It’s one thing to design an AI interaction, and another to witness it truly land with a group. This past week, during a live session for a mental healthcare network transformation in a Dutch province, I experienced one of those "wow" moments. We were working with Dembrane, our AI partner for capturing and synthesizing dialogues, and a feature we call the "echo button" led to a genuinely inspiring group reaction. It wasn't just about the tech; it was about the synergy between human facilitation, participant vulnerability, and a strategically prepared AI prompt.

The session involved a dedicated talk group, expertly guided by a facilitator, Jeroen. He has a real knack for drawing people out, asking clarifying questions like, "Can you say a bit more about that?" whenever someone shared a keyword or a concise thought that he sensed could add more value. This created an atmosphere where people felt heard, leading to a rich and meaningful dialogue. Adding to this depth was an expert by experience, Franky, who shared his perspectives with great vulnerability.

As the Dembrane analysis later confirmed, their conversation just before the "echo" moment centered on some crucial tensions:

Their implicit goal, as Dembrane helped clarify, was to find "ways to realize transformation without systemic demands and budget cuts overshadowing the human dimension." I was observing this unfold, managing the Dembrane input from my laptop, somewhat removed from the immediate facilitation but deeply engaged in the content.

At a certain point, Jeroen consciously chose to press the "echo button." Behind this button, I had prepared a prompt for our AI partner, designed to do two things: reflect back a synthesis of what had been discussed and pose a question to fuel the next round of conversations with other groups.

The question generated by the AI, based on their preceding dialogue, was:

(In English: "Given the challenges you describe, it seems crucial to start with small, feasible steps that have a direct impact in the neighborhood. Can we think of an example of such a concrete action that we could start tomorrow, without getting bogged down in systemic requirements?")

The Dembrane analysis highlighted how perfectly this question connected: "small feasible steps" resonated with the need for space for new working methods, "without getting bogged down in systemic requirements" directly addressed the system-versus-human tension, and "direct impact in the neighborhood" brought in a focus on tangible results for people.

The group's reaction was immediate and palpable. As Dembrane captured:

From my vantage point, I saw what I can only describe as mouths falling open. It was a moment where, I believe, people suddenly saw the profound potential of working this way when it's applied thoughtfully. It worked because Jeroen felt the freedom and the right moment to use it, and the prompt delivered something that the group instantly recognized as highly relevant to their conversation. "Ah yes," they seemed to think, "this is spot on."

This experience also shed light on my own role. I find the strategic, preparatory work – like designing these prompts – incredibly engaging, more so than the live operational side of checking if the AI is capturing everything correctly. This "echo button" moment was a powerful affirmation of the value of that strategic preparation.

However, it also underscored that I can only truly witness and learn from these impactful moments if I am present and involved in the live work, even if it's in an observational or support capacity. It’s pushing me to think even more deeply about the purpose of such AI interventions in a live dialogue. What exactly do I hope to achieve when that button is pressed? This experience has certainly laid that question bare in a very stimulating way.

Reflecting on this, a few thoughts emerge that might be useful for other facilitators looking to work with AI in similar ways: