08/26/2026
Prompt Engineering May Be Fading. Prompt Thinking Isn't
Jackie Kiadii, MCT
While researching for a Copilot class I was teaching, I came across the same claim from several people: prompting doesn't matter anymore because AI models are getting better. There's some truth to it.
Modern generative AI tools are much better at understanding natural language, inferring intent, engaging in conversation, asking clarifying questions, and identifying some of the steps needed to complete a task. As a result, business professionals probably don't need to memorize elaborate prompt formulas, magic phrases, or complicated "act as a…" scripts just to get useful results anymore.
But that doesn't mean prompting stopped mattering. It means the skill is changing.
Clarity is still critical. Unclear requirements waste your time and the model's computational energy, leading to more back-and-forth, more rework, and more time spent fixing assumptions that could've been caught up front.
And they produce worse output. Vague in, generic out. Your result ends up looking like everyone else's, or worse, the dreaded "AI slop."
Compare these two prompts:
Option 1
Analyze this spreadsheet and tell me what you see.
Option 2
I'm preparing for a 30-minute leadership meeting about Q2 sales performance. Analyze this workbook, focusing on the insights a VP of Sales needs to understand performance and decide where to prioritize follow-up. Keep the response concise, and flag any assumptions or data limitations I should verify before presenting.
One of these gets you a generic summary that you will spend precious minutes going back and forth to refine. The other gets you output you could walk into the meeting with.
Whether you're using Copilot, ChatGPT, Claude, Gemini, or whatever comes next, the tool matters less than what you bring to it.
Prompt engineering asks, "What words will make AI perform better?" Prompt thinking asks, "Have I defined the requirements well enough that a human or an AI could do this well?"
That second question is the one good managers have always asked before handing off work.
Better upfront thinking won't guarantee a perfect first response every time. Iteration is still worth doing, especially when you're brainstorming, exploring strategy, or figuring out what you really want as you go. But a clearer starting point means fewer wasted follow-ups either way.
The Bottom Line
When it comes to AI adoption, the organizations getting the most out of Copilot and similar tools aren't just training people to use the technology. They're teaching people to think clearly, define the work, and stay accountable for the result.
P.S. - I'm publishing a follow-up with 5 questions that will help you define requirements. Feel free to follow along if that's useful.
hashtag
