Skip to main content

Overview

Ability Testing is a feature in Z360 that allows you to quickly verify how your AI agent responds throughout a conversation flow. Before deploying an ability to live interactions, you can simulate the entire flow to ensure the agent behaves as expected at each step. During ability testing, Act nodes execute in a mock environment rather than using actual tools and integrations. This means you can safely test your ability’s logic without creating real tickets, sending actual notifications, or triggering live workflows.

How to Test an Ability

Step 1: Open the Ability

  1. Navigate to AI Studio in your Z360 dashboard
  2. Go to the Abilities section
  3. Select the ability you want to test, or create a new one

Step 2: Access Ability Testing

  1. In the Ability configuration screen, locate the Test Ability button at the bottom of the Ability Details panel
  2. Click Test Ability to open the testing interface

Step 3: Start a Test Conversation

  1. The testing interface will open with a chat window titled Ability Testing
  2. The agent will automatically send an initial message based on your ability’s configuration
  3. Type a message in the input field at the bottom to simulate a user interaction

Step 4: Interact with the Flow

  1. Respond to the agent’s prompts as a user would during a real conversation
  2. The agent will follow your ability’s flow, asking for required information and processing responses
  3. Use the Show Agent’s Thinking button to view the agent’s reasoning at each step

Step 5: Review Agent Responses

As you interact with the agent, observe how it:
  • Collects information from the user
  • Validates inputs according to your flow rules
  • Moves through different nodes in your ability
  • Generates appropriate responses at each stage

Step 6: Clear and Restart

Click Clear History in the top right corner to reset the conversation and start a fresh test from the beginning.

Understanding the Test Environment

Mock Execution

When testing abilities, Act nodes (such as creating tickets, sending notifications, or updating records) are simulated rather than executed against real systems. This allows you to:
  • Test the complete flow without side effects
  • Verify the agent’s decision-making logic
  • Check response quality and formatting
  • Identify issues before going live

Channel Selection

Use the dropdown menu (e.g., “Email”) in the testing interface to simulate how the ability would behave across different communication channels.

Tips

  • Test multiple scenarios: Try different user responses to ensure your ability handles edge cases gracefully
  • Review agent thinking: Use the “Show Agent’s Thinking” feature to understand the agent’s decision process and identify any logic issues
  • Test required fields: Ensure the agent properly collects all required information before proceeding to Act nodes
  • Iterate and refine: Use testing results to adjust your flow, improve prompts, and enhance the user experience