Testing websites manually can be time-consuming and error-prone. x360 AI RunTest simplifies automation by letting you write tests in plain English instead of fragile CSS selectors or XPath. It uses AI to translate your instructions into real browser actions, making tests self-healing and reliable.
Key Features:
AI Self-Healing Tests: Automatically recovers when selectors break due to UI changes.
No Selectors Needed: Write tests in natural language.
Playwright Native: Works alongside your existing Playwright suite.
Evaluation Engine: Classifies test results as Passed, AI Smart Passed, or Failed.
Multi-LLM Support: Azure OpenAI, OpenAI, Anthropic Claude, Google Gemini, DeepSeek.
Quick Example:
await agent.executeTask([
'Navigate to https://app.example.com/login',
'Enter john@example.com in the Email field',
'Enter SecurePass123 in the Password field',
'Click the Sign In button',
'Verify the Dashboard heading is displayed',
].join('\n'));
Benefits:
Reduces test maintenance by over 80%
Integrates easily with CI/CD pipelines
Provides detailed logs, screenshots, and evaluation reports
Learn More: https://x360aitech.com/platforms/x360airuntest
Top comments (1)
nteresting approach. 👏 The self-healing part is especially useful for reducing the maintenance cost of UI tests.
One thing I'd be careful about, though, is the word "reliable". A test recovering from a broken selector proves that the agent found something it considers equivalent, but it doesn't necessarily prove that it interacted with the intended element or that the resulting behavior is correct.
That's where I think the evaluation layer becomes really important. 🧪 The browser action can be AI-driven, but the oracle should ideally be as deterministic and independent as possible: expected state, URL, DOM properties, API side effects, authorization boundaries, etc.
Otherwise there's an interesting failure mode: the AI successfully "heals" the test, the test passes, and we've simply automated the process of convincing ourselves that the UI still works. 😄
Self-healing is valuable. Self-verifying is the harder problem. 🔐