I need to think about generating a useful set of canned comments based on operations that we've seen over the years, and that would be useful to distinguish between NCBI-destined comments vs. purely 'notes to self'.

This should also include training/documentation once it's done.