Near term and far term
APM in the near term and APM in the far term
Today and near term
Pick and place assembly of single atoms (or molecule fragments) is not at all a necessity for early forms of APM.
In fact pick and place assembly is not needed at all for early forms of APM. Assembly driven by the "vigorous" thermal motion at the nanoscale (slightly misleading tech-term: "self assembly") can do the job.
- This Thermally driven assembly is not present in macroscale manufacturing. Therefore it is not present in our (knowledge and) intuition (unless we study the nanoscale in detail). Advanced APM is sometimes claimed to be impossible due to the effects of thermal motion. Which is clearly wrong for all the points that have been pointed at (mechanosynthesis, friction, ...). What is the case is that some proponents of advanced APM may lack knowledge (and intuition) regarding thermal motion.
- Thermally driven assembly puts thing together in faulty configurations quite often (high error rates). But its just enough such that one can start climbing the "stiffness ladder" introducing more and more restrained and forced motion leading to advanced APM.
It may come somewhat unexpected but in early APM systems there is no need for the atoms to stay in place. No, that does not contradict the introduction earlier. The atoms still need to keep their nearest neighbors they are strongly bonded to. What needs to be preserved such that is counts as atomically precise (in the weak (topological) sense) is just "what links to what" (tech-term: "bond topology").
In the early atomically precise systems of today the atoms tend to be bonded together in polymer chains. The whole chains constantly deform since the (zig-zag going) bonds in these polymer chains can (and very much do) rotate and flex. Thereby atoms can be displaced much more than their own diameter. Polymer chains with (mutually puzzle piece like matching) "side groups" that cause these chains to fold up into compact lumps (such chains are called: "foldamers") restrict this unwanted freedom of motion far enough to give the folded lumps a (more or less) predictable shape. But the location of the individual atoms may (and usually will) still wiggle around way beyond the diameter of the individual atoms.
In some sense even chemistry (the deterministic parts of it) could be counted as the earliest form of APM. (This is very much excluding macromolecular polymer chemistry with statistical cross-linking.) Important to note is that a major aspect of APM is that it specifically focuses on scaling up APM capabilities to bigger sizes. Chemistry is on the very bottom and does not scale up well.
In advanced atomically precise systems the atomically precise lumps are no longer made from folded up chains. Instead of chains the chemical bonds form tight meshes. Tiny crystals with molecule character. This tight mesh of bonds prevents the bonds from rotating, excessive stretching and bending. It is stiff. Here the location of the individual atoms can finally be restrained below the diameter of the individual atoms. This is atomically precise (in the strong (positional) sense). It allows advanced force applying mechanosynthesis.
Towards the far term
There are two core ideas that determine what the R&D direction from early forms of APM to advanced forms of APM actually is. This wiki will refer to those two ideas with the shorthand "gem-gum".
Further reading on the page: The defining traits of gem-gum-tec.