From repo connect to first AI-managed pipeline in under 15 minutes.

No. SuperPlane wraps around your existing CI runner — GitHub Actions, Buildkite, CircleCI, Jenkins. It adds the intelligence layer on top. Your current runner still executes the jobs. SuperPlane decides which jobs to run, which environment to provision, and when to promote or roll back. You don't migrate away from what you have.

SuperPlane never reads your source code. It reads commit metadata (file paths changed, author, branch) and test result history (pass/fail per test per run). From that, it builds a dependency graph: which file changes tend to trigger which test failures. The model is trained on your own pipeline history — no source code, no AST, no static analysis. You opt into a 30-day history window in your policy.

Every rollback decision is logged with its full rationale — the specific metric that crossed your threshold, the threshold value, and the timestamp. You can review, override, and re-deploy via the CLI or dashboard. If SuperPlane rolled back incorrectly (e.g., a flaky external dependency caused a transient spike), you can mark the run and SuperPlane adjusts its model for that signal. Decisions are auditable and can be exported.

Use superplane policy validate to lint your policy file locally before committing. superplane run dry-run shows you exactly what SuperPlane would do on a given commit without executing anything. The run log in the dashboard explains every AI decision in plain language — not just what was done, but why, with the specific input data that drove the decision.

No. SuperPlane never stores source code. The only data that passes through SuperPlane's systems is: commit metadata (SHA, file paths, author), test result summaries (pass/fail counts per suite, not test body content), and pipeline event data (timing, env configuration). All tokens are encrypted at rest. See our Security page for the full data handling model.