The 5-Part Prompt Formula for Client-Ready Answers

Here's the truth about "AI gives generic answers": generic prompts get generic answers. "Write me an insurance post" gives you rubbish. But add five ingredients and the same tool writes like you on your best day. This formula is the foundation for every lesson in this course — master it here, use it forever.

The formula: R.S.T.F.T

1. Role — who should the AI be? "You are an experienced Malaysian insurance agent who writes warm, natural WhatsApp messages."
2. Situation — the context it needs. "My client is 40, has a medical card, just asked why his premium went up."
3. Task — exactly what you want. "Write a reply that explains repricing honestly and calms his worry."
4. Format — shape of the answer. "Max 6 lines, short sentences, no bullet points, end with a question."
5. Tone — how it should sound. "Casual Malaysian English, warm, zero pressure, no exclamation marks."

See the difference

Weak prompt: "Write a message about medical card premium increase." → You get a stiff, generic essay that sounds like a press release.

Formula prompt:

ROLE: You are an experienced Malaysian insurance agent who writes warm, honest WhatsApp messages.

SITUATION: My client Mei Ling, 38, just messaged me: "Why my medical card premium increase again?? Every year go up!" Medical inflation in Malaysia is around 12-15% per year and repricing affects the whole industry, not just her plan.

TASK: Write a reply that (1) acknowledges her frustration first, (2) explains repricing simply using the medical inflation angle, (3) offers to review her plan together to check options — without pushing anything.

FORMAT: Max 7 lines, short sentences, no bullet points, end with an easy yes/no question.

TONE: Casual Malaysian English, empathetic, honest, zero sales pressure.

Run both versions yourself and compare. The second one produces a message you could genuinely send after a 10-second edit. Same AI, same day — the only difference is the prompt.

The secret sixth ingredient: iterate

Your first output is a draft, not a verdict. Reply to the AI like you'd brief a junior assistant:

• "Shorter."
• "Too formal — make it sound like texting a friend."
• "Give me 3 different versions."
• "Add a bit of humour at the start."

Each refinement takes 5 seconds. Two or three rounds usually gets you to "send-ready". Agents who quit after one bad answer never discover this.