Skip to main content
Once a workflow is published, operating it means three jobs: starting runs from the events that should start them, governing what each run may do and who must approve it, and watching what runs cost, produce and record. Each guide below covers one part. For a workflow you have not built yet, start with Build.

Start runs

Scheduling

Cron and one-shot schedules, missed fires, pause and resume.

Ingress

Let an outside system start a run by calling a signed URL.

Email

Send mail, and start a workflow from mail Forge receives.

Slack

Install and operate the Slack gateway.

Factories

Run an autonomous production unit with its own manager and budget.

Govern and approve

Operate Trust Fabric

Decisions, agent trust and the emergency stops.

Approvals and HITL

Human tasks: how holds and requests reach people and resume work.

Budgets and cost

Envelope and org-unit budgets and the spend they track.

Watch and notify

Observability

Traces, live views and analytics for running work.

Audit

The append-only audit trail and a run’s audit packet.

Webhooks

Signed event deliveries to your endpoints, with retry and replay.