Deliberatively planning, where an agent thinks through several steps ahead before actually making the first step, isn't used in industry Game AI outside of spatial pathfinding. In this talk about the behavioral AI in the game F.E.A.R., you'll see how this style of planning comes up in a commercial game and how it entangles with the design-driven animation subsystems of the game.
Download this zip (contains a ppt file of slides, several wmv videos, and a pdf file): http://alumni.media.mit.edu/~jorkin/gdc2006_orkin_jeff_fear.zip
Navigate to https://www.gdcvault.com/play/1013282/Three-States-and-a-Plan to start the audio playing.
Click along with the slides, while the audio plays, and open videos when they seem relevant.
After laboriously watching the talk, review the pdf text. It follows the same path, but you might have different comprehension for text versus images/audio.
Respond to the talk/text as a whole in the usual format.
Jeff Orkin has some other GOAP-related resources on his academic website: http://alumni.media.mit.edu/~jorkin/goap.html