Prompt
In shortA prompt is the input you give an AI model: your question or task, plus any instructions, examples and text it should use.
In plain words
A prompt is everything you send to a model in one request. It can hold a question, a task, rules for the answer, examples, and the text the model should work on. The model continues from there. Chat apps also add hidden instructions, the system prompt, before your text.
OpenAI, Anthropic and Google each publish prompt guides, and they agree on the basics:
- State the task clearly and say who the answer is for.
- Give the context the model cannot guess, such as the source text or your goal.
- Say what format you want: a list, a table, a length.
- Show one or two examples of a good answer when the style matters.
- Test the prompt, read the result, and adjust.
Why it matters
- Quality: The same model gives a weak or a strong answer depending on the prompt.
- Cost: A longer prompt has more tokens and costs more on every call. It also uses part of the context window.
- Safety: Do not put secrets or personal data into a prompt. See what never to paste. When a prompt includes text from outside, such as a web page, that text may carry hidden instructions (prompt injection).
- Control: Settings like temperature change how much the answer varies, but the prompt sets the direction.
Example
Weak prompt: "Write about our product."
Better prompt: "You write for small-business owners. Summarise the product text below in three bullet points, in plain language, without hype. Each bullet has at most 15 words. Text: …"
The second prompt names the reader, the task, the format and the limit. The model has less to guess, so the answer needs fewer rounds of correction.
Common misconceptions
- "There are magic words." No phrase works on every model. Clear instructions and good context work better than tricks.
- "Longer is better." Relevant is better. Padding costs money and can dilute the point.
- "A good prompt works forever." Models change. Re-test important prompts after a model update.
Questions and answers
What makes a good prompt?
A clear task, the needed context, the format you want, and an example when style matters. Then test it and adjust.
What is the difference between a prompt and a system prompt?
The prompt is what you send in a request. The system prompt is a standing instruction set by the app or developer that applies to the whole conversation.
Is prompt engineering a real skill?
Yes, in the sense that clear writing and testing improve results. The big providers publish guides for it. It is not a set of secret phrases.
Read next
- System prompt
A system prompt is a block of standing instructions, set by an app or developer, that shapes how a model behaves across a whole conversation.
- Prompt injection
Prompt injection is an attack where text a model reads, such as a web page or file, contains instructions that make it act against its user.
- Context window
The context window is the maximum amount of text, counted in tokens, that a model can read and write in one request.
- Temperature
Temperature is a setting that controls how random a model's word choices are: low gives steady answers, high gives more varied ones.
- What not to paste into AI chatbots
Passwords, client data, IDs and secrets do not belong in a chatbot: a traffic-light checklist, provider data rules, Swiss and EU basics, and safer options.