Prompt
An instruction for AI
A prompt is the text or instruction you give to generative AI or a language model to perform a task. It contains your question or request and often context, examples, and rules. The way you phrase it directly affects the quality and accuracy of the output.
Why the prompt matters
A language model does not read your mind; it works only with what you write in the prompt. So the same model can give you a shallow answer or a very precise one, depending on how well you phrase the request. The prompt is essentially how you steer the model.
A good prompt usually contains more than a bare question. It helps to tell the model what role to play, what to focus on, in what format you want the output, and what the constraints are. The less room you leave for guesswork, the more consistent the results.
How to write a good prompt
A good prompt can be put together with a few simple principles. You do not need to be technical; you just need to describe clearly and specifically what you need and refine the result over time.
- State the goal clearly: what exactly the AI should create or solve.
- Provide context: who the output is for, in what tone and style.
- Specify the format: length, structure, a list, or a table.
- Attach an example of the desired result if you have one.
The prompt in practice
Picture the difference between the instruction write a text about our company and a more detailed request stating that it is a short homepage text for small business owners, in a friendly tone, and no more than three sentences. The second request almost always produces a more usable result. Crafting effective prompts for recurring tasks is its own discipline, prompt engineering.
Want to learn to write prompts that work?
Explore our AI trainingFrequently asked questions
Do I need to be a programmer to write a prompt?
No. A prompt is written in plain language, so anyone can do it. More important than technical knowledge is the ability to describe clearly and specifically what you want and to refine the result based on what the model returns.
Why do I get a different answer to the same question each time?
Language models generate text with a degree of randomness, so the same prompt can lead to slightly different results. A more specific request with clear rules reduces this variability and makes the answers more consistent.
Is it worth saving prompts that work well?
Absolutely. If you find a prompt that reliably produces a good result, save it as a template. You will save time and your whole team will work consistently. This is exactly what systematic prompt engineering builds on.