BookNest stays on Airflow 129 . Its pipeline is a fixed daily sequence; Cosmos already renders the dbt 37,942 project per model (Orchestrating dbt with Cosmos); AWS 24 , Google Cloud 1 and Astronomer 149,584 sell managed Airflow; and new data engineers are likelier to know it. Prefect 74,615 's purchase of Dagster 177,056 Labs (Dagster) keeps both products, so the other two would still win under different facts:
Choose Dagster when the questions are about data rather than jobs: many teams sharing tables, partitions and freshness to track, or a dbt project large enough that lineage and per-model checks matter most.
Choose Prefect when pipelines are Python programs with dynamic shapes (one run per uploaded file, branches decided by results), or a small team wants the lightest server.
Stay with Airflow when you rely on its providers, on managed offerings, or on DAGs that many people already maintain.
Whatever you choose, keep business logic in plain, testable functions like booknest_pipeline.py: moving BookNest between three orchestrators then took a few hundred lines of glue, not a rewrite.