Well, when we develop nanobots, they'd certainly be useful for treating any disorder, not only cancer but also ageing, at the molecular level. However, like practical fusion power, they appear to be some way off. Until then, adapting existing biological nanomachinery, for example CRISPR-CAS9, retroviruses and bacteriophages, appears to be one way forward. We'll probably need to find a way to stop collateral editing of harmless DNA sequences that happen to match those being targetted. In other words, the editing mechanisms need to act only in the correct context.
Personally, I think a combination of microbeads for high-level localisation of treatment, delivering modified bacteriophages that can recognise anomalous cells and which inject editing nanomolecules to induce cell apoptosis might be a possibility.