Build a pipeline

Work from an output

A pipeline transforms source data into an output used by other work. Output and run-history behavior depend on the selected execution path. Begin with the output schema and the source fields it needs. A visual preview is a sample for inspection; it is not a completed build or proof that every row will pass.

1. Open Pipeline Builder and select or create a pipeline in the right workspace.
2. Choose an input dataset or connection you are authorized to read.
3. Add transforms, inspect schema changes, and preview representative output.
4. Save the definition, then use its available run or build action. Inspect execution status and the resulting output separately before using it downstream.

When a run fails

• Read the failing node and error before retrying; repeated runs with the same invalid input will not correct the data.
• Check whether the source schema or credentials changed since the last successful run.
• Use the last successful output only with its recorded time and version. Do not describe it as the latest source state.