Chatbots continue to evolve, rapidly. Claude just added a conversational voice mode for mobile. Gemini now has a “Deep Think” feature and is powering a new AI Mode for Google Search. Meta AI can finally be accessed separately. Grok adds a memory. ChatGPT-5 is on the horizon. 

One thing has not changed and that’s the key to creating a better prompt. The “Golden Rule” for AI prompts: The more you give, the better results you get. 

Yes, the key to a great prompt is in the details: you must supply the details. General prompts are about as useless as asking a barista for something “caffeinated.” 

That brings us to AI’s Three Cardinal Rules that will help you perfect your prompting. 

1. Details 

Give your Gen AI model more to work with: quality in = quality out! 

If your prompt is vague, your output will be too. Chatbots like ChatGPT thrive on context. The more information you provide, the more accurate, relevant, and useful the response will be. 

If you’re working on an email, share examples of your past best work, provide details about the audience and insight to the tone you are striving to achieve. If it’s a blog post, upload a few examples of best past work or other posts that are excellent examples of what you are trying to achieve. If you have related research, provide the details and the links or upload the full PDF for review. 

Think of working with a Chatbot as a teachable moment. When you are using ChatGPT Teams, for example, it will remember these things – like your tone, style, writing rules, and the output format – for future posts. 

2. Specificity 

Say exactly what you want from the first input. 

It’s not enough to give the model background and data. You need to guide it to what you want pulled from that input. Don’t just say “summarize this.” Say, “summarize this for a real estate executive about to make a presentation to a room full of San Francisco real estate agents. Provide an executive summary followed by the most important takeaways in bullet points and finally why this information is relevant to this audience.” 

The more specific your request, the more focused your output and valuable your results. You’re narrowing things down to allow the AI model to better deliver what you had in mind and get closer to what you need. 

3. Iteration 

Never, ever stop at the first draft. 

AI isn’t a mind reader and it’s not a magician. It’s more like a lightning-fast intern who needs direction and feedback. The first output is often close, if you follow the first two rules, but “just okay” isn’t what you’re after. 

Look for the weaknesses in the reply. Zero in where it is close but not there yet. Push it to do better and explain in specifics why it missed the point or where you want to redirect it. Then do it all again because the best results happen on round two or three, and sometimes five. 

Final takeaway 

Great prompts aren’t created by luck but practice and personal experience. 

 The more effort you put into what you feed your AI chatbot, the less time you’ll spend on iteration or rewriting more of the output. 

The best approach: Treat prompting like a collaboration, not a one-and-done transaction. 

Your prompt is the steering wheel. If you don’t drive with intention, don’t be surprised where you end up. (-Kevin)