Most people's first experience with an AI assistant goes something like this: type a short request, get a generic answer, conclude that AI is not very useful. The difference between that and a genuinely helpful result is usually not the tool. It is how the request is written.

The good news is that writing good prompts is not a technical skill. It is the same skill as writing a clear brief for a new colleague. This guide covers the techniques that make the biggest difference.

Think of it as briefing a capable new colleague

Imagine handing a task to someone smart who has just joined your company. They know a lot in general, but nothing about your customers, your products, or your standards. What would they need from you to do a good job? That is exactly what a good prompt contains.

The five ingredients of a good prompt

1. The task

Say clearly what you want produced. "Write a reply" is vague. "Write a short, friendly reply to this customer explaining that their order will arrive two days late" is clear.

2. The context

Give the background the AI cannot know: who the audience is, what has happened so far, and why the task matters. Paste in the relevant email, notes, or document rather than describing them.

3. The constraints

State the limits: length, tone, what to include, and what to avoid. For example: "Under 120 words. Apologize once, do not offer a discount, and end with a clear next step."

4. An example

If you have an example of what "good" looks like, include it. One strong example often does more than a paragraph of instructions, especially for tone and format.

5. The format

Say how you want the output structured: a bulleted list, a table, three options to choose from, or a draft ready to paste into an email.

A before-and-after example

Before:

Write a LinkedIn post about our new service.

After:

Write a LinkedIn post announcing our new AI support assistant service for small online stores. Audience: owners of small e-commerce businesses. Tone: practical and warm, no hype. Mention that it answers common questions like order status and returns, and that a person always handles refunds. Under 150 words, no hashtags except one at the end. Give me two versions to choose from.

The second prompt takes thirty seconds longer to write and usually saves several rounds of back-and-forth.

Techniques that help

  • Ask for options. "Give me three different subject lines" makes it easier to pick a good one than asking for a single perfect answer.
  • Iterate in the conversation. Treat the first answer as a draft. Reply with specific feedback: "Shorter, and make the second paragraph more concrete."
  • Ask it to ask you. For complex tasks, start with: "Before you start, ask me any questions you need answered to do this well."
  • Request reasoning for decisions. "Recommend one option and explain why in two sentences" helps you judge the suggestion rather than just accepting it.
  • Break big tasks into steps. Outline first, then draft each section, rather than asking for a finished 2,000-word document in one go.

Things to avoid

  • Pasting confidential information into unapproved tools. Check your company's policy on which AI tools may be used with which data.
  • Trusting facts without checking. AI can state wrong numbers, dates, or references confidently. Verify anything that matters.
  • Vague quality words. "Make it better" or "more professional" give little direction. Say what specifically should change.
  • Starting from scratch every time. If a prompt works well, save it.

Build a shared prompt library

The fastest way for a team to improve is to share what works. Create a simple shared document with prompts for your most common tasks, such as customer replies, meeting summaries, or proposal outlines, each with a note on when to use it. New team members get good results from day one, and quality becomes more consistent across the team.

A template to start with

Copy this, fill in the brackets, and adjust as you go:

Task: [what you want produced]
Context: [who it is for, background, relevant material pasted below]
Constraints: [length, tone, must include, must avoid]
Example: [an example of good output, if you have one]
Format: [how the result should be structured]

Clear prompts are the simplest, cheapest improvement most teams can make to their use of AI. If you are rolling out AI tools across your team and want help with guidelines, prompt libraries, or training, we are happy to help.