Overview
When you use AI tools like Claude for Education, PittGPT, Microsoft 365 CoPilot Chat, and Gemini Chat, the quality of your prompt has a huge impact on the quality of the response you get. A vague prompt often leads to generic or incomplete results, but a clear, detailed prompt helps the AI understand exactly what you want — and deliver it in a useful way. The CREATE framework, developed by AI consultant and author Dave Birss, is a simple way to remember how to write better prompts. CREATE stands for: Character, Request, Examples, Adjustments, Type of Output, and Extras. Each part of this framework helps you give the AI clearer instructions so you can unlock more accurate, relevant, and valuable answers.
Detail
Below is a breakdown of each part of the CREATE framework, with explanations and examples you can adapt for your own prompts.
Character:
Start by telling the AI what role you want it to take on. Think of the AI as an actor — the more specific you are about the “character” it should play, the better it can draw on the most relevant information. This is sometimes known as Role Prompting. For example, instead of asking PittGPT for generic writing help, you might say: “You are an experienced copywriter with 20 years of experience creating persuasive sales emails for the automotive industry.” This sets the tone and helps the AI focus on the right style and knowledge area.
These character traits can also be made permanent by adding a system prompt. This is accessed through the various tools to give the AI direction on how you want to interact with it.
Request:
Next, be clear and specific about what you want the AI to produce. Include enough context and any important details. A general request like “Write me an email” will likely get you an average, uninspired result. A more effective request might be: “Write a persuasive, energetic sales email for the Cougar Hyper Sport, an electric sports car known for its best-in-class acceleration. The email should build excitement, emphasize the thrill of driving it, and highlight what it feels like to turn heads when you’re behind the wheel.” Providing clear context and key points makes all the difference.
Examples (Optional):
Whenever possible, give examples that help illustrate the tone, style, or format you want. This step is optional, but it often helps the AI match your expectations more closely. For instance, you could say: “Use an energetic tone similar to this sample line: ‘Feel the power beneath your feet.’ Keep the language bold, exciting, and emotionally engaging.” Examples are especially useful if you want a specific voice, mood, or structure.
Adjustments:
Once you get an initial draft from the AI, you may find that you want to tweak how it presents information. You can include these refinements in your original prompt or add them when re-prompting. For example, you might add: “Don’t use bullet points — write in full paragraphs. Break up the text with subheadings and keep sentences short and punchy.” Adjustments help you fine-tune the response so it aligns with your expectations. It is also often useful to give the AI an instruction to clearly respond that it does not know the answer, rather than attempting to answer a question which is ambiguous or for which there is not enough information.
Type of Output:
It’s also important to tell the AI exactly what format or length you want for the final result. This helps ensure you get something you can use right away, with less back-and-forth. For example, you could say: “Deliver your response as a 500-word article with a catchy headline, subheadings for easy reading, and a clear conclusion with a call to action.” The more specific you are, the more useful the output will be.
Extras:
Finally, there are a few additional instructions you can include to make your prompt even more effective. These “extras” can help you get better results in tricky or complex situations. Some helpful examples are:
-
“Ignore everything before this prompt.” This clears out any previous conversation or context, so the AI starts fresh.
-
“Ask me questions before you answer.” This encourages the AI to clarify missing details before giving you a final response, which is especially useful for complicated tasks.
-
“Explain your thinking.” This prompts the AI to show its reasoning or step-by-step process, which can help you verify that the information is accurate or logical.
Putting all of this together, a well-crafted prompt using the CREATE framework can take a little more effort upfront — but it almost always saves you time in the long run. You’ll get more targeted, relevant, and polished results, with fewer revisions needed.
Tips:
- If you don’t get exactly what you want on the first try, don’t be discouraged. Tweak one or more parts of your prompt — add more detail, clarify the role, or include an example — and run it again. Prompt writing is a skill you’ll get better at with practice.
- Sometimes it is useful to list what not to do.
- The context window for these tools is limited. During an extended conversation it is often useful to start a new chat to start with a short context window. This allows the AI to focus on the current prompt. As an additional tip at the end of your long chat, you can ask the AI to summarize it in a very condensed way and feed that into the next chat.