Writing plays an important role in institutionalized learning environments. However, it has to be monitored in several ways in order to be successful. We developed automated knowledge assessment tools which allow us to produce instant feedback on a text during the writing process in order to promote self-regulated writing skills. The tools may serve as a complement to human tutoring approaches in learning settings in which individual face-to-face coaching is not possible. To generate text feedback, we used natural language oriented knowledge assessment strategies based on mental model theory and graph theory. Different types of text feedback are then automatically created on the basis of graphs and presented to the learner both for reflection and preflection. So far, we have succeeded in implementing the crucial parts of the coaching into computer-based technology as well as developing and implementing both static and dynamic feedback. Hence, a study of a text-guided assessment tool is presented and discussed. Finally, limitations on the volitional level of the toolset will have to be addressed in future studies.