The 31st meeting of the dialogue series “Coaching Meets AI” focused on the topic of prompt design in the context of AI-supported coaching. The discussion covered the following key points:
1. Definition and Importance of Prompts
- Prompts were defined as inputs that precisely and effectively guide AI systems. They consist of a contextual description and a processing task.
- A Prompt Script is a structured template that combines different types of prompts for specific objectives.
2. Experiences and Recommendations
- Good prompts are characterized by clarity, simplicity, and systematics. It is essential to formulate requirements precisely and understandably.
- Redundancy and clear structure help to avoid misunderstandings in AI responses.
3. Typology of Prompts
Five types of prompts were distinguished in triadic coaching (Coach, Coachee, and AI):
- A-Prompt: Provides fundamental orientation and instructions for the process.
- B-Prompt: Assists in generating contextual information (e.g., problem description).
- C-Prompt: Directly addresses the AI with a concrete task (e.g., analysis or summarization).
- D-Prompt: Guides the coachee to reflect on the AI’s responses and ask follow-up questions.
- E-Prompt: Encourages feedback from the coachee to the AI to improve its responses.
4. Coaching Concepts in Prompts
- The prompts integrate systemic coaching methods, such as the 7-Field Shimmer Analysis (e.g., consideration of actual vs. target situations, psychological processes, and behavioral patterns).
- The goal is to ensure the highest possible quality of AI responses through precise problem definition and reflection.
5. Success Factors in Coaching
The quality of prompts was analyzed based on eight success factors from coaching research, including:
- Process management and structuring
- Emotional support
- Goal and motivation clarification
- Resource activation
- Problem analysis and self-reflection
- Affect regulation
- Implementation support
- Knowledge transfer (particularly relevant in AI coaching).
6. Practical Application
- Participants shared experiences on integrating prompt scripts into coaching tools, such as generating daily questions via AI systems.
- Differences between triadic and dyadic coaching were discussed. The latter emphasizes the interaction between the coachee and the AI.
7. Quality Control of Prompts
The quality of prompts was assessed based on the aforementioned success factors. Each prompt type addresses different factors, leading to evolving requirements throughout the coaching process.
8. Conclusion and Outlook
- The systematic development of prompt scripts is crucial for the effectiveness of AI-supported coaching.
- The next meeting will focus on image-based and constellation work to further deepen the coaching process.
The discussion highlighted that well-thought-out prompt design and the integration of coaching principles significantly contribute to the success of AI coaching.