How you phrase your requests to AI tools makes a real difference to the quality of results you get. This guide covers the techniques that help most — with practical examples and templates.
AI tools can generate confident-sounding text that is factually wrong — this is called "hallucination." Always verify important facts, figures, and claims before acting on them. This applies to every prompt technique in this guide. AI is a powerful drafting tool, not a reliable source of truth.
A prompt is just the text you type into an AI tool. It's your instruction, your question, your request.
The problem is that most people use AI the way they'd use a search engine: short, vague queries and hoping for the best. That works fine for Google, but AI tools respond very differently depending on how you ask.
The same AI can give you a generic, unhelpful answer or a useful one — the difference is often just the prompt.
The good news: you don't need to learn anything technical. You just need to give the AI a bit more to work with.
This guide covers the techniques that make the biggest difference for business use cases — writing emails, creating content, analysing information, drafting documents. No jargon, just practical examples you can use immediately.
The most common prompting mistake is being too vague. AI doesn't read between the lines — it responds to what you actually write.
Vague prompts get generic answers. Specific prompts get useful ones.
The second prompt tells the AI:
That's not complicated — it's just giving the AI the same context you'd give a human colleague.
Before you send a prompt, ask yourself: "If I gave this instruction to a new employee with no context, would they know what to do?" If not, add more detail.
One of the most powerful ways to get consistent results is to show the AI what you want, not just tell it.
This technique is called few-shot prompting — you give one or more examples of the output you're looking for, and the AI follows the pattern.
By showing the AI your preferred style — the length, the tone, the structure — it can replicate that pattern more reliably than if you tried to describe it in words.
When you show the AI examples, you're communicating more than words alone can express — format, length, tone, structure, level of detail. It's often faster to paste in a good example than to write a detailed description of what you want.
Telling the AI to act as a specific type of expert changes how it responds. It draws on different knowledge, uses different language, and prioritises different things.
This isn't magic — it's just a way of focusing the AI's response towards a particular perspective or expertise.
Role-playing makes AI more useful, but doesn't make it an actual expert. Use this for drafts and ideas — not as a substitute for real professional advice.
Roles work even better when combined with specific context about your situation:
For complex questions or analysis, asking the AI to "think step by step" or "walk through your reasoning" often produces better, more accurate results.
This technique is called chain-of-thought prompting. It works because it forces the AI to break down the problem rather than jumping straight to an answer.
You don't need fancy language. Any of these work:
Sometimes the best way to get what you want is to say what you don't want. AI has certain default tendencies — it can be overly formal, use filler phrases, or pad out responses with unnecessary content.
Telling it what to avoid helps you get cleaner, more useful output.
"Don't use corporate jargon or buzzwords. No phrases like 'leverage', 'synergy', 'circle back', or 'touch base'."
"Don't include an introduction or summary — just give me the content directly."
"Don't give generic advice. Be specific to my situation."
"Don't soften your feedback or be overly polite. Tell me directly what's wrong."
AI can output information in many different formats. Specifying what you need saves you reformatting later and often makes the content more useful.
The first response isn't always good — that's normal. Knowing how to iterate and improve is just as important as writing a good initial prompt.
Say: "Shorten this by half" or "Give me a version that's 3 sentences max" or "Remove all unnecessary words"
Say: "Be more specific" or "Give concrete examples" or "Apply this specifically to [my situation]"
Say: "Make this more casual" or "This needs to sound more authoritative" or "Rewrite this as if you're talking to a friend"
Say: "The main thing I need is X — focus on that" or "You've addressed Y, but I asked about X"
What to do: If something looks wrong or sounds too precise to be true, check it yourself against a reliable source. Asking the AI "are you sure?" may not help — it may simply restate the same incorrect information with more confidence. Try: "Don't include anything you're not certain about" to reduce the risk in future outputs.
Think of AI like a first draft from a capable but literal-minded assistant. You wouldn't expect a first draft to be perfect — you'd give feedback and refine it.
Getting good results from AI is often a conversation, not a single prompt. It's normal to need 2-3 rounds of refinement. Don't give up after a disappointing first response — tell the AI what's wrong and ask it to try again.
These techniques cover the fundamentals — being specific, giving examples, assigning roles, thinking step by step, setting constraints, and specifying format. Apply them consistently and the quality of what you get from AI tools should improve.
If you'd like guidance on how to use AI effectively in your specific business, let's have a conversation.