What it means
Prompt engineering is the practice of writing and refining the instructions you give an AI model so it produces the result you actually want.
A prompt is whatever you type into an AI tool. Prompt engineering is the craft of shaping it: what to ask, what background to include and what the answer should look like. A large language model only sees the text in front of it. When a request is vague, it fills the gaps with the most average guess.
Think of it like briefing a temp on their first morning. “Sort out the files” gets you something. “Put last year’s invoices in date order, one folder per client, and flag any without a signature” gets you what you needed. The temp is just as capable either way. The difference is the brief.
The word “engineering” does some heavy lifting. For most people, it just means writing clearly. At the technical end, it means testing many versions of a prompt and keeping the one that fails least. Here’s the everyday version.
| Weak prompt | Stronger prompt | What changed |
|---|---|---|
| “Write an email to my landlord.” | “Write a short, polite email asking my landlord when the broken boiler will be fixed. It’s been ten days.” | Purpose and facts |
| “Summarise this report.” | “Summarise this report in five bullet points for a manager who hasn’t read it. Deadlines first.” | Audience, length and priorities |
| “Give me dinner ideas.” | “Five vegetarian dinners I can cook in 30 minutes. No mushrooms.” | Clear limits |
| “Is this contract OK?” | “List any clauses that limit my right to cancel. Quote each one, explain it simply, and say if you’re unsure.” | A specific job and permission to be unsure |
Where it came from
The word “prompt” already meant a cue, as in a writing prompt or a computer’s command prompt. When text-generating AI arrived, the text you fed it became the prompt too.
The usual account is that “prompt engineering” took hold around 2020, as early users of large language models noticed how much wording changed the output. Adding a few worked examples before the real question could turn a poor answer into a good one. Later, simply asking a model to reason step by step was found to help with maths and logic.
Then chatbots went mainstream, and 2023 became the year of the “prompt engineer”. Headlines described a new job that needed no coding. A few well-publicised listings offered salaries in the hundreds of thousands of dollars. Plenty of people were sceptical, and the term picked up an ironic edge it still has.
As models improved, the magic-words side faded. Newer models follow plain instructions well. From around 2025, developers increasingly said “context engineering” instead. The hard part, they argued, isn’t phrasing one request. It’s choosing everything the model sees: the conversation so far, its tools, its standing rules, and any documents it fetches through retrieval-augmented generation.
How people actually use it
- “Can someone good at prompt engineering fix the support bot? It keeps apologising for everything.” Work chat. Fixing the instructions behind a product.
- “Prompt engineer: the new job that pays six figures and needs no coding.” A 2023-style headline. The hype version.
- “I’m a prompt engineer now, which means I say please to the computer.” Social post. The ironic version.
- “Our system prompt went through forty rounds of prompt engineering before launch.” Engineering blog. The serious, testing-heavy sense.
- “Honestly, prompt engineering is just writing a decent brief.” A common view from people who find the term inflated.
- “Nobody says prompt engineering any more. It’s context engineering.” Developer talk from 2025 onwards.
In a sentence
Priya: The bot writes three paragraphs when people ask for our opening hours.
Sam: Tell it to answer in one sentence unless asked for more. That’s half of prompt engineering.
Jon: My manager wants me to do a prompt engineering course.
Ade: Save the money. Say what you want, who it’s for and what good looks like.
Common misconceptions
- “There are secret magic words.” Some phrasings help, but there’s no password. With current models, clear goals and relevant detail beat clever tricks.
- “You need to know how to code.” Not for everyday use. The core skill is explaining a task well.
- “A good prompt guarantees a correct answer.” It improves the odds. Models can still make things up, which is why hallucination needs its own safeguards.
Related terms
- System prompt: the standing instructions a company gives its AI before you type anything.
- Context engineering: choosing everything a model sees, not just the wording of the request.
- Few-shot prompting: showing the model a few examples of what you want first.
- Chain of thought: getting a model to work through a problem step by step before answering.
- Prompt injection: hiding instructions in text an AI reads, to hijack what it does.
Questions people ask
Is prompt engineering still a real job?
Some roles still carry the title, mostly at AI companies. More often, prompt engineering is now a skill listed inside other jobs.
What makes a good prompt?
A good prompt says what you want, who it’s for and what the finished result should look like. An example of the output helps, and so does telling the model it’s fine to say it doesn’t know.
Is prompt engineering the same as vibe coding?
No. Vibe coding means building software by describing what you want and letting AI write the code. Prompt engineering is the wider skill of writing good instructions for any AI task.
What is the difference between prompt engineering and context engineering?
Prompt engineering focuses on how a request is worded. Context engineering covers everything a model receives alongside it, such as documents, earlier messages, tools and standing rules.
Do I need to learn prompt engineering to use ChatGPT?
You don’t need a course to use ChatGPT or similar chatbots well. Everyday prompt engineering mostly means being specific, giving background and following up when the first answer misses.
The short version
Prompt engineering is the skill of wording requests to an AI so the answers come back useful. It was sold as a glamorous career in 2023, then settled into something ordinary: clear writing, good examples and the right context. A good brief still beats a vague one.