What Is a Prompt? A Plain-English Explanation With Examples

A prompt is the message you give an AI tool to tell it what you want: a question, an instruction, some text to work on, or a mix of all three. When you type "Summarize this email in three bullet points" into a chatbot, that sentence is the prompt, and the summary you get back is the response. The clearer your prompt, the more useful the response usually is. This guide explains what a prompt is, what a good one contains, and gives examples you can copy and adapt.

What does a prompt look like?

A prompt can be as short as a few words or as long as several paragraphs. All of these are prompts:

• A question: "Why is the sky blue?"
• An instruction: "Write a thank-you note to my neighbor for watering my plants."
• Text plus a task: "Here is my cover letter. Point out any sentences that sound stiff: [paste text]."
• A description for an image tool: "A watercolor painting of a lighthouse at sunrise, soft pastel colors."

There is no special syntax to learn. You write in normal language, the same way you would explain a task to a helpful person who knows nothing about your situation.

Why does the prompt matter so much?

An AI chatbot only knows what is in its training and what you tell it in the conversation. It cannot see your screen, read your mind, or know who the answer is for unless you say so. When a prompt is vague, the AI fills the gaps with generic guesses. When a prompt is specific, it has less to guess.

Compare these two:

Vague: "Write something about exercise."
Specific: "Write a 100-word encouraging message for a coworker who just started walking at lunch. Keep it friendly, not preachy."

The second prompt tells the AI the audience, the length, the purpose and the tone. You will almost always get something closer to what you need.

What are the parts of a good prompt?

Most strong prompts include some of these ingredients. You rarely need all of them at once.

1. Task. What you want done: explain, draft, summarize, list, compare, rewrite, translate.
2. Context. Background the AI cannot know: your situation, your goal, what you have already tried.
3. Audience. Who will read or use the result: a child, a manager, a customer, yourself.
4. Format. How the answer should look: bullet points, a table, an email, a step-by-step list.
5. Constraints. Limits such as length, tone, reading level, or things to avoid.
6. Example. A sample of the style or format you want, if you have one.

A quick way to remember it: what, for whom, in what shape, and with what limits. For eight specific techniques with before-and-after examples, see How to write better AI prompts.

What are the main types of prompts?

Question prompts

You ask something and expect an explanation. Example: "What is the difference between weather and climate? Explain in plain language with one everyday example." Good for learning, but verify facts that matter.

Instruction prompts

You tell the AI to produce something. Example: "Write a short packing list for a two-night camping trip in cool weather."

Text-plus-task prompts

You paste content and say what to do with it. Example: "Summarize the meeting notes below into decisions, open questions and next steps: [paste notes]." Our guide on summarizing long documents with AI covers this in depth.

Role prompts

You ask the AI to take on a perspective. Example: "Act as a friendly job interviewer for a retail position. Ask me one question at a time and give feedback after each answer." A role steers tone and focus; it does not give the AI real expertise or credentials.

Example-based prompts

You show one or more samples and ask for more in the same pattern. In technical writing this is sometimes called "few-shot" prompting. Example: "Here are two product descriptions I like: [paste]. Write one for a canvas tote bag in the same style."

Step-by-step prompts

You break a bigger job into stages or ask the AI to work through a problem in order. Example: "First, list the questions I should answer before choosing a laptop. Wait for my answers before recommending anything."

Follow-up prompts

Every message after your first one is also a prompt. Follow-ups like "make it shorter," "use simpler words," or "give me three other options" are often where the real improvement happens, because the chatbot remembers what was said earlier in the same conversation.

What are some prompt examples for everyday tasks?

Copy these, then replace the bracketed parts with your own details.

Writing an email

Write a polite email to [recipient] asking to move our meeting from [day/time] to [day/time]. Keep it under 100 words and offer one alternative time.

Understanding something confusing

Explain [topic] to someone with no background in it. Use one everyday analogy, keep it under 200 words, and list two questions I could ask to learn more.

Making a plan

I want to [goal] in [time frame]. I have about [amount of time] per week. Create a simple weekly plan as a checklist, and flag any step where I should get expert advice.

Studying

Here are my notes on [subject]: [paste]. Make 10 flashcards in a question-and-answer format, then quiz me one question at a time.

More study ideas are in how to use AI to study.

Creating an image

A [style, e.g. flat illustration] of [subject] in [setting], [colors or mood], [composition, e.g. centered, wide shot].

Image tools respond best to concrete visual details: subject, style, lighting, colors and framing. See AI image generators for beginners for more.

What is a system prompt?

Behind the scenes, AI providers give their chatbots background instructions, often called a system prompt, that shape how the chatbot behaves in every conversation. You do not usually see it. Many chatbots also let you add your own standing preferences, sometimes called custom instructions, such as "I am a beginner, so avoid jargon" or "keep answers brief." Where this setting lives and what it is called varies by product, so check the tool's help pages. For more terms like this, see our AI glossary for beginners.

Do you need to learn "prompt engineering"?

"Prompt engineering" is the practice of carefully designing and testing prompts, especially for apps and automated systems. For everyday use, you do not need a course or a list of magic phrases. Clear writing, a bit of context, and willingness to ask follow-up questions get you most of the way. Be wary of anyone selling "secret prompts" that promise guaranteed results.

What are common prompt mistakes?

• Too vague. "Make it better" does not say what better means. Say "shorter," "more formal," or "clearer for a beginner."
• Too many unrelated requests at once. Split them into separate prompts.
• Leaving out key facts. If the email needs a date, a name or a deadline, include it rather than letting the AI invent one.
• Giving up after one try. Treat the first answer as a draft and refine it.
• Trusting the output blindly. A good prompt improves answers but does not guarantee accuracy. Check names, numbers, quotes and sources. See how to check AI answers for mistakes.

What should you never put in a prompt?

Anything you type may be stored by the service, and depending on the provider and your settings it may be reviewed or used to improve the product. Keep these out of your prompts:

• Passwords, PINs and security codes
• Full financial account or card numbers
• Government ID numbers
• Private medical details tied to your name
• Confidential work documents, unless your employer has approved the tool

When you need help with something personal, remove or replace identifying details first. Our guide Is it safe to share personal info with AI chatbots? explains the settings worth checking.

Frequently asked questions

What is a prompt in AI, in one sentence?

A prompt is the text you give an AI tool, such as a question, instruction or pasted content, that tells it what to produce.

How long should a prompt be?

As long as it needs to be to remove guesswork. A simple question can be one line. A task with an audience, format and constraints might take three to five sentences. Longer is not automatically better; relevant detail is what counts.

Do prompts work the same way in ChatGPT, Claude, Gemini and Copilot?

The same basic habits work across all major chatbots: be specific, give context, ask for a format and refine with follow-ups. Each tool may respond a little differently to the same prompt, so if one answer misses the mark, rephrase or try another tool. If you are new, start with how to use ChatGPT for the first time.

Is it okay to be polite in prompts?

Yes. Polite wording does no harm, and writing in a natural, clear way tends to produce clear answers. What matters most is that your request is specific.

Can a prompt make the AI give a wrong answer?

A prompt can make errors more likely, for example by including a false assumption ("Why did X happen in 2019?" when it did not). AI tools may go along with the assumption. If you are unsure of a fact, ask the AI to check the premise first, then verify it yourself.

Key takeaways

• A prompt is anything you give an AI tool to tell it what to do: a question, instruction or content plus a task.
• Good prompts state the task, context, audience, format and constraints.
• Follow-up prompts are where most improvement happens.
• Keep sensitive personal and confidential information out of prompts.
• Even a great prompt does not guarantee accuracy, so verify what matters.