Designing AI for Reflection, Not Advice: Conversation Architecture, Avatars, and Ethics | with Dr. Vanessa Mai

Dr. Vanessa Mai is a systemic business coach, communication scientist, and researcher at the University of Cologne—and since 2019, she has been pioneering the empirical design of AI coaching chatbots. Her mission is clear: creating conversational systems that foster genuine human self-reflection rather than delivering pre-packaged advice. In this conversation with Rebecca Rutschmann, she bridges the gap between coaching psychology and technical architecture. She unpacks the lessons learned from evolving early rule-based chatbots into generative AI sparring partners, explaining why digital onboarding serves as vital psychological contracting, why anthropomorphic avatars and faux-empathy damage client trust, and how coaches can lead ethical AI product design. What you'll take away: Why true AI coaching must provoke inquiry and self-reflection rather than spit out advice The critical role of digital onboarding: why the very first prompt sets the entire psychological contract The empathy paradox: why AI that tries too hard to sound human breeds sycophancy and breaks authenticity Avatar and persona design: why transparent, non-human characters build safer spaces for vulnerability How user drop-off data exposes latency issues, context limits, and client hesitation to challenge AI Resources & Links: Rebecca & Vanessa Mai's Springer Paper (Published Jan 2023) David Clutterbuck's 2019 Paper on AI in Coaching Jonathan Gratch's Research on Human–Machine Self-Disclosure Joseph Weizenbaum's Original ELIZA Paper (1966) Vanessa Mai's Profile and Research on Chatbot Self-Disclosure & Rapport Dr. Nicky Terblanche's AI Coaching Research on Research Gate Vanessa Mai on LinkedIn