Skip to main content

Scheduled runs

Scheduled runs execute an application — or an evaluation — automatically on a recurring schedule, with no one in the loop. Use them for daily/weekly digests, periodic data refreshes, recurring report generation, or regularly re-running an evaluation suite to catch quality regressions.

What you can schedule​

  • Applications — a published Tool, Agent, or Voice Agent runs on the schedule with a fixed set of inputs.
  • Evaluations — an evaluation suite runs on the schedule so you get an ongoing quality signal over time rather than a one-off score.

How it works​

  1. Pick the published application (or evaluation) to schedule.
  2. Define the cron-based schedule (frequency and time).
  3. Provide the inputs the scheduled run should use.
  4. Each firing executes like any other run — consuming credits, applying guardrails, and producing a trace in Run history — so you can review what every scheduled run did.
  • Bulk Run — combine with scheduling to process a batch on a cadence.
  • Evaluation — the suites you can run on a schedule.