Do you mean a writer for each one (your question says reader)?
To create multiple writers is a simple task using a fanout tool.
Here I am writing each postcode from an address dataset to a separate GML file. So you could point all your data sources into a single feature type on a single writer and - by setting a fanout - create a separate writer/output for each at run time.
Probably the best way is to use fme_dataset or fme_feature_type as part of the fanout expression
Obviously , this does mean you need all of your source data tables to have the same attribute schema as defined in the one output feature type.
If that's not the case then you'll need to flick the switch and make it a dynamic translation:
Here I'm saying to write to a different output, but using the same schema as the original table had for everything (so if I had multiple source tables they would still get the correct schema)