Model driven engineering aims to shorten the development cycle by focusing on abstractions and partially automating code generation. We long lived in the myth of automatic Model Driven Development (MDD) with promising approaches, techniques, and tools. Describing models should be a main concern in software development as well as model verification and model transformation to get running applications from high level models. We revisit the subject of MDD through the prism of experimentation and open mindness. In this article, we explore assistance for the stepwise transition from the model to the code to reduce the time between the analysis model and implementation. The current state of practice requires methods and tools. We provide a general process and detailed transformation specifications where reverse-engineering may play its part. We advocate a model transformation approach in which transformations remain simple, the complexity lies in the process of transformation that is adaptable and configurable. We demonstrate the usefulness, and scalability of our proposed MDD process by conducting experiments. We conduct experiments within a simple case study in software automation systems. It is both representative and scalable. The models are written in UML; the transformations are implemented mainly using ATL, and the programs are deployed on Android and Lego EV3. Last we report the lessons learned from experimentation for future community work.