At the molecular level, multicellular eukaryotic lineages and bacterial biofilms show predictable evolutionary footprints in their development. For instance, the zygotic embryogenesis ofArabidopsis, which is initiated by gamete fusion, shows hourglass-shaped ontogeny-phylogeny correlations at the transcriptome level. However, many plants are capable of yielding a fully viable next generation by somatic embryogenesis — a comparable developmental process that usually starts by the embryogenic induction of a diploid somatic cell. This leads to the question: is the hourglass-shaped ontogeny-phylogeny correlation preserved in somatic embryogenesis? To explore the correspondence between ontogeny and phylogeny in this alternative developmental route in plants, we developed a new and highly efficient model of somatic embryogenesis in grapevine (Vitis vinifera) and sequenced its developmental transcriptomes. By combining the evolutionary properties of grapevine genes with their expression values, which were recovered from early induction until the formation of juvenile plants, we found a strongly supported hourglass-shaped developmental trajectory. However, in contrast to zygotic embryogenesis inArabidopsiswhere the torpedo stage was evolutionary the most inert, we found that in the somatic embryogenesis of grapevine the heart stage expressed evolutionary the oldest and the most conserved transcriptome. This is a surprising finding because it suggests a better evolutionary system-level analogy between animal development and plant somatic embryogenesis than zygotic embryogenesis. We conclude that macroevolutionary logic is deeply hardwired in plant ontogeny and that somatic embryogenesis is likely a primordial embryogenic program in plants.