A data team's first pipeline is often a few scripts on a timer: export the sales at 1:00, load them at 1:30, build the report at 2:00, and hope each step finished before the next began. Apache Airflow replaces the hope. You describe the work as a Dag (a directed acyclic graph): a set of tasks and the order they run in. Airflow's scheduler starts a Dag run when the schedule says so, runs each task once its upstream tasks have succeeded, retries what fails and keeps a record of every attempt.
Each task in each run is a task instance, and the Grid shows them as squares: one column per run, one row per task, colored by state.
This workspace simulates Airflow 3.3 in your browser: a Python editor on the left, and on the right the Dags list, Grid, Graph, Assets, Admin and a terminal with the airflow command line. New Dags start paused, as they do in Airflow, until you switch them on.
A daily Dag for Marigold Grocers, a fictional grocery chain: two tasks, and the second reads what the first returns.