Bulk & Scheduled Execution

Most of the platform is built around running a Tool, Agent, or Voice Agent one invocation at a time — from the playground, an API call, or a chat surface. This section covers the three ways to run applications at scale or unattended:

• Bulk Run — process a whole batch of inputs through an application in one job, with resilience against pod restarts and deployments.
• Event-based execution — fire an application asynchronously in response to an event or long-running workflow, and receive the result via callback.
• Scheduled runs — cron-based recurring runs for both applications and evaluations.

All three reuse the same execution engine, credits model, guardrails, and tracing as single runs — so anything you can observe or govern on a single run also applies here.