Original Insight
“Instead of showing a really great analysis, I can use this tool to formulate super sharp questions, because at the end of the day, we don’t have to deliver the answers upfront. Asking a super smart question is really good.” — Dirk Ohlmeier
Expanded Synthesis
The dominant paradigm for using AI in professional work is the answer paradigm: you ask a question, AI provides an answer, you evaluate and use the answer. This is reasonable, and for many tasks it works well. But it misses one of AI’s deepest capabilities — and one of coaching’s deepest levers.
In the June 19 session, Dirk Ohlmeier shared a discovery that reframed how he was using AI in his executive search work: rather than using an AI research prompt to produce a final analysis of competitor businesses or ideal client profiles, he found that the prompt’s real value was in generating the questions he should be asking — questions sharp enough that he could then bring them directly to his clients.
This is the questioning machine use case, and it has profound implications for coaches.
In coaching, the question is often more valuable than the answer. A well-designed question creates movement. It opens a perspective the client hadn’t considered. It reveals an assumption that was invisible. It invites the client into their own wisdom rather than importing someone else’s. The coaching question is the leverage point.
What Dirk discovered — and what Lou affirmed — is that AI, when given the right research prompt and instructed to produce not just analysis but inquiry, can surface questions that a skilled interviewer might take years to develop. The O3 model with web search enabled, when asked to research 15 different CEOs and return their most pressing strategic concerns, produced a list that looked generic at first. But the reframe was: take those findings and ask “what are the three questions no one is asking this person that they most need to hear?” Suddenly the output shifts from analysis to provocation.
For coaches, this opens a specific practice: use AI to design your session pre-work. Before meeting with a client, use an AI research prompt to surface:
- The most important questions in this person’s industry right now
- The blind spots common to people at their career stage or business stage
- The assumptions that are likely driving their current thinking
- The contrarian perspective on the problem they’ve described
Then bring those questions — not as the AI’s answers, but as your questions — into the session. The client experiences you as unusually perceptive. The insight is theirs; you were just the architect of the inquiry.
There is a second dimension to this insight that emerged in the same session: the “stochastic average” problem. Lou noted that when AI produces an analysis without being instructed to find unique things, it returns the most common, well-represented views in the training data — the center of the distribution. For CEO analysis, this meant “digitalization, supply chain, cost reduction” appeared across the board because those are the dominant themes in published business literature.
The fix is simple but counterintuitive: prompt AI to find what is missing from the dominant narrative, not just what is in it. Ask it to find: the underrepresented concerns, the ignored perspectives, the questions no major publication is asking. This pushes the output toward the edges of the distribution, where insight actually lives.
This matters acutely for coaches because your clients are usually stuck in the stochastic average of their own industry’s thinking. Everyone else is saying the same things, using the same frameworks, chasing the same metrics. The coach who can bring questions from the edge of the distribution is the coach who creates breakthrough.
The practical discipline is: before each client engagement, run a targeted AI research query not to get answers, but to get better questions — especially the questions that push against the prevailing narrative in your client’s world.
Practical Application for PowerUp Clients
The Pre-Session Question Generator
Before any significant coaching session, client meeting, or strategy call, run this 3-step AI process:
Step 1: Context Prompt “I’m meeting with a [title/role] in [industry] who is working on [challenge/opportunity]. Based on current patterns in this space, what are the 5 questions most people in this situation are being asked? What are the 5 questions almost no one is asking them?”
Step 2: Assumption Audit “What assumptions is someone in this position most likely making that might not hold in [current year/environment]? Turn each assumption into a question I could ask to surface it.”
Step 3: Edge-of-Distribution Questions “What perspective on this situation would be contrarian, surprising, or counterintuitive — but worth exploring? Frame each as a question rather than a statement.”
Then: Select 3-5 questions from this output that resonate with what you know about the client. These become your session frame.
The deeper coaching application — for coaches coaching coaches: Teach clients to use this same process before their client sessions. The coaching prompt becomes a meta-coaching tool: “How do I design better questions than I would naturally think of?”
Journal prompt: Where am I currently relying on my own mental library of questions, when a 10-minute AI research sprint could give me 20 questions I’ve never thought to ask?
Additional Resources
- O3 model in ChatGPT with web search enabled — best current model for research-quality analysis with citations
- Insight - Teach One Era Ahead of Your Audience, Not Eight — related insight about staying at the cutting edge without overwhelming; applies to how you use AI research to frame questions
- Lou’s research prompts shared in Telegram — if shared, these are worth adapting for pre-session inquiry
Evolution Across Sessions
First appearance of the questioning frame as a distinct AI use case. Dirk’s observation in June 19 is spontaneous — not a framework Lou introduced, but a member discovery that surfaced organically. This is a sign of the mastermind functioning at its best: a member finds a use case that reframes the entire purpose of a tool.
Note also the “stochastic average” concept introduced by Lou in the same session. This is a useful technical framing that has implications beyond research: any time you use AI for brainstorming, content generation, or strategy work without specific instructions to push to the edge of the distribution, you are likely getting average outputs. The antidote is always some variation of “find what’s missing” or “find the contrarian view.”
This insight pairs naturally with the Process Prompt Hierarchy (June 5): the questioning machine use case is a specific, high-value application of process-level prompting designed for the coaching context.
Next Actions
- For me (Lou): Design a “Pre-Session Question Generator” as a reusable process prompt; share as a resource for mastermind members; potentially use as a coaching demonstration at a future session
- For clients: Run the 3-step question generator before their next significant client meeting and compare the quality of the session to usual; report back to the group on what surprised them