Kestra

Airflow 2 reaches end of life in April 2026. Most teams are treating it as a version upgrade. It's not. It's an architectural decision that will shape who can own workflows and how failures get diagnosed for years to come.

Whether you're on open-source Airflow, Amazon MWAA, Astronomer, or Cloud Composer, the underlying model is the same: orchestration lives inside executable Python. That means only Python developers who know Airflow internals can own workflows, and what works at 50 DAGs starts to strain at 500. Airflow 3 improves the ergonomics, but the model is the same.

Before you commit engineering resources to an upgrade or a migration, it's worth understanding where that model holds up and where it doesn't.

Benoit and Will will walk through where Airflow's architecture holds up and where it strains, compare it against a declarative, language-agnostic orchestrator like Kestra, and demo AI-assisted DAG migration from Python to YAML in minutes.
