A prompt starts a conversation with an AI assistant. Four areas help make that conversation useful: persona, task, context and format.
The four main areas
1. Persona
Define who the AI should be. Example: “You are a programme manager in healthcare…”
2. Task
Describe what you want the AI to do. Example: “Draft an executive summary email…”
3. Context
Provide relevant background information. Example: “Based on Q3 programme documents and recent feedback…”
4. Format
Specify how you want the output structured. Example: “Limit the response to bullet points…”
Example prompt using all four
You are a programme manager in healthcare. Draft an executive summary email to stakeholders based on Q3 programme documents and recent feedback. Limit the response to bullet points.
You do not need all four elements in every prompt. Using several of them usually helps. Include a verb or command in the task so the requested action is clear.
Quick tips
1. Use natural language
Write as if you are speaking to another person. Express complete thoughts in full sentences.
2. Be specific and iterate
Tell the AI what you need it to do: summarise, write, change the tone or create something. Provide the context needed to make those instructions meaningful.
3. Be concise and avoid complexity
State your request in brief, specific language. Avoid unnecessary jargon.
4. Make it a conversation
Refine your prompt when the result misses the mark. Follow-up instructions can identify what to preserve, change or remove.
5. Use your documents
Ground the output in your own material. Name or attach the documents the AI should use.
6. Ask the AI to edit the prompt
Try: “Improve this prompt, identify any missing context and explain each change: [original prompt].” Review the revision before using it.
Advanced techniques
1. Provide context
Before:
Tell me about climate change.
After:
Explain three major impacts of climate change on agriculture in tropical regions, with examples from the past decade.
2. Show examples
Before:
Convert this technical statement to plain language.
After:
Here are two examples of converting jargon into plain language:
- “Utilise familiar design paradigms” → “Use familiar design patterns”
- “Commence the authentication process” → “Sign in”
Now convert: “The platform implements end-to-end encryption.”
3. Break complex tasks into steps
Before:
Analyse this quarterly sales data.
After:
Analyse this quarterly sales data by:
- identifying the top-performing products
- comparing the current quarter with the previous quarter
- highlighting unusual patterns or trends
- suggesting possible reasons for those trends
4. Ask it to examine the problem first
Before answering, identify the relevant evidence, constraints and plausible approaches. Then recommend an option and explain the trade-offs.
5. Define the AI’s role
Explain how rainbows form from the perspective of an experienced science teacher speaking to a curious 10-year-old.
6. Assign a role for creative work
Try: “You are the head of a creative department at an advertising agency…” Then add the audience, objective, source material and constraints that govern the work.
7. Give constraints
Include measurable requirements such as a character limit, number of options or required details: “Generate five options, each under 100 words.”
8. Specify tone when it matters
Name the qualities the writing needs: formal, informal, technical, casual, professional or friendly. Concrete examples often communicate tone more precisely than labels alone.
The golden rule
Show your prompt to a colleague who has little context on the task and ask them to follow it. If the instructions confuse them, the AI will probably struggle too.
Ask the AI for help with prompting
When you are unsure how to frame a request, try:
I am trying to get your help with [goal]. What information and constraints do you need from me to produce a useful result? Then help me draft the prompt.
This exposes missing context before the assistant starts the work.
Put the method to work
Use Key Prompt Indicator to examine how different roles and tasks are framed. For image direction, the Midjourney prompt builder turns scene, camera, style and format decisions into one structured prompt. When a project needs wider direction, start with the creative brief builder before drafting individual prompts.