I've been thinking more about the concept of "purpose" and think that it could be said that an agent has a purpose in that it's activity is necessary (/sufficient) to a system/thing/property (still meaningless w/out an observer), in which way it seems intent is not necessarily implied by purpose and the agent need not be aware of a similar "bigger picture"/"goal" like the observer.
Also, wanted to post a link to an interesting set of rules for The Game of Life called Brian's Brain/Brain Rules
(I think I saw a module for this in NetLogo as well). The spin on this set of rules is that there are intermediate stages to "life" and most everything reasonably winds up as a glider. I was looking through the modules from the Game of Life again last week and although the ones that start out as already configured patterns are neat the ones that are still most fascinating are the ones that start out very simple and lead to complexity/organization that rivals the former, since these second type are able to account for both a (likely/random/simple) origin and from there a progressive development of the "system" to higher complexity/organization.