Episode 182
Building Believable AI Coaches with Luka Abramovic
How do we design AI that people trust, feel understood by, and can genuinely grow with?
What does it really take to build an AI coach that feels believable, empathetic, and trustworthy — without ever pretending to be human?
In this rich and forward-thinking conversation, Luka and Zoran take listeners behind the scenes of designing Leelou AI Coach, exploring:
- The major shift from AI “giving answers” to AI asking powerful, developmental questions
- How empathy can be modelled and expressed through algorithms
- Ethical boundaries in AI coaching — and how to build them from day one
- Designing safety, crisis protocols, and “edge case” responses
- What L&D leaders need to know before adopting AI coaching solutions
- How human coaches and AI can work together to expand coaching access
- The future of trustworthy, ethical, human-centred AI in leadership development
Luka shares how the architecture, data structures, and behaviour design of Leelou were built to enhance human capability — not replace it — and how teaching AI to coach actually made the entire team better coaches.
This episode brings together technology, psychology, ethics, and the soul of coaching into one inspiring dialogue.
Key Talking Points
- Moving AI from authority to curiosity
- Teaching AI to ask coaching questions, not give advice
- How AI can mirror emotions, values, and meaning authentically
- Building empathy into AI behaviour loops
- Ethics-by-design: boundaries, safety, and non-directive intelligence
- Technical + behavioural foundations behind Leelou
- What L&D leaders must evaluate in AI coaching tools
- Where AI coaching is heading — and where humans remain irreplaceable
- The future partnership between AI coaches and human coaches
- Why AI can expand coaching access globally without losing quality