Learning is an important activity from elementary school to university and longer (long-life learning). A variety of learning models and approaches are used with more or less active and autonomous orientations. The objective of technology-supported learning is to assist and empower these approaches by proposing Learning environments. Collaboration is a key aspect of learning in that it allows learners to share and exchange learning contributions. In group learning (class learning), size is often a problem for the teacher, making it hard for them to assist and supervise all learners. In this paper we propose a Collaborative Learning Environment using technology-supported tools organized as a system, taking into account different learning approaches. Class learners are dynamically split into smaller groups, which are either managed by the teacher (for learners needing close supervision and guidance) or by the Virtual Assistants – Chatbots (for more autonomous learners). This split is not only based on actual working results but also on learning emotion perception. We thus aim to combine pedagogy and a neuroscience-based view of emotions and emotional approaches. We provide the main principles, the system architecture, the orchestration process and the first results.