Check That Your Website's Key Journeys Work

How the testing works

1. 1

   It explores

   An AI model walks through up to 25 pages of your application and builds a map of what is there. If you added a login, it signs in first and proves it worked with a screenshot.
2. 2

   It writes test cases

   From what it found, it writes cases for the features of your application, such as searching, filtering, submitting a form or updating a profile.
3. 3

   It runs them

   Each case runs in a real Chromium browser, with a screenshot at each step. No CSS selector is stored, so a small design change does not break a test.
4. 4

   It decides and explains

   Each case gets a verdict. Failures become findings, written in plain language with the evidence and a suggested fix.

What it can find at each level

Examples from real checks

A site served from two servers can answer with a page from one and fail to supply the stylesheet from the other, so some visitors see an unstyled page while everything looks fine to the owner. Provelex reports this as a page asking for files it did not receive, because that shows up in the browser's network log rather than the page itself.

Another common case: a button disables itself and sends someone to a payment page. They press Back, the browser restores the page exactly as it was, and the button is still disabled. Nothing is broken underneath, and developers rarely see it because they tend to reload rather than go back.

What to keep in mind

• AI-written tests cover what the model chose to try. They will not cover every path, and two runs will not be identical.
• It is not a link checker: it does not visit every link on your site.
• CAPTCHA and one-time codes can block automated sign-in, though you can sign in yourself in a live browser instead.
• Check a failed test against the site before you treat it as a defect. Complex journeys and business rules still need a person.