User guideAI Engineering Platform
Back to siteSite

Analytics

Analytics answers two questions: how fast is delivery, and is the quality improving?

Analytics

Delivery and outcome metrics for the selected period.

Metrics

MetricWhat it means
Lead timeFrom ready task to merged change
Approval timeTime spent waiting for human decisions
OnboardingTime from project creation to the first accepted task
Run successCompleted runs divided by all runs
Cost per accepted changeProvider cost divided by merged changes
First-pass mergePRs merged without a rework cycle
Proof validityAcceptance proofs that referenced real evidence
Runs per taskHow much agent work a task needed
Change failure rateMerged changes that required a fix
Escaped defectsDefects found after merge, per 100 merges

Metrics marked measured come from recorded events. Metrics marked na have no data yet; the platform does not invent a value.

Filtering

Use the period selector to compare delivery windows. Project filters narrow the view to one product or client.

Reading the numbers

  • A short lead time with a high rework rate is not an improvement.
  • A high proof validity rate with a low first-pass merge rate suggests the plan or context needs work.
  • Approval time is often the largest lever; it is a human process metric, not an agent metric.

Exports

The API exposes the same analytics data. Use it for a dashboard or a weekly report. Large exports should page through the API rather than scrape the UI.

Next