Life is a complex phenomenon that not only requires individual self-producing and self-sustaining systems but also a historical-collective organization of those individual systems, which brings about characteristic evolutionary dynamics. On these lines, we propose to define universally living beings as autonomous systems with open-ended evolution capacities, and we claim that all such systems must have a semi-permeable active boundary (membrane), an energy transduction apparatus (set of energy currencies) and, at least, two types of functionally interdependent macromolecular components (catalysts and records). The latter is required to articulate a 'phenotype-genotype' decoupling that leads to a scenario where the global network of autonomous systems allows for an open-ended increase in the complexity of the individual agents. Thus, the basic-individual organization of biological systems depends critically on being instructed by patterns (informational records) whose generation and reliable transmission cannot be explained but take into account the complete historical network of relationships among those systems. We conclude that a proper definition of life should consider both levels, individual and collective: living systems cannot be fully constituted without being part of the evolutionary process of a whole ecosystem. Finally, we also discuss a few practical implications of the definition for different programs of research.