Abstract. Soft Computing (SC) is a concept with constantly evolving semantics, as researchers have adopted its main philosophy while adding various interpretations and facets to this concept. Originally defined as a loose association or partnership of components, SC has gone through several transformational phases. This paper will trace some of the phases experienced by the author as part of his understanding of the evolution of SC and its role in constructing decision-making models. The first phase is the hybridization phase, driven by the inherit ease of integration of SC components. The second phase is a two-level model characterization, based on the split between object-level and meta-level reasoning. This phase, inspired by traditional AI problem formulation, led to a third phase, in which we addressed the knowledge and meta-knowledge representation required by each of these reasoning levels using a linguistics analogy. The fourth phase is the extension of the heuristics used at the metalevel, e.g. Metaheuristics (MH's) from evolutionary algorithms to other search methods. The fifth and last phase, further described in this paper, is the proposal for a strong separation between offline MH's (used for design and tuning) and online MH's (used for models selection or aggregation.) This last view suggests a broader use of SC components, since it enables us to use hybrid SC techniques at each of the MH's levels as well as at the object level. Furthermore, this separation facilitates the model lifecycle management, which is required to maintain the models vitality and prevent their obsolescence over time.