Prompt
A prompt is the text you give an AI model to get a response: a question, an instruction, or a longer brief with background, rules, and examples. The model reads only what is in the prompt (plus any standing instructions), so the words you choose decide what comes back.
The prompt is your steering wheel. Better prompts get better results from the exact same model.
What does prompt mean in AI?
In everyday English, prompt means on time, or a cue that gets someone to act. In AI, the second sense stuck: a prompt is the cue that gets the model to respond. You type it into ChatGPT, Claude, or Gemini, or an app sends it for you behind the scenes.
A prompt can be one line ("Summarize this email") or several pages. Longer prompts usually carry more of the things a model cannot guess on its own: who the answer is for, what good looks like, and what to avoid.
What can a prompt contain?
Most useful prompts mix a few of these pieces. You rarely need all of them at once.
- An instruction or question: the thing you want done.
- Context: background the model needs, such as your business, audience, or goal.
- Input data: the email, document, or numbers you want it to work on.
- Examples: a sample or two of the output you want.
- Format and constraints: length, structure, tone, and anything to leave out.
Prompt examples
Three prompts for the same model, from bare to fully specified:
"What is a good name for a dog grooming business?"
"Suggest ten names for a mobile dog grooming business. Short, friendly, easy to spell."
"I am opening a mobile dog grooming van in Henderson, Nevada, aimed at busy families. Suggest ten names, under three words each, playful but not silly. Avoid puns on 'paws'. Here is one I like: Suds on Wheels."
Types of prompts
You will see these names for different kinds of prompts. Each has its own glossary entry.
- User prompt: what you type in each message.
- System prompt: standing instructions that run before every conversation and set the AI's role and rules.
- Zero-shot prompt: a task with no examples, just instructions.
- Few-shot prompt: a task with a handful of worked examples to copy.
- Chain-of-thought prompt: a prompt that asks the model to reason step by step before answering.
- Prompt template: a reusable prompt with blanks for the parts that change.
- Prompt chain: a series of prompts where each output feeds the next.
Prompt vs prompting vs prompt engineering
A prompt is the thing you write. Prompting is the everyday skill of writing prompts that get a useful answer, and our guide to prompting walks through the structure, before and after examples, and how to iterate. Prompt engineering is the same craft done systematically, with testing, for prompts that run again and again inside apps and agents.
Frequently asked questions
What is a prompt in AI?
A prompt is the text you give an AI model to get a response. It can be a short question, an instruction, or a longer brief with background, rules, and examples. The model works only from what the prompt and any standing instructions tell it.
What is an example of a prompt?
A simple prompt is "Suggest ten names for a mobile dog grooming business." A stronger one adds context and limits: who the business serves, how long the names should be, the tone you want, and one name you already like.
Is a prompt just the question I type into the AI?
Often, yes, but it can be much more than a single question. A prompt can include background context, input data to work on, rules to follow, and examples to copy, all of which steer the answer you get back.
What is the difference between a prompt and a system prompt?
The prompt is what you type in each message, while the system prompt is a standing set of instructions that runs before every conversation and shapes the AI's role, tone, and rules. You write the prompt each time; the system prompt is usually set up once.
What are the main types of prompts?
Common types are user prompts, system prompts, zero-shot prompts with no examples, few-shot prompts with a handful of examples, chain-of-thought prompts that ask for step-by-step reasoning, prompt templates with fill-in blanks, and prompt chains where each output feeds the next prompt.
How do I write a better prompt?
Give the model a role, the context it lacks, one specific task, the format you want, an example, and any limits, then refine the answer with specific corrections. Our step-by-step prompting guide covers each part with before and after examples.
New to all this? Start with what an AI agent really is, browse the full glossary, or explore the learning hub.
← Back to the glossary