If you’ve ever felt like AI gives you “meh” answers—or worse, confidently wrong ones—you’re not alone. Most of the time, the problem isn’t the AI. It’s the prompt.

Using a generative AI tool like ChatGPT without a prompt framework is like saying to a friend: “Please make me food.” You have no idea what you're going to get.

Using a prompt framework is like saying to a friend instead: “Please make me a perfectly-cooked large cheese pizza with pepperoni at 5pm.” You know exactly what you're getting.

Same kitchen. Way better results.

ChatGPT (any large language model, or LLM) is not "thinking," it is reviewing millions of words to predict what word should come next in your sentence. In the example above, it has no guidance about the output you need. The second prompt provides needed context. To be effective, LLMs need a certain amount of context in order to give you the results you're looking for. This particularly holds true when using industry jargon and acronyms, which may mean different things in different settings.

Great prompts aren’t wordy or overly clever. They’re structured.

Over the past few years, working as a Chief Marketing & Membership Officer and building multiple custom GPTs, I’ve tested dozens of prompt patterns. Three frameworks consistently rise above the rest—APE, CARE, and RTF. They help me communicate clearly with AI and dramatically improve the quality, accuracy, and usefulness of what I get back.

Here’s a quick guide to using them yourself.

Action → Purpose → Expectation

APE is my go-to when I want the AI to do something specific—summarize, rewrite, analyze, categorize, or create something with guardrails.

Tell the AI exactly what you want it to do. Example: “Summarize this 50-page financial policies and procedures manual.”

Explain why you need the output. This gives the AI context so it "knows" what to emphasize. Example: “This is for a grant application to the XYZ Foundation.”

Tell the AI what good looks like. Outcome clarity always improves accuracy. Example: “Highlight the provisions that impact the safety and security of potential grant funding.”

It mirrors how you would brief a colleague. Action prevents vagueness, purpose gives direction, and expectations define success.

Great for strategy, planning, and creative work.

CARE ensures your prompt includes what the AI must understand before it produces anything.

Where are we? What’s happening? Example: “You're launching a new online learning course.”

What should the AI produce? “Draft a marketing plan with messaging, channels, and tactics.”

What does success look like? “Incorporate the WIIFM and use behavioral economics techniques for copywriting.”

Give it an existing file, link, or model to follow. “Make it similar to this (uploaded) marketing plan.”

AI is pattern-based. When you feed it a pattern through context and examples, the output is more tailored, more polished, and more useful.

Role → Task → Format

RTF is perfect when tone, perspective, or channel really matters. I use this one constantly inside my custom CMO GPT.

Tell the AI who it is supposed to be. “You're a chief marketing officer at a professional association.”

Tell it what to do and what inputs to use. “Review the uploaded materials on brand voice and messaging for our webinar program and write four social posts emphasizing value and WIIFM. Use behavioral economics techniques like loss aversion.”

Tell it how the final output should be delivered. “The posts will go on LinkedIn.”

Defining the role changes the lens the AI uses—like switching hats. You get content that sounds like the expert you need, not generic output.

How to Start Using These in Your Daily Work

To get the most out of APE, CARE, and RTF:

AI performs best when it can see what you're talking about.

Channel, audience, tone, length—these all matter.

For example:

• Use RTF for tone/role
• Use APE for task clarity
• Use CARE when you need a strategic plan

This is exactly how I structure my Chief Marketing Officer GPT—repeatable patterns that raise the floor and ceiling on output quality.

Like this post? Follow me on LinkedIn!