Copilot's repair suggestions
- Collaborated with engineering teams to implement feature for attended and unattended modes
Impact
The Self-healing with Copilot feature has transformed UI automation reliability:
- Significantly reduced automation failures due to UI changes
- Enabled continuous operation in both attended and unattended scenarios
- Decreased maintenance time for UI automation flows
- Improved automation reliability across different applications
- Reduced the need for manual selector updates
- Enhanced user confidence in long-running automation processes
- Established a foundation for self-healing UI automation
Project Videos (1)
Power Automate Self-healing System
Introducing AI-powered self-healing for cloud flows