Set the Scene Before You Ask the Machine
Before AI tools became part of my everyday work, most of my writing time went into editing—tweaking, rewording, fixing tone. These days, it’s flipped. I spend more time upfront: drafting, brainstorming, finding sources, thinking through the angle. The polish comes later.
I'm not against using AI (unlike some purist writers), but I’m wary of letting it think for me. That’s because AI is good at producing average results. It fills in the blanks with whatever’s typical. If I want to be better than average—and I do—that means I still need to bring the thinking.
In a previous post, I laid out four ways to raise the bar when working with AI:
Define your voice
Write the first draft yourself
Go deeper with research and examples
Edit and refine together with AI
The first three all come down to one thing: context.
When I say “context,” I mean the stuff that surrounds the AI prompt—the problem you're working on, your goals, your tone, the audience, what you’ve already tried, what success looks like. Without that, AI is just guessing.
Context is Everything
Lately, more people are starting to talk about context when it comes to AI.
One quote that stuck with me comes from Nick Milo, who works in the world of knowledge management among other jobs. He put it simply in a recent newsletter:
The quality of your relationship with AI depends entirely on the quality of context you give it.
If AI doesn’t know what matters to you, it’s just going to play it safe. You’ll get whatever it can average out from the internet.
Think of it this way: working with AI is like handing off a task to a junior colleague or a freelancer. If you just say “write something” or “make this better,” you're setting them up to fail. But if you give them a clear, thoughtful brief, with goals, background, audience, and examples, they’ll almost always deliver something useful. AI’s the same.
That’s why writing things in your own words still matters. As Milo puts it, the value now is in what he calls the “notemaking”: thinking through your ideas, connecting the dots, shaping your own take. AI can help with the rest, but not if you skip that part.
Same thing in technical work. On the “This Day in AI” podcast, one of the hosts said they’ll often spend 15 minutes or more crafting a prompt when asking for code or a detailed answer. Why? Because the setup, or the framing, is what determines whether the output is usable or a mess.
Another phrase I’ve heard:
“If you want 1,000 words out, put 1,000 words in.”
And it tracks with my experience too. The more I feed in—problem specifics, goals, tone, background—the better the response.
Do a Warm-Up Round with AI Before Jumping In
If you’re doing something complex, such as building a business case, writing a talk, or tackling a messy strategy, it’s not always the best idea to start by asking AI to do it. That’s like handing over your to-do list to someone who’s never seen your job.
Instead, do a warm-up round. Use AI to help you think.
Here are a few ways to do that:
Define the problem
Start with a brain dump: your goals, rough ideas, insights, market research, blockers. Then prompt AI to help you clarify what problem you're actually solving.
List the assumptions
Ask AI to find 10 assumptions you might be making. You’ll be surprised how often this surfaces blind spots. It also sharpens your focus—and makes you look more prepared in meetings.
Use the “5 Whys”
Take one assumption or key idea and ask AI to run a “5 Whys” on it. It’s a simple but effective way to dig deeper and build stronger arguments or frameworks.
Ask what’s missing
After your initial inputs, ask: “What haven’t I considered?” or “What perspectives am I missing?” AI won’t get it perfect, but it can help broaden your view.
This warm-up isn’t fluff—it’s the thinking most people skip. And it shows in their output.
Bonus for you visual learners: Here’s a solid demo of this in action from another Youtuber
Treat It Like a Brief, Not a Magic Trick
Using AI without context is like briefing a junior employee with zero prep. Or worse, like giving a client project one vague sentence and expecting a perfect result. Good inputs = good outputs.
A simple example:
If you just say, “Summarize this financial report,” AI will do it, but on its own terms. It’ll guess what’s important, usually defaulting to a generic bullet list. Sometimes it gets it right. Often, it misses what you care about.
Now compare that to a prompt like this:
[Insert financial report document]Focus on: Q1 financial highlightsReview: Major revenue driversTop expense areasProfit trend
Output format: 2–3 sentence summary3–5 bullet points2–3 actionable recommendations
This prompt helps AI know what to focus on, what to ignore, and what kind of output you want. You’re setting the stage.
This is how to treat any serious task you hand over to AI; treat it like writing a creative brief. Whether you’re asking for code, content, or analysis, spend the extra 10–15 minutes defining:
The goal (what you’re trying to do)
The audience (who it’s for)
The format (how you want it structured)
The focus areas (what to zoom in on)
The “do nots” (things to skip or avoid)
You don’t have to write a novel—but you do need to give it enough. And yes, sometimes that means putting in as many words as you want out.
Also: if you hate typing, try dictation. Talking through your prompt can be faster and often clearer than writing it out. Tools like Microsoft Copilot Voice or your phone’s voice input can help you talk your way into better prompts.
Wrap-Up: Good Prompts Start With Good Thinking
AI can clean up your writing, generate ideas, and fill in blanks—but it can’t replace your judgment. Not yet. And definitely not if you’re handing over half-baked prompts and expecting gold.
Want better results from AI? Treat it like a creative partner. Start with a thoughtful brief, do a warm-up round, and don’t skip the thinking. Just like with real collaborators, clarity upfront saves you a lot of pain later.
Quick Takeaways
Don’t start with the output—start by understanding the problem.
Use AI to help surface assumptions and blind spots.
Be specific: tell it the goal, the format, and what to focus on.
Write prompts like briefs, not vague wishes.
Spend more time setting the scene—you’ll save time later.