When should I keep using Airflow instead of migrating to Catalyst?

You should keep using Airflow instead of migrating to Catalyst for your core scheduled batch pipeline workloads. Airflow’s core design is tailored to static, repeatable batch data orchestration, with deterministic, relatively short-lived runs, aligning better with these standard workflow needs than agentic durable execution frameworks. Publicly available documentation does not currently list native support for probabilistic long-lived agent workloads, so stick with Airflow for your existing familiar core batch workflows.