I have no idea how to put this into words.
Or: the accidental prompt, and why it worked perfectly.
It was 9:55 on a weeknight. I’d been building AI prompts all day. Carefully crafted ones — structured, specific, layered with context and intent. The kind that make an AI sit up straight and deliver exactly what you asked for.
I was wrapping up a particularly productive session when I noticed something I’d been meaning to address for a while. I kept writing really good prompts and then losing them. Using them once and never finding them again. Reinventing the wheel, week after week.
So I typed this:
“I have no idea how to put this into words so that I can prompt an AI professionally. I’m building prompts all day long and sometimes I feel I’m repeating myself, so I want to build a library of the really well written prompts that I will use over and over.”
The AI’s response came back immediately.
“Ha! The person who has been building prompts all day, who just received a polished prompt store, who ran a beautifully written meta-prompt across an entire session — captured her need for a prompt library in the most un-prompted way possible.”
Reader, I laughed out loud.
Because it was completely true. And I hadn’t noticed.
What I actually did — without realising it
Let me tell you what that message contained, even though I wrote it in thirty seconds without thinking:
It told the AI what I was doing — building prompts all day long. It described the problem — repeating myself, losing the good ones. It explained what I wanted — a library I could return to. And it conveyed why it mattered — efficiency, quality, not starting from scratch every time.
Context. Problem. Goal. Stakes.
That’s a well-structured prompt. I wrote it in plain English at five to ten at night while thinking I had no idea what I was doing.
The thing you think disqualifies you — not knowing the “right” way to ask — is often exactly the right way to ask.
The myth of the professional prompt
There’s an entire industry that has sprung up around “prompt engineering.” Courses, certifications, job titles, LinkedIn posts full of arcane syntax and capital-letter instructions. ALWAYS RESPOND IN BULLET POINTS. ADOPT THE PERSONA OF A SENIOR CONSULTANT. THINK STEP BY STEP BEFORE ANSWERING.
Some of that is genuinely useful, in specific technical contexts, for people building complex automated systems.
But for the rest of us — the people who just want to get things done, think better, write faster, solve problems — it’s mostly noise. And worse than noise: it’s a barrier. It makes people believe they’re not qualified to use a tool that asks only one thing of them.
A sentence. A thought. An honest description of what they need.
The AI doesn’t care whether you’ve taken a course. It cares whether you’ve given it enough to work with. And “enough to work with” looks a lot less like a structured framework and a lot more like how you’d explain something to a smart friend over coffee.
What actually makes a prompt work
In my experience — and I’ve written a lot of them — the prompts that work best have three things in common.
They’re honest. They say what’s actually needed, not what the person thinks sounds impressive or technical. “I’m not sure how to approach this” is useful information. “I need help structuring my thinking” is a perfectly good prompt opener.
They have context. Not a wall of background — just enough that the AI knows who it’s talking to and what the situation is. A sentence or two, usually. What’s happening, and why it matters.
They’re specific about the outcome. Not “help me with my email” but “help me write an email to a client who missed a deadline, firm but not unkind, keeping the relationship intact.” The more clearly you can describe what done looks like, the better the response.
That’s it. No certification required. No special syntax. Just honesty, context, and a clear picture of what you’re hoping for.
Which, come to think of it, is also the recipe for a good conversation with a person.
You already know how to do this. You’ve been doing it your whole life. You just haven’t applied it here yet.
The prompt I accidentally wrote
I went back and looked at my 9:55pm message with fresh eyes.
“I have no idea how to put this into words” — that’s honesty. It signals uncertainty, which gives the AI permission to help shape the thinking rather than just execute an instruction.
“I’m building prompts all day long” — that’s context. It tells the AI who I am in relation to this problem. Not a beginner. Someone with a specific, professional need.
“I want to build a library of the really well written prompts that I will use over and over” — that’s the outcome. Clear, specific, actionable.
Thirty seconds. No planning. No framework. Just a tired person at ten to ten being honest about what they needed.
And the AI knew exactly what to do with it.
So here’s your permission slip
Stop waiting until you know the right way to ask.
Start with what’s true. What are you trying to do? What’s getting in the way? What would done look like? Type that — in whatever words come naturally — and see what comes back.
You might be surprised to discover that the thing you thought was an awkward, unprofessional half-thought was, in fact, everything the AI needed.
I was building prompts all day and the best one I wrote that session came out at 9:55pm when I’d stopped trying.
It’s just a conversation. You already know how to have one.
Sandi is a Melbourne-based problem-solver, crisis-averter, and translator of the technical into the human. She spent decades being the person everyone called when something was broken, confusing, or just needed explaining properly — earning a reputation that preceded her wherever she went. Now she’s channelling that same instinct into AI: making it accessible, practical, and genuinely useful for people who think it isn’t for them.