Your First Test

Creating tests in tapioka.ai is as easy as describing a user journey. Our AI handles the technical execution so you can focus on the business logic.

1. Create a Project

The first step is to create a Project and select your platform (Android, iOS, or Browser).

  • Important: The platform cannot be changed later, so ensure you select the correct environment for your tests.

2. Create a Test Suite

Inside your project, create a Test Suite. Suites are logical groupings of test cases, typically organized by:

  • Feature: (e.g., "Profile Tests", "Checkout Flow")
  • Test Depth: (e.g., "Smoke Tests", "Regression Pack")

3. Create a Test Case

Now you can add individual test cases to your suite. You have three options:

  • Manual Definition: Write steps directly in the dashboard using natural language.
  • Copy & Iterate: Reuse steps from existing test cases to build complex scenarios.
  • File Upload: Import multiple tests at once using files exported from tools like Zephyr.

Writing Instructions

Use plain English to define your test steps and expected results, for example:

  • Scenario: "Go to your account, go to settings"
  • Expected Result: "Information about billing is visible"

For more complex scenarios and formatting tips, see our detailed guide on Writing Instructions.

Preconditions

To streamline your testing, any test case can be marked as a Precondition.

  • Usage: Preconditions are typically used for common setup logic, such as logging in or skipping registration flows.
  • Execution: Once set as a precondition for a specific test, it will run automatically before that test starts.
  • Standalone: Tests marked as preconditions cannot be run as standalone tests; they only execute as part of the setup for other test cases.
  • Repetition: A precondition will run for every test execution it is assigned to.

4. Running the Test

To execute your test:

  1. Select Tests: Choose the specific test cases or suites you want to run.
  2. Setup Environment:
    • For Android/iOS: Select or upload the application artifact (.apk or .ipa).
    • For Browser: Select the target environment/URL.
  3. Execution Mode: Choose to run the test immediately or schedule it for a later time.

Standalone Execution & Clean State

Every test is executed as a standalone run. To ensure reliability and avoid test pollution, we reset the app state (cleaning cache and data) between each test. This means every test starts from a totally clean, predictable state.

Execution Phases

The test execution happens in two distinct phases:

  • Learn Phase: When you first run a test, the AI agent explores the UI to identify elements and valid interaction paths. It proposes actions based on your instructions (e.g., "I clicked element '24' because it matches 'Back Button'").
  • Run Phase: Once a test is "Learned", it becomes a stable part of your regression suite, achieving a high success rate by reusing the knowledge gained during the learning phase.

5. Reviewing Results

After execution, you can review the results in the Runs section. The dashboard provides visual evidence for every step:

  • Screenshots: Captured "Before Action" with interactable elements automatically numbered and highlighted.
  • Detailed Logs: Step-by-step technical logs of AI decisions.
  • Video Playback: Watch the full agent session to understand exactly what happened.