I wasn’t involved in the decision to create an agent secretor, so I don’t know what the idea behind it was, but I’ll throw in my two cents:
I would think that before you can evolve a specialized part of your cell to secrete agents, you’d first evolve the mechanism of secretion, which would happen all over your surface until you see fit to evolve, say, a polarization mechanism to make the secretory vesicles congregate at one region. The only way for secretion to arise in a localized part of your membrane is for it to evolve by co-opting an already-localized mechanism to make it secrete stuff. For example:
Let’s say your cells are sessile and form monolayer colonies. Each cell is polarized with an apical side and a basal side. You perform pinocytosis on your apical side, so you have vesicles constantly budding off, fusing with endosomes to release their captured solutes, then returning to that membrane. Then, through some fluke of evolution, a digestive enzyme which was previously sent to the late endosomes (in preparation for fusion with a lysosome) instead starts getting sent to the early endosomes, and thus gets carried out of the cell by pinocytic vesicles returning to the apical membrane. Thus, by co-opting an already localized process, you secrete something locally instead of globally.
But I think such an example would be too complicated for us to easily model in this game, so I’ll finish with a simple point: I think that, instead of creating a localized structure immediately and then specializing it to secrete agents at certain times or whatever, we should reverse the order – first evolve secretory vesicles to secrete agents all over your surface, then evolve mechanisms to regulate location, timing, type of agent, etc.