<p>To support effective collaborative learning, it is essential to enhance collective efficacy and provide timely, tailored feedback throughout the learning process. However, it is often difficult for a single instructor to monitor multiple groups and deliver feedback that meets the specific needs of each team in real time. This study aims to develop and validate a systematic design for an AI chatbot that enhances collective efficacy in collaborative learning environments. The design will be devised by developing and validating design principles. The design-based research methodology was conducted to develop the design principles, which were then validated by experts and evaluated for usability. This process involved a review of relevant literature and case analysis, followed by expert validation and usability test to confirm the design principles’ validity. The final design principles are composed of four design elements— (1) support for group cohesion, (2) support for affective cohesion, (3) support for collaborative learning activities, and (4) facilitation of dialogue—and ten design principles. These include the principle of belongingness formation, interdependence, positive atmosphere, empathy formation, sharing, collaborative problem-solving, social regulation, immediate scaffolding, familiarity, and personification. The design principles are further supported by 46 actionable sub-guidelines. Validation was conducted through three rounds of expert review and usability testing with educators and developers, whose diverse perspectives informed both the pedagogical soundness and technical feasibility of the principles. This study contributes to the field by offering theoretically grounded and practically applicable guidelines for designing AI chatbots in education. Importantly, the principles are designed to be usable by educators without requiring technical expertise, thus supporting scalable and accessible implementation in real classrooms.</p>