A custom agent can do anything you write it to do. The vendor agents are more limited. This could include reading the log file to extract the PDF file name and writing it to the SQL database. Our "archive" process stores the original DAT file in a SQL database and then uses it to create PDFs as the archive.
Currenlty we have a HDLogger custom agent that writes all the Job information, this includes the path to the PDF file, index details like Creditor / Invoice Numbers, document type.
We then run a external scripted process that archives the Log information and the PDF to a database. I just thought it would be nice to have the log and the pdf files archive to the db by a custom agent... that way as soon as the job is printed / processed users can also find it in the db.
My only concern is the number of connections that will be opened to the DB at any one time...