Learners' concentration is an essential factor for learning and acquisition. The duration of concentration varies from one individual to another. Some learners have a long duration of concentration; whereas, others have a short one. Leaving the learner in front of a screen for a random duration is a strategy that does not optimize online learning. In this perspective, we have implemented an approach based on three intelligent agents customizing and adapting the learning time according to the learner's concentration time. The first agent is called "Detector Agent" (DA).The task of this agent is to measure the learner's concentration time and to detect the factors that may influence it. The second agent is named "Scheduler Agent" (SA). Its function is to cut the time of a session in proportion to the concentration time measured by DA and to program relaxing breaks in order to regenerate the degree of concentration. And finally, the third agent is called "Rectifier Agent" (RA). The latter is responsible for rectifying the factors that negatively influence the concentration of the learners. These agents continuously communicate between each other in order to ensure an efficient treatment. The experienced results show that the approach contributes effectively in the acquisition and the performance of learners. Success rates have risen sharply and the learners express a growing satisfaction.