Thanks to their intrinsic network topologies, dynamic combinatorial libraries (DCLs) represent new tools for investigating fundamental aspects related to self-organization and adaptation processes. Very recently the first examples integrating self-replication features within DCLs have pushed even further the idea of implementing dynamic combinatorial chemistry (DCC) towards minimal systems capable of self-construction and/or evolution. Indeed, feedback loop processes - in particular in the form of autocatalytic reactions - are keystones to build dynamic supersystems which could possibly approach the roots of "Darwinian" evolvability at mesoscale. This topic of current interest also shows significant potentialities beyond its fundamental character, because truly smart and autonomous materials for the future will have to respond to changes of their environment by selecting and by exponentially amplifying their fittest constituents.