Well, I haven't yet worked with the runtime, but I guess there's no AI implemented at all. So basically that's what your up to. Only following pathes should however be pretty simple.
For the AI:
*you'll need some primary loop, that checks every ... what the pawn is going to do (like a tick function, but not necessarily that often). At the end of this loop, when all actions are planned you should execute them.
Example:
loop
-pawn shall go somewhere
->check if pawn has a destination (no)
->determine where pawn should go (may be location of next PathNode)
->start moving
loop
-pawn shall go somewhere
->check if pawn has a destination (yes)
->if pawn is not yet within a specific range (say about 32UU)(no)
->start moving
loop
-pawn shall go somewhere
->check if pawn has a destination (yes)
->if pawn is not yet within a specific range (say about 32UU)(it is)
->stop moving
->clear destination
loop
=>do it all over again
So I hope that helps you to figure out the basics of the AI you need.