Yes — that’s perfectly fine in Prefect 3.x. A flow can contain all of your logic with no tasks at all. - What you keep: scheduling/deployments, parameters, flow-level retries/timeouts, logging, artifacts, and (if configured) flow result persistence/caching. - What you lose: fine‑grained observability and control (no task-level retries/caching/concurrency or parallelism managed by a task runner), and the UI will just show the single flow run node. Example:

from prefect import flow, get_run_logger

@flow(retries=2, retry_delay_seconds=10, log_prints=True)
def main(x: int = 1) -> int:
    logger = get_run_logger()
    <http://logger.info|logger.info>("Starting flow-only run")
    y = x * 2  # all your logic here
    if y % 3 == 0:
        logger.warning("Multiple of 3!")
    return y

if __name__ == "__main__":
    main()

If you later want granular retries, caching, or parallelism for specific steps, you can refactor those pieces into `@task`s while keeping the rest of the logic in the flow.