Stop calling it 'prompt engineering'. It's just writing.
We invented a job title for the act of being clear. Then we put it on resumes. The good news is the skill is real. The bad news is, we already had a name for it.
I've been on hiring committees for three years. Every fifth resume that crosses my desk in 2026 says "prompt engineering" somewhere. Sometimes under skills. Sometimes as a job title. Once or twice as a self-applied credential, the same way someone with a piano in their living room puts "musician" on their LinkedIn.
I want to be honest. I cannot remember the last time a candidate could finish a coherent answer to "what is prompt engineering, actually?" without me feeling like we were both pretending.
That's not the candidates' fault. The phrase has eaten itself. It started as a joke about how funny it was that prompts mattered. It became a job posting. Then a course. Then a credential. Then a resume keyword. The thing the phrase pointed at is real — the activity is real — but the word "engineering" did something quiet and bad to it.
What you're actually doing when you "engineer" a prompt
You're doing one of two things.
You're being more specific than your normal speech allows. That's writing.
You're giving the model the context it needs to do the work — examples, constraints, format hints, the audience for the output. That's also writing. Specifically, it's the kind of writing a good editor or a good product manager does for a living. It used to be called writing a brief.
There's no third thing. There's no secret syntax. There's no incantation. There are tactics that help in specific tools — temperature, top-p, a system prompt with hardcoded rules — but those are the configurations, not the craft. The craft is the sentences in the middle of the message, and the craft is older than the tool by about a thousand years.
The reason this matters: when we name something "engineering", we imply that it has the structure of engineering. That it's learnable through a textbook. That there are right answers and wrong answers. That you should hire for it like you hire for a Java developer.
None of those things are true. Prompting is closer to teaching. It's closer to writing a memo. It's closer to phrasing a question for a smart but very literal junior employee. And the people who are good at it are mostly the people who were already good at being clear before any of this happened.
"You are a world-class marketing copywriter with 20 years of experience. Take a deep breath. Step by step, write 3 headlines for our SaaS product that maximize engagement and reflect Q4 priorities. Be creative."
32 words of incantation. The model still has no idea what your product does, who it's for, or what "engagement" means here.
Written"We sell automated invoice reconciliation to mid-market finance teams. Write 3 LinkedIn headlines aimed at CFOs whose teams currently reconcile by hand. Keep each under 12 words."
Specific audience. Specific outcome. Specific constraint. The model has everything it needs because you wrote like a person.
If the prompt requires engineering to make it work, the model probably can't do the task anyway. The advanced prompting techniques you're learning are mostly a way to push slightly past the model's real ability — and they fail silently when the underlying capability isn't there.
A simple test
If I write you a one-sentence prompt and you understand exactly what I want, the model probably will too.
If you have to ask me three clarifying questions to understand my prompt, the model will silently make up answers to all three. You will not know which ones.
The fix is not better engineering. The fix is to write the prompt the way you'd brief a junior person. What's the goal. Who's the audience. What does done look like. Here's one example of the right shape.
This is the same skill we've always wanted from senior individual contributors. The same skill we want from technical writers. The same skill we want from any product manager whose product manager actually knows what they're building. The AI hasn't created a new skill. It has just made the absence of this old skill suddenly, brutally legible.
What we should hire for instead
We should hire for: can this person write a clear two-paragraph brief. Can they look at a vague request and find the three questions to ask back. Can they take a piece of unfamiliar text and quickly distill what's important.
These are skills you can test in twenty minutes. They are also the same skills the best writers, editors, journalists, and senior engineers have had for decades. There is no shortcut to them. AI doesn't make them obsolete. It makes them more obvious.
The brief-to-prompt swap
The advice nobody wants to hear
If you're struggling to get the model to do what you want, you don't need a course on prompt engineering. You need to:
- Write down what you actually want, in plain English, without using bullet points to disguise vagueness.
- Read it back. Notice the parts where you said "good" or "high quality" or "well structured" without saying what those mean.
- Replace those vague parts with examples. One good example is worth a page of adjectives.
- Send it.
That's it. That's prompt engineering. There is no next step. There is no advanced technique. The thing that's actually rare and valuable is steps one through three — and we don't need a job title for it. We need an English class.
What to put on your resume instead
If you've been writing "prompt engineering" on your resume, here's a friendly suggestion. Replace it with the actual thing.
"Designed an LLM-powered customer support workflow that cut response time 40%" is a real skill. It tells me you built something. It tells me you measured it. It tells me you can describe both halves.
"Prompt engineer" tells me you wrote some prompts. Everybody writes prompts now. Your dentist writes prompts.
The phrase will have a normal lifecycle. It will be hot for another year. Then it will become slightly embarrassing, the way "blockchain expert" is now. The people who'll look smart in 2028 are the ones who, in 2026, were quietly building things and describing them precisely — not the ones who optimized their resume for an AI ATS scanning for the word "prompt".
The skill is real. The job title isn't.