This presentation concerns a theory on how gestures (accompanying speaking and silent thinking) are generated and how gestures facilitate the gesturer’s own cognitive processes. I will present evidence that gestures are generated from a general-purpose Action Generator, which also generates “practical” actions such as grasping a cup to drink, and that the Action Generator generates gestural representation in close coordination with the speech production process (Kita & Ozyurek, 2003, Journal of Memory and Language). I will also present evidence that gestures facilitate thinking and speaking through four functions: gesture activates, manipulates, packages and explores spatio-motoric representations (Kita, Chu, & Alibali, 2017, Psychological Review). Further, I will argue that schematic nature of gestural representation plays a crucial role in these four functions. To summarise, gesture, generated at the interface of action and language, shapes the way we think and we speak.