Quick answer
A prompt is the input you give an AI model: a question, instruction or context. The quality of the prompt decides the quality of the output.
What it is
A prompt is the text you send to an AI model. It can be a question, instruction, example or context. System prompts set behaviour; user prompts are the actual request. Prompt engineering is the practice of designing prompts that produce reliable, useful outputs.
Why it matters
The same model can produce very different outputs depending on the prompt. Good prompts are clear, specific and grounded in examples.
How to use it
- Use a system prompt to set role, tone and constraints.
- Provide 3 to 5 examples for tasks that need consistent format.
- Specify the output format (JSON, markdown, length).
- Test prompts against edge cases and iterate.
Examples
- A system prompt for a senior partner voice: 'You are a senior partner at a digital studio. Calm, evidence-led, sentence case. Never use banned words.'
- A user prompt with examples: 'Classify the following support tickets into 3 categories. Examples: ...'