Drexler had two ideas for how molecular manufacturing would work: (i) something like a robot arm that would pick and place atoms, and (ii) a mill with molecular gears, conveyor belts and stuff like that.

Circa 1990 some researchers at IBM pushed some atoms around on a surface to spell "IBM" and visualized it with a scanning probe microscope, which seems like a baby step towards (i). To accomplish (i) you need a tip that can pick up an atom and push it to a reaction site and make the reaction happen, much like an enzyme. If you had the tip you could put it on a probe and start building, but we don't. (I do wonder though how you sense the environment to know that the last reaction went right, where the source atom is, etc.)

So far as (ii) I think to biological systems for inspiration and it's something chemists are making gradual progress on

https://www.nature.com/articles/s41467-020-17814-0