Skip to main content

How Wopee.io works

Wopee.io works in four stages: it crawls your web app, builds an app-context document, derives user stories, and generates test cases. Every artifact is editable before anything runs. The crawl takes 5–15 minutes. Test generation takes 10–30 seconds per test in Direct mode, or 5–10 minutes in Agent mode, where the AI runs the scenario live in a browser.

No credit card required

The 4-stage pipeline

From a URL to a reviewed test suite. You stay in control at every stage.

  1. 1

    Crawl

    5–15 minutes

    Point the agent at a single URL. It explores your web app autonomously, capturing screenshots and HTML along the way. No scripts, no selectors.

  2. 2

    App context

    1–2 minutes

    The agent turns what it found into an app-context document describing your application. Edit it or regenerate it until it matches reality.

  3. 3

    User stories

    1–2 minutes

    From the app context, the agent derives user stories. Review, edit, delete, or add your own before a single test is generated.

  4. 4

    Test cases

    10–30 s or 5–10 min per test

    Each user story becomes a test case. Direct mode writes it in 10–30 seconds. Agent mode runs the scenario live in a browser first, taking 5–10 minutes per test.

Every artifact (the app-context document, the user stories, the test code) is editable and regeneratable. Nothing is a black box. Realistic time to a first generated suite: about 10–30 minutes, not seconds.

Supervised autonomy: the interactive cockpit

Autonomous does not mean unsupervised. In the Wopee.io cockpit you watch the agent work in a live-streamed browser, take over control mid-run, steer it with a short note, and answer the multiple-choice questions it raises when a decision is yours to make. When the agent hits something it cannot pass on its own, like a CAPTCHA, multi-factor authentication, or missing credentials, it asks for your help, hands you the browser, and resumes autonomously once you are done.

Your tests stay yours

Sign in with GitHub, GitLab, or Google. Tests live in a Git repository you can open, diff and roll back. Passing tests can also be generated as Playwright code that uses the Wopee.io SDK.

No credit card required

Video thumbnail: workshop on building a first AI testing agent with Playwright and Wopee.io

See it in action

Follow a full workshop from a prompt and a URL to generated tests, agent runs and a Playwright pull request on GitHub.

Watch the 41:55 video →

Guides for your stack

All guides →

Guides for the framework you build on, the industry you ship into, and the testing job in front of you.