This 3-part prompt structure fixes 80% of bad AI outputs (live example)

​

“The Framework That Fixes Messy Prompts”

Hi Reader,

​

When someone says “ChatGPT didn’t give me what I wanted,” it’s typically not the model that failed.

It’s the prompt, and more specifically it’s the lack of structure in the prompt. Building the skill to design strong prompts is becoming an essential differentiator across several professions and fields, and soon could be the difference between winning project, landing a job, and remaining personally competitive.

The good news is that among the several frameworks out there, there are simple, easy to use frameworks that you can begin using instantly to nurture and build prompt design skills.

Let's look at one of these frameworks, the RTF Framework:

RTF = Request – Task – Format

It’s one of the simplest and most powerful ways to get better results from AI, and fast.

Here’s how it works:

🟢 R = Request

👉 What do you want the AI to do?

Use a clear verb and object.

Example: “Help me write a summary…”​
​

🟢 T = Task

👉 What’s the purpose, audience, or context?

Give it critical details.

Example: “…for a presentation to mid-level managers about our Q3 strategy.”

​

🟢 F = Format

👉 How should it be delivered?

Define structure, length, tone, or style.

Example: “Make it 3 bullet points, no more than 100 words total.”

Without RTF:

With RTF:

🎯 Better structure = better outputs. Every time.

🔍 In fact, I’d wager that most “prompt hacks” on the internet are based on RTF with extra glitter.

This one works because it’s foundational. You can use it with any model, on any platform, for nearly any task, and its a great entry point to building prompt design into your process when working with AI.

​

I recently published a brief video on the RTF Framework in action here. The prompt used in this video is pasted below.

Next time you’re stuck staring at a blinking cursor in ChatGPT… try RTF.

​

—Adam

Founder, The Viable Edge

​
​

🧠 PS: Want me to break down more frameworks like this in future emails? Just reply “More RTF” and I’ll know you’re in.

RTF PROMPT:

Request:

Create a daily plant watering reminder table that tracks the following plants: tomato plant, basil plant, and hydrangeas.

Task:

• Set up a table to show 15 days of plant watering activity and weather data.
  ​
  ​
• Include the past 7 days of actual weather conditions and watering history, as well as the next 7 days of weather forecasts, and anticipated actions that might be needed based on the current forecast. Source real world data from the internet.
  ​
  ​
• Update the table every day at 6am and send that as a task reminder, ensuring all entries are accurate and visually engaging.
  ​
  ​
• Provide daily watering instructions, such as “Water today” or “No need to water today,” based on the best plant species care practices, and very specifically accounting for previous watering actions and the current forecast.
  ​
  ​
• If any historical watering data is missing, ask for that information before proceeding.
  ​
  ​
• Include weather emojis and plant icons to make the table easy to understand at a glance.
  ​
  ​
• Ask any clarifying questions as needed, such as geographic location, time, and/or previous watering history to ensure this task kicks-off diligently up-to-date.
  ​
  ​
• Reproduce the full, updated table each day with the reminder.
  ​
  ​
• On the initial launch of this task reminder
  • ask for the user’s location to ensure accurate weather data
  • Any previous watering activity so you know the last time watering took place and can account for that in the ongoing recommendations.
  • The current date

Format:

• A well-organized table with columns for date, weather conditions, and the watering status of each plant.
  ​
  ​
• Use emojis to indicate actions taken (e.g., ✅ for watered, ❌ for not watered) and weather conditions (e.g., ☀️ for sunny, 🌧️ for rainy).
  ​
  ​
• Place today’s watering instructions at the bottom of the table.
  ​
  ​
• Ensure the table is visually clear and easy to read.

​