WCAG 2.2 audits, on request

AI mobile app testing platform for iOS, Android and web

Every test, written and run by mini.

Real devices, simulators and browsers. Cross-app, GPS and multi-device flows included.

Book a demoSee a real report

The hard parts, handled.

  • No flaky tests.

    No selectors or coordinates. After a redesign, mini checks the same criteria.

  • Two phones. One test.

    Up to three phones on the same OS.

  • Out of the app and back.

    Email codes, magic links and Sign in with Google.

  • Anywhere on the map.

    A mocked location or route, iOS and Android.

  • mini speaks up before you ask.

    Flags issues no test asked about, and asks in Slack for accounts.

mini writes the suite.

Each test is a scenario and its acceptance criteria.

  • From what you already have.

    Your app, repo, Notion specs and test cases. Jira and Linear on request.

  • Outcomes, not clicks.

    mini finds the way through on every run.

  • Yours to change.

    Edit tests in plain words from the dashboard, Slack or your agent.

Real devices and simulators, in parallel.

The whole suite at once. Physical phones per workspace, on request.

  • Browsers too.

    Desktop Chrome and Firefox, Safari on iPhone, Chrome on Android.

  • Your build, or ours.

    Upload from CI, or mini builds each commit.

  • Test data, set up first.

    mini seeds accounts through your test endpoints.

Every finding comes with its proof.

One issue per failed criterion, with the recording, what mini expected and what it saw.

  • Ready for a developer.

    Steps to reproduce and a fix prompt for your agent.

  • Not a bug, said once.

    Dismiss with a reason; mini remembers.

  • Beyond pass and fail.

    UI/UX issues, plus iOS and Android WCAG 2.2 audits, on request.

Tested before it merges.

mini tests a pull request when you ask and comments with what broke.

  • Start it your way.

    Comment @mini test, or use the GitHub Action.

  • Other CI, coding agents too.

    Any CI via the CLI, coding agents via MCP.

  • In your Slack.

    Runs and findings land in your channel.

Ship a redesign. Keep the tests.

Tests say what must be true, not what to tap.

  • New features, new tests.

    mini adds, edits and retires tests as your code and specs change.

  • Every change listed.

    See every edit mini made, and revert any.

  • Works without GitHub.

    Start it from the CLI; your coding agent proposes the edits.

Every capability.

Teams that ship with mini.

“For the first time, I ran a pilot without a single technical bug. Testing my native iOS and Android apps in real conditions let me catch bugs upstream. Minitap saves me time and gives me confidence in every release.”
Sandra KhirecheFounder, Impakt
  • 2 days → 1 hourFull regression
  • 48 hoursTo full coverage
  • WeeklyReleases, from fortnightly

From a pilot customer.

  • Helmit
  • Memrise
  • Nanogram
  • Impakt
  • Trippin
  • MindStash
  • Clutch

Good questions.

Regression testingAccessibility testingPricingSecurity and enterprise
Which devices and browsers does mini run on?

iOS simulators and Android emulators in the cloud, plus physical iPhones and Android phones, per workspace on request. On the web: Chrome and Firefox on desktop, Safari on iPhone and Chrome on Android. Not tablets, watches, TV, Edge or desktop Safari.

What build do we hand mini?

An iOS simulator build or an Android APK that includes x86_64, uploaded from your CI or the CLI. Or connect your repository and mini builds each commit, monorepos included. Physical iPhones, switched on per workspace on request, install from TestFlight.

How long does a run take?

Tests run in parallel, so run time depends on the size of your suite and how many devices and browsers your plan runs at once. A full regression usually takes about an hour. On a GitHub pull request, mini can run only the tests the change affects.

An AI agent sounds flaky. How do we trust a pass?

mini’s tests have no selectors or coordinates to go stale: it reads the screen on every run and checks the same acceptance criteria. Each criterion carries its evidence: what mini saw, a screenshot, the moment in the recording and its confidence. Every test ends Passed, Warning, Failed or Unprocessable. If mini can’t grade a test, because the build broke or sign-in failed, it’s marked Unprocessable, never passed.

Will it block our merges?

Only if you ask it to. The GitHub check informs by default. Turn on blocking for an app and a failing test fails the check, so branch protection can require it. In the GitHub Action, wait-for-result and fail-on-failure hold a deploy.

We’re not on GitHub. Does it still work?

Yes. GitLab, self-hosted GitLab and Bitbucket connect your repository, so mini builds from it and writes the suite from your code. Pull request comments and checks are GitHub only; any other pipeline starts runs and reads verdicts through the minitest CLI.

Can we run it every night?

Yes, from your CI on a schedule, or by pushing a release tag. mini doesn’t schedule runs by itself yet.

What can’t mini test?

Face ID and fingerprint, NFC, Bluetooth, real SMS and calls, real payments, flows that finish in an external browser, tablets, watches and TV. Payments run in the store’s sandbox or with a test card.

Bring your hardest flow.

Watch mini run your build, and keep the report.

Book a demoSee a real report