Challenging parent-teacher conversations are a recurring part of professional practice. They are emotionally charged, shaped by asymmetric expectations and attributions of responsibility, and can escalate quickly. At the same time, they require structured facilitation, perspective-taking, and workable agreements. In teacher education, authentic practice opportunities often remain episodic and difficult to scale because they depend on trained role-players and intensive coaching (Lane & Rollnick, 2007).

Against this background, we present an AI-supported learning environment that enables repeatable, situated practice with embodied Parent Avatars and structured reflection with an embodied Coach Avatar in both VR and desktop modes.

Prior work indicates that virtual role-play can enable repeatable, situated communication practice, and that generative, embodied agents can support unscripted dialogue and scalable coaching beyond predefined branching scripts (Othlinghaus-Wulhorst et al., 2019; Li et al., 2025). Debriefing is pivotal for consolidating simulated experience into learning, and recent VR evidence suggests chatbot-guided debriefing can yield competence and self-efficacy gains comparable to facilitator-led formats when prompts follow established structures (Evangelou et al., 2026; Palaganas et al., 2016). We therefore position embodied conversational agents as dual-purpose components for interactive simulation and structured debriefing at scale (Rudolph et al., 2006).

We treat both modes as pedagogically equivalent options: VR supports presence and conversational proximity, while the desktop mode is expected to be used more frequently due to lower access barriers and easier integration into regular modules (Hamilton et al., 2021; Witmer & Singer, 1998). The project connects to the broader need to foster transversal competences through scalable training opportunities (Calero López & Rodríguez-López, 2020).

Our guiding question is how a VR- and desktop-based tool can (a) create realistic practice conditions for demanding conversations, (b) support reflective learning and transfer, and (c) remain feasible for routine use in modules with larger cohorts. As the didactic frame we use Nonviolent Communication (NVC) as a practice-oriented model for de-escalation and relationship-building (Rosenberg, 2015). The metaphors of giraffe language and jackal language serve as memorable cues to shift from judgements, blame, and demands toward observations, feelings, needs, and requests.

The tool follows a short learning journey:

• a brief orientation by the Coach Avatar,
• a spoken simulation with a Parent Avatar, and
• a debrief with the Coach Avatar that reconstructs critical moments, poses reflection questions, links them to NVC concepts, and plans next steps for transfer.

The tool offers three freely selectable scenarios that represent typical challenging cases in lower secondary schooling in Switzerland: (1) track transition and performance pressure, (2) school anxiety and pressure for immediate relief, and (3) cyberbullying and escalation. The AI-enabled, embodied avatar approach aims to provide scalable conversational practice while preserving realism and variability, which current work on educational chatbots and conversational agents identifies as a key design opportunity and challenge (Okonkwo & Ade-Ibijola, 2021; Wollny et al., 2021).

While the primary use is in teacher education, the competence target is transferable to VET and workplace learning contexts where professionals regularly navigate demanding stakeholder conversations. The contribution addresses a widely shared challenge: scaling authentic communication practice without losing didactic quality and reflective depth, and securing this through iterative field development (Design-Based Research Collective, 2003; Wang & Hannafin, 2005; McKenney & Reeves, 2012).

The prototype was developed through an iterative Design-Based Research (DBR) process together with lecturers and technologists (Design-Based Research Collective, 2003; Wang & Hannafin, 2005; McKenney & Reeves, 2012). DBR cycles combined co-design, scenario authoring, prototype refinement, and pilots in existing modules. Implementation provides two equivalent modes. In VR, embodied avatars are encountered with high conversational proximity and increased sense of presence (Witmer & Singer, 1998). On desktop, the same scenarios and debrief structure can be accessed with minimal setup, supporting frequent practice and cohort-scale integration. Both modes share the same learning journey and debrief logic. After several smaller formative evaluations, a formative evaluation with 17 pre-service teachers was conducted at the end of prototyping. Participants completed all three scenarios in a within-session design: one scenario in VR and two scenarios on desktop (with headphones). Per scenario, approximately four minutes of simulation and approximately four minutes of debrief with the Coach Avatar were conducted. Data sources were short, open-ended voice responses on perceived realism, learning value, usefulness of the debrief, and improvement suggestions. Voice responses were transcribed and subsequently anonymised or pseudonymised. Analysis follows an established approach to thematic analysis to identify recurring patterns in early-stage user feedback (Braun & Clarke, 2006).

