How do you schedule Prefect flows in 2026?

As of April 2026, Prefect flows are scheduled by attaching a schedule (cron, interval, or RRule) to a deployment using `flow.serve()`, `prefect deploy`, or `prefect.yaml`. The Prefect API or local agent then triggers runs according to the schedule and routes them to a configured work pool.

Schedule Prefect Flows (2026)

Scheduling Prefect Flows

Prefect 3.x decouples flows (the code) from deployments (how, where, and when the code runs). Schedules attach to deployments.

Step 1 — Define the Flow

Decorate a Python function with @flow:

from prefect import flow

@flow
def daily_etl():
    extract()
    transform()
    load()

Step 2 — Create a Deployment With a Schedule

The fastest path is flow.serve(), which runs a long-lived process that schedules and executes runs locally:

daily_etl.serve(name="daily-etl", cron="0 6 * * *")

Three schedule types are supported:

Cron — Standard cron expression (e.g., "0 6 * * *")
Interval — Recurring every N seconds/minutes/hours
RRule — iCalendar RRULE (e.g., business-day-only)

Step 3 — Production Deployments

For production, use prefect deploy with a prefect.yaml file. The deployment registers with Prefect Cloud or self-hosted Prefect Server, and runs are dispatched to workers in a work pool (Process, Docker, Kubernetes, or ECS).

deployments:
  - name: daily-etl
    entrypoint: flows/etl.py:daily_etl
    work_pool:
      name: kubernetes-pool
    schedules:
      - cron: "0 6 * * *"
        timezone: "UTC"

Step 4 — Pause, Skip, and Backfill

Schedules can be paused via the UI or prefect deployment schedule pause. Backfills are triggered with prefect deployment run and an explicit start/end window.