A data team's first pipeline is often a few scripts on a timer: load the orders at 6:00, build the report at 6:30, and hope the first one finished. An orchestrator replaces the hope. It knows the report depends on the orders, runs them in that order, retries what fails, keeps a record of every run and tells you when something is late or broken.
Dagster organizes that work around assets: the tables, files and models your pipeline produces. You declare each asset as a Python function, Dagster reads the dependencies between them, and the asset graph shows the whole pipeline. To materialize an asset is to run its function and store what it returns.
This workspace runs a simulation of Dagster 1.13 in your browser: a Python editor on the left, and on the right the asset graph, runs, jobs, automation and a terminal with the dg command line.
Three assets for Juniper & Pine, a fictional shop. Each parameter names the asset it reads, so Dagster runs them left to right.