From User to Builder: What AI Actually Demands of You
A collaborative Q&A by Lee Whitmore and Muhammad Ziyan
A line gets drawn between AI users and builders. Consumers ask AI to summarise and generate; builders design systems and compound advantage. Using AI as a tool is becoming the floor, not the ceiling.
Lee is a leadership coach and writes about AI reshaping organisations. His interest is what these tools ask of the person building them: thinking, judgement, leadership capacity no workflow can replace. https://substack.com/@leveluplee
Muhammad is a clear, concise thinker who helps founders and builders make sense of AI without the noise. Through 60 Words of AI, he brings a practical, stripped-back perspective on what is happening in the field and why it matters. https://substack.com/@muhammadziyan
Both believe AI output is only as good as what you bring. This is our attempt to be honest about what that looks like.
Q1: We have both argued 2026 marks a shift - from using AI to building with it. What has that shift required?
Lee:
More than expected. Tools are easier, but that surfaces a harder problem: you must know what you are building before you start. The temptation is reaching for AI with a vague idea and letting it fill the shape. The output looks convincing. It is usually hollow.
Getting specific before opening AI is what the shift has required. Not a topic, but a clear position: something I believe someone else might push back on. Once that exists, AI becomes useful. Without it, AI produces average thinking at speed.
Muhammad:
For me, the biggest shift wasn’t learning new AI tools. It was changing how I think before using them.
When I started, I believed better prompts were the answer. Now I spend far more time deciding what I’m trying to build than writing the prompt itself.
Building with AI means taking responsibility for the outcome. The AI can generate, research and execute, but it cannot decide what deserves to exist. That part still belongs to you.
Q2: We have both said AI output is only as good as what you bring. What does that require before opening the tool?
Lee:
Rough notes without the tool. Even a few lines. Awkward, incomplete, but mine. Those notes must contain an actual position - not a topic - before anything else. If I cannot articulate what I think in one sentence, the AI finds consensus and hands it back looking authoritative.
That is the most dangerous output: confident, well-structured, empty. Everything after is negotiation between roughness and fluency. The roughness matters. Where thinking lives.
Muhammad:
I start with one question: What’s the one idea I want someone to remember after reading this?
If I can’t answer that in a sentence, I don’t open AI.
I rarely start with the headline. I start with the argument. Once that’s clear, the headline almost writes itself.
AI is much better at helping me explain an idea than helping me discover one.
Q3: Agents went live this year. Work is automated at role level, not just task level. What does that mean for humans still in the room?
Lee:
It raises the floor on what leadership means. When AI handles execution, people are left for judgement: deciding what to build, why it matters, holding a team together. Those are not soft skills. They are the hardest skills.
Leaders I work with are discovering AI has not made their jobs easier. It has made non-automatable parts more visible. The gap between a leader who can navigate ambiguity and one who cannot is getting wider.
Muhammad:
It means your expertise becomes more valuable, not less.
Agents can execute. They can’t decide what’s worth executing. They can’t tell you when the system is solving the wrong problem.
The humans still in the room are there for judgement. That requires real domain knowledge. People without it will struggle to direct agents meaningfully.
Q4: The internet is filling up with content that sounds fine and means very little. What is our defence against contributing to that?
Lee:
Reading it out loud. Every time. AI-shaped prose is smooth and even in a way human writing rarely is. When it flows too easily, without hesitations or deliberate pauses, that signals too much of the tool’s fingerprints remain.
The other defence is publishing less. The pressure to produce is why writers reach for AI as a shortcut rather than scaffold. I would rather publish one piece that has something to say than three that fill space.
Muhammad:
One question before publishing: could anyone else have written this?
If yes, it’s not ready.
AI can copy a writing style. It cannot copy lived experience or a genuine opinion. That’s what I try to put into everything. Simplicity is just the format. The thinking underneath has to be mine.
Q5: What do we know now about working with AI that we wish we had understood at the start?
Lee:
It is a mirror before it is a tool. The first thing it reflects is the quality of your thinking, and that reflection is uncomfortable if thinking is not yet clear. I spent early months improving prompts. That helped, but improving clarity was the bigger lever.
The editing pass is where work happens. I used to treat editing as a final check. Now I treat it as primary authorship. The AI drafts; I write.
Muhammad:
I wish I’d understood that AI rewards clarity more than complexity.
I spent months learning prompt frameworks, templates and tricks. Most of them became irrelevant once the models improved.
The lasting skill wasn’t prompting. It was thinking clearly enough to give AI something worth expanding.
Today I spend less time asking better questions to AI and more time asking better questions to myself.
Closing Thoughts
AI does not remove the need for judgement. If anything, it makes judgement more visible, because the machine can move quickly only after a human has decided what matters.
That is the shift we keep coming back to: from using AI well to building with it deliberately. The difference is not in speed or output alone, but in judgement, clarity, and the willingness to bring a point of view. That is what will separate the people who simply use these tools from the people who actually shape what they make with them.
About the Authors
Lee Whitmore is a leadership coach, author, and podcast host specialising in AI, strategic disruption, and major change programmes. His book, Enhanced Leadership, is available at https://mybook.to/EnhancedLeadership Find his newsletter at:
Muhammad Ziyan writes 60 Words of AI on Substack, covering everything happening in AI in 60 words or less. He writes for founders and builders who want AI made simple. Find his newsletter at
