This paper presents a novel method to generate optimal presentations for a Guide-Robot that explains the exhibition to different types of audience. The generation of automatic presentations are selected dynamically regarding different criteria and an intelligent algorithm is implemented based on fuzzy logic to decide which presentation is the optimal. Thus, the decision-making mechanism prioritizes values of the presentation by means of a quality index that the fuzzy logic algorithm generates. The learning phase is produced using feedback information from the public that can modify the previous quality criteria to evaluate if the task is good or bad. Thus, the robot can learn by means of the interaction of the public and with the combination of fuzzy logic for selecting the optimal time and presentation that require a specific guided visit. To ensure that the learning phase is working properly, the robot has been tested in museums where there are interactions between the public and the robot.