Yes. In Prefect 3.x you can have one flow trigger/schedule runs of other flows (via their deployments) with custom parameters, and keep them independent (not subflows). Two good options: 1) High-level helper: run_deployment (recommended) - Use

as_subflow=False

to avoid linking as a subflow in the UI - Pass

parameters={...}

to override defaults - Use

scheduled_time=<datetime>

to schedule for the future - Use

timeout=0

for fire-and-forget (don’t wait) Example:

from datetime import datetime, timedelta
from prefect import flow
from prefect.deployments import run_deployment

@flow
def orchestrator():
    # Schedule another deployment for 2 hours from now with custom params
    run_deployment(
        name="target-flow/prod-deploy",
        parameters={"date": "2025-11-21", "priority": "high"},
        scheduled_time=datetime.utcnow() + timedelta(hours=2),
        as_subflow=False,   # keep it independent
        timeout=0           # return immediately
    )

2) Low-level client: create_flow_run_from_deployment - Gives full control; async only - If you don’t set

parent_task_run_id

, it will not be a subflow - Set initial state to

Scheduled(scheduled_time=...)

for future runs Example:

import asyncio
from datetime import datetime, timedelta
from prefect.client.orchestration import get_client
from prefect.states import Scheduled

async def trigger():
    async with get_client() as client:
        dep = await client.read_deployment_by_name("target-flow/prod-deploy")
        fr = await client.create_flow_run_from_deployment(
            deployment_id=dep.id,
            parameters={"date": "2025-11-21", "priority": "high"},
            state=Scheduled(scheduled_time=datetime.utcnow() + timedelta(hours=2)),
            name="orchestrated-run",
            tags=["orchestrated"],
        )
        return fr

asyncio.run(trigger())

Notes and tips: - The target flow must have a deployment defined (created via

flow.deploy(...).apply()

or

flow.serve(...)

). -

run_deployment

can also accept

flow_run_name

,

tags

,

job_variables

,

idempotency_key

, and

work_queue_name

. - If you do want to wait for completion, omit

timeout

or set it to a positive number;

timeout=0

is fire-and-forget. Docs: - Run deployments programmatically - Create and manage schedules - Chaining deployments with events (alternative pattern) Would you like an example tailored to your specific deployment names/parameters or scheduling pattern (e.g., a batch of future times)?