Search

Training the AI

Functionize uses advanced machine learning models to keep your tests accurate as your application evolves. These models dynamically adapt to UI changes by analyzing thousands of attributes for each element, reducing the need for brittle selectors like XPath.

However, machine learning models occasionally make mistakes — for example, by selecting the wrong element, misinterpreting a UI state, or failing to adapt correctly to a subtle UI change. Train the AI is your feedback mechanism to correct these mistakes. By reporting errors, you contribute real-world data that improves Functionize’s accuracy over time, reducing false positives and ensuring test resilience.

When to Use Train the AI

  • A test interacted with the wrong element.
  • A click or input action failed unexpectedly.
  • The UI changed and self-healing didn’t resolve the issue.
  • The expected verification didn’t match reality due to model error.

By reporting these cases, you help the Functionize Data Science team refine the ML models with additional training data.

How to Use Train the AI

You can submit feedback in two different contexts: directly from the test in Slider View, or from Orchestration Results after execution.

Option 1: Report from Slider View

  1. Open a test in a Project.
  2. Click View Results (or the View button on any action) to open Slider View.
  3. Navigate to the action you want to report.
  4. Open the Action menu → select Train the AI.
  5. Choose the type of error that occurred (e.g., wrong element, timing issue).
  6. Choose the priority of the issue.
  7. (Optional) Provide additional comments or context.
  8. (Optional) Allow the Functionize team to rerun your test for verification.
  9. Click Submit.

Option 2: Report via Orchestration Results

  1. Open the Orchestration Results page.
  2. Find the execution you want → click the Actions button.
  3. Click Details to open the test execution.
  4. From the read-only Test Detail page, click View Results for step-by-step review in Slider View.
  5. Select the action you want to report.
  6. Open the Action menu → select Train the AI.
  7. Follow the same submission process as above.

Why Train the AI Matters

Every submission you make feeds into Functionize’s continuous learning loop.

  • Helps eliminate recurring false positives.
  • Strengthens element recognition models for your application.
  • Improves self-healing accuracy for future runs.
  • Reduces long-term maintenance effort.

Tip: Think of Train the AI as “teaching moments” — the more feedback you provide, the better Functionize learns your application and adapts without manual intervention.