Relay AI is generally available on Pro and Enterprise plans. It analyzes failed runs, surfaces flaky tests, and suggests pipeline changes you can apply in one click.

Intelligent test selection

When you push to a pull request, Relay AI maps changed files to test suites and skips work that cannot be affected by the diff. Teams in beta cut median CI time by 34% without dropping coverage gates.

You stay in control: accept suggestions per repo, per branch, or per pipeline stage.

Flaky test detection

Relay AI tracks tests that pass and fail on identical commits, then groups them by root cause — timing, shared fixtures, runner drift, or external dependencies.

Failed runs now include a Flaky badge with a link to quarantine or retry policy recommendations.

Pipeline optimization

After enough runs, Relay AI proposes concrete diffs: parallelize slow jobs, split oversized steps, or reorder cache warm-up. Every suggestion links to the run data that produced it.

Getting started

Enable Relay AI from Settings → Automation on any Pro workspace. It starts in suggest-only mode so you can review recommendations before they affect live pipelines.