Prompting Framework For Top 10% AI Users

T - Task. What do you actually want AI to do? Be specific here. The more precise you are the better the output. Give it a persona, something like "act as an experienced financial analyst" and tell it the format you want the output in. List, paragraph, table, whatever fits your use case. C — Context. The more context you give, the better your results. Background information, who the audience is, what constraints you're working with, what you've already tried — all of it helps. Think of it like briefing someone new on a project. The more they understand your situation, the better work they can do for you. R — References. You can describe what you want with words all day, but an actual example captures nuances that words just can't. Paste in a sample of writing you like, a format that worked, a previous output that was close. References do a lot of heavy lifting. E — Evaluate. After you get the output, actually stop and ask yourself — is this what I wanted? Does it hit the brief? Is something off? Most people just take the first output and run with it. This step is what separates intentional prompting from just hoping for the best. I — Iterate. Prompting is rarely a one-and-done thing. It's a loop. You evaluate, you adjust, you go again. Always be iterating.