Most people get bad AI output because they give AI a bad brief. This framework fixes that — permanently. Master these five ingredients and you'll get dramatically better results from any AI tool, every single time.
Think of AI like a brilliant new hire on day one. Fast, capable, eager — but knows nothing about your business until you brief them. Your prompt is the brief. The better the brief, the better the output.
The Role is the persona you assign before anything else. It's the single highest-leverage ingredient — changing the Role alone can completely transform the quality, tone, expertise level, and framing of the output.
Context is the briefing. AI knows nothing about your business, your customer, what happened yesterday, or what's at stake — until you tell it. The more specific your context, the more specific the output.
The Task is the specific action you want AI to perform. Most people make their Task too vague. They ask for a "summary" when they mean a "3-bullet executive brief" or "help with the email" when they need something far more specific.
Format controls the structure, length, tone, and style of what AI produces. Without it, AI defaults to its own preferences — often too long, too formal, or structured in a way that doesn't match your actual use case.
Constraints are the guardrails. They protect your brand voice, prevent AI from using language you hate, and stop it from making assumptions that undermine the output. Most beginners skip this step. Experts never do.
Every prompt below is fully built using R·C·T·F·C. Click the copy button, fill in any brackets with your real details, and run it in ChatGPT or Claude.
Three real tasks — built without the framework first, then rebuilt with R·C·T·F·C. The improved version has a copy button so you can use it immediately.
Every one is fixable with one extra line in your prompt. Recognising them is half the battle.
Screenshot this. Print this. Pin it next to your screen. The entire framework in the smallest possible format.
The R·C·T·F·C Framework is what we teach in our Role-Based Prompt Engineering Workshops — built around your team's actual roles, workflows, and use cases. Your people walk out with 25–50 prompts they'll use the same day.