Most people use AI assistants like a slightly smarter search engine. They type a question, get an answer, and move on. And most of the time, the answer is... fine. Usable. Forgettable.
The people getting dramatically better results from the same tools aren't using different AI — they're asking differently. Here's the framework that makes the biggest difference.
The RCTF Formula
Every high-quality AI prompt has four components. You don't need all four for every request, but the more of them you include, the better the output:
- R — Role: Tell the AI who to be
- C — Context: Give it the background it needs
- T — Task: State exactly what you want
- F — Format: Specify how you want the output
What This Looks Like in Practice
Before (vague): "Write me an email to my boss about taking a day off."
After (RCTF): "You are a professional business writer. I work in a mid-sized marketing agency and I have a good relationship with my manager. Write me a brief, professional email requesting one day off next Friday to attend a family event. Keep it to 3 sentences, casual but professional in tone, and end with a line offering to arrange coverage if needed."
The second version produces something you can send directly. The first produces a generic template you have to rewrite anyway.
Three More Techniques That Work
1. "Give me your honest assessment." AI models are trained to be agreeable. Explicitly asking for an honest, critical take — "Tell me what's wrong with this plan" or "What am I missing?" — consistently unlocks more useful analysis.
2. Ask for options, not one answer. "Give me three different approaches to this, with the tradeoffs of each" consistently outperforms "tell me the best way to do this." You'll see considerations you wouldn't have thought of.
3. Iterate, don't start over. Your first prompt is a starting point, not a final request. Follow up: "Make it shorter." "Make the tone more confident." "Add a specific example." Each refinement costs you nothing and takes seconds.
The Single Most Underused Feature
Most people never use the system prompt or custom instructions. In ChatGPT, you can set persistent instructions that apply to every conversation: your profession, your preferred writing style, what you don't want the AI to do. Setting this up once saves you from repeating context every session.
In Claude, you can paste a full document at the start of a conversation and refer back to it. In Gemini, you can save a Gem (a custom configuration) for specific recurring tasks.
Students applying this structure to coursework and revision quizzes can also pair these prompts with our guide to AI study tools for notes, research, and active recall.
The tools are powerful. How you talk to them determines how much of that power you actually get. Try the RCTF formula on your next prompt and see the difference for yourself.