This paper presents an architecture designed for simulation serious games, which automatically generates game-based scenarios adapted to learner's learning progression. We present three central modules of the architecture: (1) the learner model, (2) the adaptation module and (3) the logs module. The learner model estimates the progression of the development of skills targeted in the game. The adaptation module uses this estimation to automatically plan an adapted sequence of in-game situations optimizing learning. We implemented our architecture in Game of Homes, a simulation serious game, which aims to train adults the basics of real estate. We built a scripted-based version of Game of Homes in order to compare the impact of scripted-based scenarios versus generated scenarios on learning progression. We qualitatively analyzed logs files of thirty-six adults who played Game of Homes for 90 minutes. The main results highlighted the specificity of the generated pedagogical scenarios for each learner and, more specifically, the optimization of the guidance provided and of the presentation of the learning content throughout the game.