You're using AI WRONG: Stop Delegating and Start Thinking!
Geoff Woods calls it an "AI Thought Partner ™️." I call it the fastest way to stop building the wrong thing.
📌 THE POINT IS: AI gets exponentially more useful when you stop asking it to produce answers and start asking it to sharpen your thinking — one question at a time — until your real objective reveals itself.
Like and Share this article!
Having a partner is way better than having a digital intern
In today's AI world, people have learned to ask their AI tool of choice to just execute something. This makes sense as early adoption patterns encouraged people to ask AI to write the first draft. However, in his book, “The Ai-Driven Leader,” Geoff Woods leans into advising readers to think of their favorite AI tool as a thought partner. When you tell the assistant to ask questions, it forces you to refine your point, get ready to write that first draft, and provide editorial guidance. This results in a much more inspired dialogue.
“How can AI help me do this?” - Geoff Woods
The innocent question that saved me hours of wasted effort
You can try this today: next time you're about to ask ChatGPT to build a business plan, strategic plan, coaching plan, etc., ask it to act as your thought partner and help you think through that activity. Give it background information, like you would any human advisor. Then tell it to ask you additional questions, one at a time, to help it fill in any other relevant blanks that you didn't think of (I recommend giving it a cap so that it doesn't get stuck in an endless question loop). Notice how different the process is; how many things pop into mind as you're answering the questions; and the quality of the output that it provides.
For me, this was very relevant last week when I spent about 5 minutes writing a prompt (it was a long prompt because of the significant context that I provided). After I asked my AI tool to ask me questions, one at a time to act as my thought partner and agent co-builder, it immediately began with a clarifying question:
“Let's be outcomes-focused here. What are the 3-5 outcomes you'd like this coaching agent to help you determine each week?”
I stared at the screen…
Then out of the blue, I had this epiphany. I didn't need an agent to tell me I should delegate more or protect calendar time for strategic thinking. I already knew that. I was doing this for the exercise of building an agent, but this agent wasn't going to have any ROI.
Had I not asked the AI tool to be my thought partner, it would have dutifully started telling me how to build this agent. But by asking it to help me think, it very innocently asked me a question that opened my eyes. After sharing the epiphany with the AI, it told me that we should not continue building the agent at all.
Based on its question, I asked it to help me review productivity in relation to my strategic plan for the year. I got meaningful ideas in a one-page report and the activity took far less time! It was also a much more strategic exercise.
“Stop looking at [AI] as an assistant and… start looking at it as a thought partner.” – Geoff Woods
Questions don't just improve prompts. They help leaders have those ah-hah moments that lead to inspiration
This idea of asking the AI agent to think about the context and activity first, then create its own questions for you to go deeper before presenting ideas is a strategic unlock. Like using ChatGPT or other tools in Study Mode, the AI tool now moves from a mindless answer machine that may be totally, confidently wrong to a thought-inducing tool which helps you discover insights you may have been overlooking.
In a recent exercise I did with ChatGPT, by asking it to help me think through a business problem in order to make a decision, the questions that it created for me were logical, changed based on my answers, and were mostly multiple choice in order to force me to decide (or give it a write-in answer but only if I had additional details to support it). The next question was always related to the answer that came before it, almost like the model was following a pre-defined decision tree that ultimately helped me realize where my own biases, preferences, and experiences were taking me.
It may be good to note here that using a prompting framework like C-R-A-F-T is particularly useful. Prompt frameworks help structure your thinking so you don't omit critical instructions to the AI.
How leaders can apply these techniques to up their game with their AI thought partner
Leaders can apply these immediately:
Use Thought Partner mode before you build anything. If the goal isn’t crisp, automation just scales confusion.
Force prioritization early. “Help me think about the 3 outcomes that matter most.” beats “build me an agent.”
Add a “tradeoff question.” “What must be deprioritized for this to work best - give me options to consider?” (prevents fantasy planning).
Switch from “more productivity” to “better alignment.” My ah-hah moment above is a great example of this.
Log the questions, not just the answers. Over time, your best questions become your leadership operating system.
Teach your team these frameworks and patterns. It’s a cultural upgrade: better questions → better thinking → better work.
AI is moving from being your assistant to being your mirror + coach + strategic lens
Thought partner mode with AI tools moves from blind robotics to creative interactions that unlock insights. It doesn't just improve output quality; it improves decision quality. The next time you're sitting down to try something just to try it, think about your time's ROI and ask the AI to help you take your use-case to the next level! You might be surprised what it suggests. Then you can ask it to be your thought partner to work through the next steps and notice how different the experience is working with it. This is more than a productivity hack; it's a decision-quality upgrade.
As always, thought-provoking insights Matt. Very timely. As engineers, we often default to building and delegating to tools as quickly as possible. Having someone consistently challenge the why and the time ROI forces a shift in mindset — from “can we build this?” to “should we, and what’s the best way?” That pause is where better decisions (and better use of AI) actually emerge.