16/04/2026
I’ve built 80+ AI agents and spent 10,000 hours prompting. These 4 rules never fail.
Here’s what works for me
The 4 Golden Rules of Prompting
(works for text, code, images, video everything)
▢ Specific Goal ↦ What do you actually want?
▢ Context ↦ What does it need to know to help?
▢ Role ↦ Who should it pretend to be?
▢ Examples ↦ What should the output look like?
Use these 3 frameworks:
Saves you time and gets consistently sharp outputs - everytime
1./ R.A.C.E.
🔹Role → Tell AI who to be
🔹Action → What you need it to do
🔹Context → Key details or data
🔹Expectation → Say exactly how you want the output
E.g. "Act as an experienced BA. Analyse this CSV of support tickets. Focus on onboarding issues. Give me 5 top pain points with quotes and prioritisation."
2./ R.I.S.E.
🔹 Request – Set the expert persona
🔹 Input – Give it the raw data
🔹 Scenario – Describe the situation
🔹 Expectation – Ask for actionable outcomes
E.g. "Draft a user story. Input: Acceptance Criteria. Scenario: Under 8 points. Expect: Story + edge cases."
3./ C.O.A.T.
🔹 Context – Background info
🔹 Objective – Your end goal
🔹 Actions – How AI should think
🔹 Tasks – What you want back
E.g. "We need a GTM plan to get 1000 paid users in 3 months. Pick channels, budget, and create a week-by-week plan."
Prompting is the highest ROI skill you can learn.
It’s the new literacy.
Don’t overthink it.
Just test, tweak, and talk to it like a smart intern.
Even asking AI to make your prompt better is a prompt.
Try this:
"You're a world-class prompt engineer. Optimise this prompt for clarity and results: [Paste Prompt]"
You’ll be shocked at the improvement.
📌 There is something better than prompting and its creating AI agents.