Hello,
I can create summary from given text, using exiting training data. Or if you want to just detect keywords from given text, we might simple consider term frequency and inverse document frequency.
About your second point, find target audience from text. Do you have any previous dataset with which we can train a Machine Learning model for prediction?
Regards,
Samiran