Personally, I don’t think you could get this to work in large enough scale to be usable.
Back 20 years ago, IBM spelled out IBM in, I think gold atoms using an electron microscope and some other equipment.
Creating functional molecular bonds is more difficult than just moving atoms around.
Interesting idea though
The apparatus would at the very least require a database-like library of known pathways and kinetic data to synthesize a practical route from reactants to products.