I Built an AI Skill to Challenge My Thinking Raising the Barr Weekly Memo: Issue 716

I want more from AI than a fast answer. I want a thinking partner that questions my assumptions, challenges my conclusions, and tells me when my reasoning falls short. I challenge its answers just as firmly, because confidence is not proof that either of us is right.

But a productive exchange has value beyond the work in front of me. When a question exposes a weak argument or a correction improves the result, I want that lesson carried into the work I do next. That is why I create AI Skills: to build a library of processes I can put to work whenever I need them. I preserve what works, refine what falls short, and expand each Skill as my needs change.

Make Better Thinking Repeatable

A good prompt can describe an entire process. A Skill brings that process, my examples, and supporting materials together in a form I can install, use, and refine.

With my Article Interview Skill, I don’t have to reconstruct the interview process or explain how I want to be challenged every time. When something falls short, I improve the Skill itself. The correction becomes part of the instructions for future articles, rather than a fix buried in yesterday’s conversation.

A Skill is not automatically better than a good prompt. For an occasional task, a prompt may be enough. For work I repeat, the advantage is having the whole process ready to run, refine, and use again.

I don’t just save the answer. I create a process that helps me think better.

My Article Interview Skill

I designed and created the Article Interview Skill to ask me one question at a time. The goal is to draw out what I know, why I believe it, and what my experience actually supports. I instructed it to push me further when my answer is vague or a conclusion needs more thought. I do not want a weak idea turned into a polished paragraph. I want to find out what would make the idea stronger.

Suppose I say consultants need to become more visible. That is familiar advice, not yet an article worth reading. What holds them back? What have I seen in my work that explains their reluctance? Am I identifying their real problem, or repeating advice I have not questioned? The answers could change my original argument, not just help me explain it.

AI also helps me draft, organize, and revise, including this article. It works from my ideas, my experience, and my own words, with instructions to preserve my voice rather than substitute language I would never use. I review whether the writing sounds like me, whether the reasoning holds up, and whether the article delivers something practical and valuable. I remain responsible for what I publish.

For a consultant or advisor, producing more content is valuable when it brings more of your unique ideas and insights to the people you serve. That is what I wanted this Skill to help me do repeatedly: create stronger content that helps my readers while pushing my own thinking further.

Build the Process Before the Skill

I do not start by asking AI to create a Skill. I start with a conversation about the work I want to improve. We explore what the Skill should accomplish, how it should work with me, and what I may have overlooked. I ask AI to challenge the approach, not simply turn my first idea into instructions.

You do not need to arrive with the entire process figured out. Developing it with AI is part of the value. Here is how I approach it:

1. Choose a recurring task worth improving. Preparing for client conversations, strengthening proposals, or developing articles are useful starting points. Explain how you handle the task today, where it falls short, and what a better result would look like.
2. Work through the approach together. Ask AI to question you one step at a time and recommend improvements. Consider what information it needs, where it should challenge you, and when it must stop for your input.
3. Bring your judgment into the instructions. Share examples of work you value and explain why they work. Identify what AI must preserve, what it must avoid, and what it should never assume. This is where the process becomes yours.
4. Test the process on real work. Have AI walk through it before creating the Skill. Notice what it misses, where the questions could be stronger, and whether the result is actually useful. Revise the approach, not just the answer.

5. Approve the process before building. Ask AI to summarize the proposed instructions so you can review them. Once they reflect what you want, have it create the Skill.
6. Test, correct, and keep testing. Never assume a Skill works correctly because AI created it or because it worked once. Test it on real work to uncover missed instructions, gaps, and results that fall short. In my experience, some problems only surface later, as I continue using the Skill. When that happens, I modify the Skill itself and test it again to check that the correction works without disrupting what already worked.

Start with this:

Preserve Your Best Thinking. Challenge the Rest.

A Skill can help me work faster, but speed is not its most interesting benefit. When I build challenge into the process, it can help me question familiar conclusions, discover unexpected connections, and develop ideas I would otherwise leave unexplored.

Refining the Skill also makes me examine the thinking behind my instructions. Am I asking for something that improves the work, or simply something that feels familiar? I want to preserve my best judgment without making yesterday’s thinking the limit of what I create next.

That is the greater opportunity: not just producing more, but strengthening our thinking and developing original ideas and insights that benefit the people we serve. Inside the AI Insider Lab, we use real business challenges to put that kind of thinking into practice.

Explore the AI Insider Lab

This Week’s Carousel

I Built an AI Skill to Challenge My Thinking