Find bugs in your app without writing tests

Tests only check what someone thought to check. The login works, the checkout works, the settings screen saves. The bugs that reach users are usually somewhere else: a button on a screen nobody tested, an error message that only shows on an empty account, a sheet with no way to close it.

Explore covers that gap. You give it a device and some time, and it uses your app the way someone who has never seen it would, without a script. Afterwards you read a short list of findings instead of a long test log. Nothing becomes a test unless you choose it.

Start an exploration

Open Create test on the Tests page and switch to the Explore tab, or click Explore this app on the app’s page. Then choose three things:

Click Start exploring and watch it work, or go do something else and come back to the results.

What comes back

Each finding is one of five kinds:

• Crash: the app closed, went blank, or something else took over the screen.
• Error screen: the app showed an error message or an error page.
• Broken element: a control that does nothing, text that is cut off or overlapping, an empty placeholder.
• Can’t leave the screen: the AI tried several ways out, Back included, and nothing changed.
• Error in the app’s log: your app logged an error. The lines appear with the finding, with secret values masked.

The list stays short. A problem the AI runs into again is shown once, with a count (“Seen 3 times”), and permission prompts and system alerts are not reported at all.

Show me on a finding plays the run’s video from just before it happened, so you see how the AI got there as well as what went wrong.

Turn a finding into a test

Once you fix a bug, you want a test that catches it if it comes back. Save as test builds that test from the finding. It opens Review steps with one step for each thing the AI did on the way to the problem, then a Verify step describing what a working app would show. Edit anything you like, and nothing is saved until you click Save.

The new test starts with a head start. Explore already found the route, so the steps you leave as they are replay that route on the next run without calling the AI. Only steps you change are worked out by the AI again, and once a run passes, VibeView saves those steps too (how that works).

It behaves like a careful tester

The AI follows a few firm rules:

• It stays in your app, and never buys anything, deletes data or messages real people.
• It never makes up passwords, emails, card numbers or codes. If your app needs a sign-in, store the account as test data. The AI types it without seeing it: a secret reaches it only as a name such as {{password}}.
• It reads your app notes, so a sentence like “Use the demo account; the Pro tab needs a subscription” changes where it goes.
• On a TV it moves focus with the remote, the way a viewer would.

What it costs

Explore uses the AI on every turn, so it needs usage credit or your own AI provider key. It is not covered by the Free plan’s free test. With VibeView-provided AI, the run page shows the usage credit the run actually used. With your own key, VibeView charges nothing for the AI. The device time counts toward your streaming minutes like any session. See Usage credits.

A run stops when its time is up, when it reaches the action limit for that length, when the AI has covered everything it can reach, or when you click Cancel. A spending safety limit, based on the model’s price, also stops a run that gets stuck in a loop. Whatever it found before stopping is kept.

Good to know

• Explore is in beta. Treat a finding as a lead, and watch the clip before you file it.
• Runs are kept for 30 days. A test you saved from a finding stays.
• Screenshots are not edited. If a secret was visible on screen, it is visible in the screenshot.
• Explore runs don’t affect your test health. They appear in the Runs list marked Explore, and never count toward pass rates, flaky tests or Needs attention.

Frequently asked questions

Does Explore save tests on its own?

No. Each finding has Save as test, and nothing is saved until you review the steps and click Save.

Can it explore a TV app?

Yes, on Apple TV, Android TV and Roku. The AI moves focus with the remote, and only reports a screen it can’t leave after it has tried Back.

Who can start an exploration?

Anyone in your organization with the Developer or Admin role.

What if it runs out of things to do?

The run ends early once the AI has covered what it can reach. The findings so far are kept.