Get better results from AI

1. Share examples

Show AI a result you like and explain why.

This could be a piece of writing, a website screenshot or a finished document. For example: “I like the short sentences and real-life examples in this text.”

This makes it easier for AI to understand what “good” means to you.

Ready-to-use prompt

I need [task].

Here is an example I like: [attach text, an image or a file].

Use its [tone / structure / way of explaining things]. Pay particular attention to [specific detail].

Create the result for [your audience]. Do not copy the example word for word.

2. Save it in files

Write down your background and work preferences so you do not have to explain them every time.

Work in a folder on your computer using Codex or Claude. Ask AI to create two text files in that folder:

• “About me”: what you do, who you work with and your goals.
• “My preferences”: the results you want and what to keep in mind.

For example: “Use a friendly tone. Keep the term AI. Write briefly and include specific examples.” Update the files as you work.

Ready-to-use prompt

Review my explanations and feedback in this conversation.

Suggest what to save in “About me” and “My preferences” for future tasks.

Do not save assumptions as facts. Separate ongoing preferences from requirements that only apply to this task.

Show me the additions, then save them in the project folder after I approve them. Create the files if they do not exist yet.

At the start of a new task, ask: “Read the ‘About me’ and ‘My preferences’ files in this folder and use them for this task.”

3. Give specific feedback

Explain exactly what to change and what already works.

Instead of “rewrite this,” say: “Shorten the introduction to two sentences. Add an example from a freelancer’s work. Keep the tone and structure.”

You do not have to fix everything at once. Start with the structure, then the examples, then individual sentences.

Ready-to-use prompt

Improve the current version.

Keep: [what you like].

Change: [what needs fixing and how].

Here is an example of the result I want: [example].

Leave the rest unchanged.

4. Describe the task

Tell AI what to make, who it is for and what format you need.

For example, expand “write a post about AI” into: “Write a short post for freelancers about using references. Include one example and invite the reader to try the tip.”

Clear requirements make the result easier to review.

Ready-to-use prompt

Create [specific result].

Audience: [who it is for].
Goal: [what the reader should understand or do].
Format: [length and structure].
Language and tone: [requirements].

Make sure to include [essential elements].

The result will be successful if [specific criteria: for example, it includes a clear example and takes one minute to read].

5. Ask for questions

Ask AI to clarify important details before it starts.

For example, a course page needs to explain who the course is for, what it teaches and how to sign up. If you left these details out, add them before AI creates the page.

Ready-to-use prompt

I need [task].

Here is what is already known: [information].

Before you start, ask up to five questions about the information you need to complete the task.

Only ask about details that affect the result. Suggest answer options where helpful.

Start working after I answer.

6. Work in stages

Split large tasks into smaller parts and review each one separately.

For example, when making a website, agree on the structure first, then the copy, then the design. This makes it easier to spot a problem before it affects the whole project.

Ready-to-use prompt

Help me complete [large task].

Split the work into 3–5 stages. Briefly explain what result I will get at each stage.

Show me the plan for approval. Then complete the first stage and wait for my feedback.

Use the decisions we have already agreed on in the following stages.