Understanding AI Grounding in Mobile Testing

Traditional automated testing relies on fragile element selectors that break when UIs change. Revyl takes a different approach using AI grounding.

What is AI Grounding?

AI grounding is the process of connecting natural language instructions to specific UI elements on screen. When you say “tap the login button,” our AI:

1. Captures the current screen state
2. Identifies all interactive elements
3. Maps your instruction to the most relevant element
4. Executes the action with precision

Why This Matters

Resilience to UI Changes

When a developer moves a button or changes its text from “Login” to “Sign In,” traditional tests break. AI-grounded tests understand intent, not just selectors.

Natural Language Interface

Write tests the way you think about them:

• “Scroll down until you see the Subscribe button”
• “Type ‘[email protected]’ in the email field”
• “Verify the success message appears”

Cross-Platform Consistency

The same natural language test works on both Android and iOS, even though the underlying UI structures differ.

The Technical Stack

Revyl’s grounding pipeline uses:

• Vision models for screen understanding
• BAML for structured LLM interactions
• Platform-specific automation (UIAutomator2, XCTest) for reliable execution

Results in Practice

Our customers see:

• 80% reduction in test maintenance
• 3x faster test creation
• Higher confidence in test reliability

Stay tuned for more deep dives into our technology stack.