How to Write Better AI Prompts
The difference between a useless AI answer and a great one is usually the prompt. Five habits that reliably get better results, with examples.
Eddie Ochieng
July 3, 2026
People who get disappointing results from AI tools usually blame the tool. Nine times out of ten the problem is the prompt. A vague request gets a vague, generic answer. A specific, well framed one gets something you can actually use. Prompting is not a dark art, it is a handful of habits, and once they stick you will get far more out of every AI tool you touch.
The one principle
The core idea behind all of this, the model cannot read your mind. It only has the words you give it. Everything below is really just ways of putting more of what is in your head into the prompt.
Be specific about what you want
Tell the model exactly what you are after, including the format, length, and audience. A request like write about email marketing gets a bland essay. Compare it to this.
Write a 150 word intro for a blog post about email marketing, aimed at small business owners who are new to it. Keep the tone friendly and practical, and end with a question.
The second prompt names the length, the format, the audience, the tone, and how to finish. That is why it works.
Give it context
The model does not know your situation unless you tell it. Add the background that matters, who you are, what you are trying to achieve, and what you have already tried. A sentence of context often doubles the quality of the answer.
I run a small pottery studio and I am writing my first newsletter. My customers are hobbyists who took a class with me. Help me write a warm welcome email that tells them what to expect each month.
Assign it a role
Telling the model who to be shapes how it responds. Asking it to answer as an experienced editor, a patient teacher, or a sceptical reviewer changes the entire tone and depth of the reply. It is a simple trick that consistently lifts quality.
Act as an experienced copy editor. Review the paragraph below for clarity and flow, and explain each change you suggest so I can learn from it.
Show it an example
If you want output in a particular style, show it one. Paste an example of the tone or format you are after and ask it to match. Models are excellent at imitation, so a single good example steers them better than a paragraph of description.
Treat it as a conversation
The best results rarely come from the first prompt. Read the reply, then refine. Tell it what to change, ask it to go deeper on one part, or push back when it is wrong. Each follow up gets you closer, and the model remembers the thread as you go.
That is close, but make it shorter and less formal, and cut the last sentence. Then give me two alternative opening lines.
A prompt template
A quick template that covers most of this at once. Role, then task, then context, then format. "You are a [role]. Help me [task]. Here is the situation, [context]. Give it to me as [format]." Fill in the blanks and you are most of the way to a great prompt.
+ Pros
- + Far better output from every AI tool
- + Less time spent fixing generic answers
- + Works across ChatGPT, Claude, Gemini, and the rest
- + Costs nothing to learn
– Cons
- – Takes a little more effort up front
- – No single prompt works for every task
- – Still needs you to check the result
FAQ
Do I need to learn prompt engineering?+
Not formally. The habits in this article, being specific, adding context, assigning a role, giving examples, and iterating, cover almost everything most people need.
Why does the AI give generic answers?+
Almost always because the prompt was generic. Add detail about the format, audience, and your situation, and the answers get sharper immediately.
Does the same prompt work on every AI tool?+
Mostly, yes. ChatGPT, Claude, and Gemini all respond well to clear, specific, context rich prompts. Minor tweaks help, but the core habits transfer.
Should I write long prompts?+
Long enough to include what matters, no longer. Context and specifics help. Padding does not. Aim for clear, not lengthy.
These habits pay off everywhere. See them applied in our guides to using ChatGPT for programming help and ChatGPT for data analysis.
