Rebo is the first reflection guidance chatbot that systematically guides appren-tices through successive levels of reflection on a learning experience. I firstly investigate general conditions for designing computer-mediated learning inter-ventions for apprentices. I aim to predict their computer literacy by determin-ing computer usage and self-efficacy, and I investigate apprentices as a com-munity of practice, focussing on their learner identities. My main area of inter-est is to develop Rebo, the reflection guidance chatbot that guides them to re-flect on practical learning tasks, working through defined, subsequent levels of reflection. This active guidance from one level of reflection to the next is new, thus a major contribution to existing research on reflection guidance in the field of technology-enhanced learning. The dialogue structure based on de-fined reflection levels, Rebo Junior, is evaluated in multiple field studies, where the reflective conversations between apprentices and Rebo Junior are analysed for coherence, reflection level and engagement. Furthermore, I inves-tigate which adaption mechanisms are relevant for successful reflection guid-ance with a conversational agent, continuously improving and enhancing chat-bot Rebo.