I did it in a class once, and I know some people that do it routinely, but this idea of colony PCR seems unnecessary, at least for the things I do. Putting a gene insert into a plasmid, transforming that plasmid in to competent cells, expressing proteins.
The things I have read suggest that you would perform colony PCR to see if a particular colony contains the plasmid you are looking for, OR to see if your cloning actually worked, by checking to see if the plasmid size on an agarose gel matches what you would expect for correctly ligated plasmid/insert product.
For the first step, I don’t really see necessity, particularly due to the usage of antibiotic resistance marker. If you transform in the evening on an amp plate and the plasmid confers amp resistance, doesn’t the presence of colony in the morning virtually guarantee the presence of plasmid within that colony’s cells, especially if you are going to transfer to liquid cultures right away in the morning?
The second purpose could be of merit, especially if you are not using ligation-independent methods, or if the insert is very large…
Cloning isn’t as easy as it sounds. When ligating an insert to a plasmid, some plasmids will self-ligate (when you have blund-ended ligation, TA-cloning, non-dephosphorylated ends…) and thus plasmids without an insert will also be transformed into the bacteria. And sometimes you get “junk” ligated into your vector. Thus, by doing a colony PCR you will find out if the colony is a “good” transformant, i.e. carrying the plasmids + insert (then you get a PCR product of the expected size), or if it is a “bad” transformant, i.e. carrying plasmid only (small PCR product or no product at all depending on the primers) or plasmid + junk insert (band of unexpected size).