Latent is an AI-powered QA platform that turns plain English into reliable end‑to‑end tests. Use it to create projects, write tests in natural language, run them in the cloud, and watch live execution with action logs, screenshots, and results — all from your browser.

Quick links

• Getting Started
• Using the Dashboard
• GitHub Integration
• FAQ

What you can do

• Create a project for your app or site
• Add tests by describing the user journey in natural language
• Run tests and view real‑time execution (streamed step‑by‑step)
• Track status, duration, pass/fail and recent activity
• Integrate with GitHub Actions to run tests on every push/PR

Tip: The clearer your test goal, the better the results.

1. Sign up or log in

   • Visit latentqa.com and create an account or log in.

2. Create your first project

   • Click “New Project”, give it a name, and save.

3. Create your first test

   • Inside the project, click “New Test”.
   • Provide a testing URL (e.g., your staging site) and describe the goal/test-flow in natural language, e.g.: “Login with user X, add item Y to cart, and check out.”

4. Run the test and watch live execution

   • Click “Run”. You’ll see step-by-step actions, and screenshots streamed in real time.
   • When the run completes, you’ll see pass/fail, duration, and details.

Next steps

• Learn the dashboard: Using the Dashboard
• Integrate with CI/CD: GitHub Integration
• Troubleshooting: FAQ

Projects:

• Create and manage projects for each app/site you test
• View aggregate stats like total tests and recent activity

Tests

• Create tests by entering a testing URL and a natural-language goal
• Edit tests to refine goals or flows
• Delete tests you no longer need

Running tests

• Click “Run” on a test to start an execution
• Live view shows each step, with action logs and screenshots
• Final status includes success/failure, execution time, and captured details

Plans and API keys

• Free plan: requires your own Gemini API key in Settings
• Pro plan: can use hosted AI or your own API key

Tips

• Be explicit in your test goal; include field names, button labels, and expected outcomes
• Use your staging URL for stable results

Related: Getting Started, GitHub Integration

Run AI-powered tests automatically on every push or pull request.

Add repository secrets (GitHub Repo → Settings → Secrets and variables → Actions):

• LATENT_API_KEY: your Latent API key (from Integrations)
• LATENT_PROJECT_ID: your linked project
• STAGING_URL: optional, override URL to test (otherwise uses test’s configured URL)

In your GitHub repository, create a new file at .github/workflows/latent-ci.yml and paste the YML below:

name: Run Tests

on:
  push:
    branches: [ main, develop ]
  pull_request:
    branches: [ main ]
  workflow_dispatch:

jobs:
  test:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Run Latent Tests
        id: latent
        uses: latent-qa/latent-ci@v1.0
        with:
          api_key: ${{ secrets.LATENT_API_KEY }}
          project_id: ${{ secrets.LATENT_PROJECT_ID }}
          website_url: ${{ secrets.STAGING_URL }}  # Optional

      - name: Display test results
        run: |
          echo "Tests passed: ${{ steps.latent.outputs.passed }}"
          echo "Tests failed: ${{ steps.latent.outputs.failed }}"
          echo "Total tests: ${{ steps.latent.outputs.total }}"
          echo "Execution time: ${{ steps.latent.outputs.execution_time }}s"

      - name: Update status check
        if: always()
        run: |
          if [ "${{ steps.latent.outputs.failed }}" -eq "0" ]; then
            echo "All tests passed! ✅"
          else
            echo "${{ steps.latent.outputs.failed }} tests failed ❌"
          fi

How it works

• You write tests in simple natural language directly in the Latent dashboard (no code needed).

• On every push, pull request, or manual dispatch, this workflow calls Latent to run all active tests for your linked project.

• Latent executes tests in the cloud and streams results; the action collects a pass/fail summary as outputs.

• Optionally set STAGING_URL to override the testing URL per CI run; otherwise, the test’s configured URL is used.

• Your pipeline gets immediate feedback if anything broke or if all tests passed, enabling you to gate merges with status checks.

Getting started

• How do I start?
  • See Getting Started to sign up, create a project, add a test, and run it.

Running tests

• Why does my run stop immediately?
  • Ensure you are logged in; runs require an authenticated session.
  • Check that your test has a valid URL and a clear, actionable goal.
  • If you are on a free plan add your API key.

Plans and keys

• Do I need my own API key?
  • Free plan requires your own Gemini API key (add it in Settings). Pro users can use our hosted AI systems or their own key.

CI/CD

• How do I run tests in CI?
  • Set up the GitHub Integration and add repository secrets. On push/PR, we run your active tests and return pass/fail.

Results and analytics

• Where can I see coverage and results?
  • The dashboard shows recent runs, pass/fail counts, durations, and activity.

General

• So… what even is Latent?

  • Think of Latent as your AI QA buddy. It writes, runs, and fixes test cases for you so you don’t waste hours clicking around or debugging flaky tests.

• How is this different from Selenium/Playwright?

  • Instead of writing complex code, you describe tests in English. Instead of debugging cryptic failures, you watch Latent work in real-time and override anything visually.

• How fast do tests actually run?

  • First run: Agent figures out the flow. Every run after: Lightning fast using saved execution paths—no compilation o interpretation delays.

• Will it work with my current workflow?

  • Yes. Latent plugs into CI/CD (e.g., GitHub Actions) with a simple workflow file.

• Do I need to know how to code?

  • No. Describe flows in plain English.

• Do I need to re-write test flows when the app updates?

  • Not usually. Latent adapts to UI/screen changes; more detailed goals yield better results.

• How fast can I start using it?

  • Instantly. Sign up, add a project, write a test, and run.

• What if I get stuck?

  • We’re here to help 24/7.

• Is there a free trial?

  • Yes — there’s a free plan with no credit card required.

More help

• Website: latentqa.com