A pipeline is a series of modules executed in sequence that results in final output documents. A given Wyam configuration can have multiple pipelines which are executed in sequence, and subsequent pipelines have access to the documents from the previous pipelines.
Conceptually, a simple pipeline looks like:
In the visualization above, the first module may have read some files (in this case 2 files) and stuck some information about those files such as name and path in the document metadata. Then the second module may have transformed the files (for example, from Markdown to HTML).
It's not unusual for a real-world generation to contain many different pipelines. Many times this is helpful if you need to reuse the output from one of the pipelines or want to separate the different generation steps.