Tests

What it is

Tests define individual test scenarios for your AI assistants. Each test specifies what to check, including conversation flows, expected behaviors, and evaluation criteria.

How to use

1. Navigate to Testing → Tests in the sidebar
2. View all tests across your organization
3. Click on a test to view details or edit
4. Create tests from within a Suite

Test Structure

Each test includes:

• Title: A descriptive name for the test
• Steps: The sequence of interactions and checks
• Evaluation criteria: What determines pass/fail
• Tags: Categories for organization

Test Steps

Tests can include various step types:

• User message: Send a message as the simulated user
• Assistant response: Expect and evaluate the assistant’s reply
• Action: Execute an external action (API call, etc.)
• Assertion: Check specific conditions

Key Features

• Organized by suites: Tests are grouped into suites for logical organization
• Folder support: Organize tests into folders within suites
• Reusable patterns: Share test configurations across scenarios
• Detailed results: See exactly what happened in each test run

Tips

• Write clear, specific test titles that describe the scenario
• Use assertions to verify specific response characteristics
• Organize related tests into folders within suites
• Include both happy path and edge case tests