So, it's a little different on the attacker side. Post-shell automation has been a thing for a long time, right? Every attacker has their own in-house scripts that get automatically run whenever they pop a shell. It's not much different than that, but it does allow for much easier (and probably more effective) context-aware decision making in post-access data collection.
Once I get that initial shell, my main problem is finding an artifact that allows lateral. Having something that can do all the grunt work of analyzing the system (probably faster and more reliably/thoroughly than I can) without me needing to sit down and code 100s of pages of IF-Then clauses would be a huge aid.