We live in a world replete with complex system of systems that need to operate in a dynamic and uncertain environment thus necessitating continuous adaptation so as to deliver the stated goals which may also change over time. These systems, characterized by large size and non-linear interactions, are spread across spaces such as information-only, cyber-physical, societal, and biological. They raise new concerns to be addressed such as: (i) Why things are the way they are? (ii) What are the right interventions to bring the system back to the desired state? (iii) Is a better state possible? And (iv) How best to reach the desired to-be state from as-is state? With current practice found wanting, we present a digital twin centric simulation-based approach and associated technology that builds further upon proven results from Modelling & simulation, AI, Control Theory, and Software Engineering while also leveraging Generative AI in supporting Digital Twin life cycle. We illustrate utility and efficacy of the approach with a representative sample of real world use cases and highlight open challenges that need to be overcome going forward.