First results suggest that participants experienced the simulated conversations as sufficiently authentic to create productive pressure and to rehearse de-escalation strategies. Many comments highlight the Coach Avatar as particularly helpful for structuring reflection, identifying escalation moments, and reframing them with NVC concepts (Rosenberg, 2015). Participants described both modes as useful: VR was associated with stronger conversational proximity and engagement, consistent with presence-oriented accounts of immersive interaction (Witmer & Singer, 1998), while the desktop mode was valued for ease of access and for enabling repeated practice across modules. At the same time, feedback indicates typical limitations of an early prototype, including occasional latency or turn-taking friction, the wish for more concrete and action-guiding debrief prompts, and greater variability in parent responses to widen the practice space. These observations are treated as first results that inform ongoing refinement of scenario logic, behavioural boundaries of the avatars, and debrief scaffolding. Given the dynamics of AI-based dialog approaches, interaction quality control and pedagogical framing remain central design concerns (Okonkwo & Ade-Ibijola, 2021; Wollny et al., 2021). Next steps include continued iterative development and a planned summative, quantitative evaluation focusing on learning outcomes, transfer indicators, and sustainability conditions for routine module use. The stated goal is to refine the prototype further and implement the training sustainably as a recurring element of the programme.

Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101. https://doi.org/10.1191/1478088706qp063oa Calero López, I., & Rodríguez-López, B. (2020). The relevance of transversal competences in vocational education and training: A bibliometric analysis. Empirical Research in Vocational Education and Training, 12, Article 12. https://doi.org/10.1186/s40461-020-00100-0 Design-Based Research Collective. (2003). Design-based research: An emerging paradigm for educational inquiry. Educational Researcher, 32(1), 5–8. https://doi.org/10.3102/0013189X032001005 Evangelou, D., Mulders, M., & Träg, K. H. (2026). Debriefing in Virtual Reality Simulations for the Development of Counseling Competences: Human-Led or AI-Guided? Technology, Knowledge and Learning. https://doi.org/10.1007/s10758-025-09941-8 Hamilton, D., McKechnie, J., Edgerton, E., & Wilson, C. (2021). Immersive virtual reality as a pedagogical tool in education: A systematic literature review of quantitative learning outcomes and experimental design. Journal of Computers in Education, 8(1), 1–32. https://doi.org/10.1007/s40692-020-00169-2 Lane, C., & Rollnick, S. (2007). The use of simulated patients and role-play in communication skills training: A review of the literature to August 2005. Patient Education and Counseling, 67(1–2), 13–20. https://doi.org/10.1016/j.pec.2007.02.011 McKenney, S., & Reeves, T. C. (2012). Conducting educational design research. Routledge. Okonkwo, C. W., & Ade-Ibijola, A. (2021). Chatbots applications in education: A systematic review. Computers and Education: Artificial Intelligence, 2, 100033. https://doi.org/10.1016/j.caeai.2021.100033 Palaganas, J. C., Fey, M., & Simon, R. (2016). Structured debriefing in simulation-based education. AACN Advanced Critical Care, 27(1), 78–85. https://doi.org/10.4037/aacnacc2016328 Rosenberg, M. B. (2015). Nonviolent Communication: A Language of Life (3rd ed.). PuddleDancer Press. Rudolph, J. W., Simon, R., Dufresne, R. L., & Raemer, D. B. (2006). There’s no such thing as “nonjudgmental” debriefing: A theory and method for debriefing with good judgment. Simulation in Healthcare, 1(1), 49–55. https://doi.org/10.1097/01266021-200600110-00006 Wang, F., & Hannafin, M. J. (2005). Design-based research and technology-enhanced learning environments. Educational Technology Research and Development, 53(4), 5–23. https://doi.org/10.1007/BF02504682 Witmer, B. G., & Singer, M. J. (1998). Measuring presence in virtual environments: A presence questionnaire. Presence: Teleoperators and Virtual Environments, 7(3), 225–240. https://doi.org/10.1162/105474698565686 Wollny, S., Schneider, J., Di Mitri, D., Weidlich, J., Rittberger, M., & Drachsler, H. (2021). Are we there yet? A systematic literature review on chatbots in education. Frontiers in Artificial Intelligence, 4, 654924. https://doi.org/10.3389/frai.2021.654924