In or ordinary life, we ascribe action and doing to other humans, and lower organisms, even bacteria swimming up a glucose gradient to get food. Yet physics has no "doings" only happenings, and the bacterium is just a physical system. I have struggled with the question "What must a physical system be to be able to act on its own behalf?" Call such a system an autonomous agent. I may have found an answer, such systems must be able to replicate and do a thermodynamic work cycle. But of course I'm not sure of my answer. I am sure the question is of fundamental importance, for all free living organisms are autonomous agents, and with them, doing, not just happenings, enters the universe. We do manipulate the universe on our own behalf. Is there a better definition of autonomous agents? And what does their existence mean for science, particularly physics?