I've observed that ants - those minuscule insects moving about alone or in swarms, in or away from colonies, are some of the most interesting automatons in nature.
I know that humans have been clever enough to begin to study and emulate these as artificial automatons, and this could be possible, eventually, even at nano scale.
One of the things to note about these kinds of little intelligent creatures, is that they somehow operate in the smallest of spaces, though capable of results that can have implications at a macro scale. They coordinate their collective work - which often is at a scale impressively great relative to their sizes, and yet, they go about their work nearly "silent".
Natural ants might use chemical footprints to track and coordinate action with each other, but could also leverage more sophisticated signals sent to each other via antennae (do they use electromagnetism, sound, heat or light to pass information to each other - interesting problems to research).
Obviously, learning from these, and then applying the ideas to artificial automatons - think swarms of autonomous or semi-autonomous bots and iot devices (especially those capable of movement, and which are equipped with both sensors and transmitters). Could have interesting applications in automated autonomous information gathering, communications network extension (by leveraging the low resource, low-scale coordination and information sharing capabilities of such automatons), security systems - signal detection, alarm mechanisms, defense, and more.
Another interesting idea about these sorts of systems occurred to me as I pondered these little creatures in the night, as they played about on the tiled floor of the sitting room; it should be possible to model a computational divinatory system upon these.
How this would work? Basically, assign concepts to each of the cardinal vertices or directions relative to the grid. And then, observing and reading off the location/position of each ant or automaton on the grid, compute the distance from each concept, which can then be reduced to a single or aggregate value from which information can then be inferred.
This could form the basis for an even more general computational divinatory framework applicable anywhere such a reference space can be projected, and where these autonomous automatons can be observed "writing in the space".
Also, thinking of how humans are social, and how individuals and groups function and move about in closed and open spaces (markets, streets, restaurants, etc.), one could envision a higher observer - say one using surveillance cameras, satellites, or even bare perception from an elevated vantage point, who can study and infer or direct behavior as though humans were merely ants. I postulate that although the average human doesn't acknowledge ant-like behavior in humans (down to the possibility of telepathic coordination via "invisible antennae"), such is more than likely to be plausible - if not true. A systematic study of this is in order.