In Prefect 3.x, the simplest fire-and-forget pattern is to trigger a deployment with

timeout=0

.

python
from prefect import flow
from prefect.deployments import run_deployment

@flow
def parent_flow():
    flow_run = run_deployment(
        name="child-flow/my-deployment",
        parameters={"x": 1},
        timeout=0,
    )

    print(f"Triggered child flow run: {flow_run.id}")
    print("Continuing immediately...")

What this does - Starts an independent flow run from a deployment - Returns immediately - Does not wait for completion - Gives you a

FlowRun

object back, so you can keep the run ID if you want to inspect it later Important note -

run_deployment()

does require a deployment - A worker must be running for that deployment’s work pool If you call a flow directly like this:

python
child_flow()

that is not fire-and-forget — it behaves like a nested subflow and the parent waits for it. If you want the child to appear as related to the parent in the UI,

run_deployment()

does that by default when called from inside a flow. If you want it completely detached, you can set:

python
run_deployment(
    name="child-flow/my-deployment",
    timeout=0,
    as_subflow=False,
)

Recommendation - Use direct flow calls for true subflows when you want to wait - Use

run_deployment(..., timeout=0)

when you want background / independent execution Relevant docs: - Run deployments - Deployments If you want, I can also show: 1. how to deploy the child flow, or 2. how to trigger it asynchronously from async code