So, I've been thinking, right.
First I was thinking about Attacks of Opportunity, what with the whole peek around the corner exploit becoming only more frustrating once the AI is (hopefully) smart enough to exploit it. First I thought using them in the limited way that (I think) most systems...