Skip to content
TestPerch
Sign in

CI time and costs skyrocketing due to AI?
Solve both with TestPerch.

TestPerch uses AI to select the Playwright tests that matter for each pull request. Get faster feedback and spend less on CI.

14 days free

TestPerch customers have saved

Hours of CI time
657.8
Estimated CI costs
$868

* CI $ savings assume a GitHub Actions Linux x64 8-core runner at $0.022/minute (USD). Actual CI bills vary. Refreshed daily.

More tests.
Less waiting.

AI is writing more code. Teams are opening more PRs. And e2e suites keep growing. Running the whole suite on every change makes the feedback loop slower and more expensive.

TestPerch uses AI to choose which Playwright specs to run for each PR, so developers and coding agents can get back to their next change sooner.

  1. Bring your existing suite.

    Add the TestPerch reporter to Playwright and connect GitHub. Your tests stay in your repository and run in your CI.

    pnpm add -D @testperch/playwright
  2. Every PR runs only the tests that matter.

    TestPerch uses the diff to choose relevant specs.

  3. Flaky tests?

    When a build fails, re-run only flaky tests instead of wasting time on tests that already passed.

  4. Broken tests?

    Put a test into quarantine without having to spin up a new PR for it. No more "rebase please".

Improve your suite.

Find the flakes wasting your time.

Surface recurring flaky tests, understand their history, and quarantine them when needed. Keep noisy tests from repeatedly interrupting the team.

Give agents the evidence to fix failures.

Make test attempts, steps, logs, and attachments available through MCP. Your coding harness gets the debugging context it needs to investigate and fix tests.

See whether your suite is getting healthier.

Follow flaky behavior, test durations, and results over time. Know where the suite is improving and where to focus next.

From failure to contextExample
Flakycheckout.spec.ts

applies a discount code

Attempt 1
Failed
Attempt 2
Passed
expect(total).toHaveText("$90.00")

Received: "$100.00"
Timeout: 5,000ms

The result is only the start.

  • Steps & logs
  • Traces & screenshots
  • Previous attempts
  • Test history
Available to your coding agent via MCP

Where you already work.

MCP
Connect your coding harness to test results and debugging evidence. Give agents context without the copy-paste.
GitHub
See run results and the rationale behind AI test selection alongside the pull request that triggered them.
Slack
Investigate test results with your team in Slack. Quarantine a flaky test via a Slack mention.

Less CI spend.
Simple pricing.

One workspace subscription.
Unlimited projects and team members.
Every feature.

No credit card required.

$20/ month
Test attempts included
40,000 / month ($0.50 / 1,000 after)
Attachment storage included
20 GB ($0.20 / GB-month after)
AI usage
Billed by usage

USD. Taxes may apply. Additional test attempts round up to 1,000 attempts.

Estimate your monthly cost

Let your coding agent estimate monthly costs from the last seven days of CI runs. Paste this prompt into an agent with access to your repo and CI.

A few more details.

Do I have to change my tests or my CI setup/provider?

Keep your existing tests and CI provider. Add the TestPerch reporter to your Playwright configuration and give your CI job the ingestion credentials.

How does pricing work?

A workspace costs $20 per month and includes test attempts and attachment storage. You pay for usage above those allowances and for AI usage. Each retry counts as another test attempt.

Cuts on CI time, pays for itself, a no-brainer.