If the files are small and the process can be set up to do it, I will bring everything in in one reader and use Group By, but this sometimes is too taxing on the machine or doesn't work at all.
My only other option, then, is to write a batch file which invokes fme.exe multiple times, passing in different inputs and outputs. But writing these batch files is fiddly and time-consuming.
Is there a better way?
Have you tried using a Directory and File Pathnames reader in conjunction with the WorkspaceRunner transformer?
I'm assuming you have one reader for your Large Process workbench. First, convert all your parameters to private. Then create one new published parameter, and call it something like Filepath.
Second, go to the Navigator, and open the Advanced Options portion of the reader. Select Feature Types to Read, click the drop down arrow, and link it to the Filepath parameter.
Third, create a new workbench. Put a Filepath and Directory reader into it, and select the folder where you store your many files that need to be processed. Connect that reader to a WorkspaceRunner transformer, point that at your Large Process workbench, and feed the path_windows attribute to the published parameter.