Prompt
Last updated June 10, 2026
What is Prompt in simple terms?
In simple terms, a prompt is whatever you ask an AI to do — the question or instruction you type in. It's your half of the conversation, and how clearly you word it shapes how good the reply is.
What is Prompt?
A prompt is the input you give an AI system — usually the question, instruction, or piece of text you type or speak — that tells it what you want and that it responds to.
A prompt is simply what you give an AI to get it going. When you type "explain photosynthesis to a ten-year-old" into a chatbot, that sentence is the prompt. When you ask an image generator for "a watercolor of a harbor at dawn," that line is the prompt too. It's the input the model responds to — the request, the question, the instruction, the starting text — and everything the AI produces is its attempt to give you a fitting continuation or answer to what you provided. In the simplest terms, the prompt is your side of the exchange, and the model's output is the other side.
What surprises many people is how much the wording of a prompt shapes the result. The same underlying model can give a vague, generic answer or a genuinely useful one depending entirely on how the request is framed — how much context you include, whether you specify the audience, the format, the length, or the tone you want. A bare "write a birthday message" yields something bland and forgettable; "write a short, warm birthday message for my 70-year-old dad who loves fishing and bad puns" yields something he might actually keep. The model didn't get smarter between those two requests. The second prompt simply gave it far more to work with. This sensitivity to phrasing is exactly why the deliberate craft of writing good prompts — prompt engineering — became a recognized skill.
It helps to know that a prompt is usually more than just the words you type. Many AI applications quietly wrap your input inside additional hidden instructions — a system prompt set by whoever built the tool — that shape how the model behaves before your request even arrives. In a longer conversation, the prompt the model actually receives often includes the earlier back-and-forth too, so it can stay on topic. All of that has to fit within the model's context window, the limit on how much text it can consider at once. But from your point of view, the prompt is the everyday thing: the question or instruction you hand the AI, and the clearer and more specific you make it, the better the odds of getting back what you actually wanted.
Real-world example of Prompt
A home cook stares into a half-empty fridge and types into an AI assistant: "recipe?" Back comes a generic pasta dish needing three things she doesn't have. She tries again with a fuller prompt: "I have eggs, spinach, half an onion, some cheddar, and bread. Suggest a quick dinner for one using only those, ready in 15 minutes." This time she gets a genuinely usable cheesy spinach scramble on toast. Same AI, same fridge — the only thing that changed was the prompt. The first gave the model almost nothing to go on; the second handed it the ingredients, the constraint, and the goal, so it could actually help.
Related terms
Frequently asked questions about Prompt
What is the difference between a prompt and prompt engineering?
A prompt is the actual input you give an AI — the words you type or speak. Prompt engineering is the skill of crafting those inputs well: deciding what context to include, how to frame the task, and how to phrase the request so the model returns what you want. So the prompt is the thing itself, and prompt engineering is the practice of writing good ones. Everyone who uses AI writes prompts; prompt engineering is doing it deliberately and skillfully.
How does a prompt work?
When you submit a prompt, the model reads it as a sequence of tokens and generates a response that it predicts best fits what you've asked. Often your visible prompt is combined behind the scenes with hidden instructions from the app and the earlier part of the conversation, and the whole bundle is what the model actually responds to. The model isn't looking anything up by default — it's producing an answer shaped by your wording, which is why two differently phrased prompts can lead to very different results.
What makes a good prompt?
Clarity and useful detail. A strong prompt says what you want, gives the model relevant context, and specifies things like the audience, format, length, or tone when they matter. Telling the AI what role to take, breaking a complex request into steps, and showing an example of the output you're after all help. The goal isn't to be wordy — it's to remove guesswork. Vague prompts force the model to fill in the blanks, and it often fills them in differently from what you had in mind.