Read time: 5 minutes | Subscribe to this newsletter

Hey warrior,

I’m thinking of hosting a workshop about how to build a one-person business to $40k+ revenue by leveraging AI.

If you’re interested, just click here and I’ll send you some more details.

This week’s newsletter is sponsored by Attention.​

We’re all too familiar with the headaches of monotonous work tasks. Updating your CRM. Writing emails. Watching call recordings. Yawn!

Attention automates it all for you. This AI wizard can:

• Automatically log in crucial info into your CRM.

• Write your follow up emails while adopting your writing style.

• Guide your sales reps with in and post-call coaching.

Plus, they feed insights that have pulled in an extra $250k in ARR per rep.

Start automating your sales today!

And a big thank you to another newsletter sponsor for today: Softr.io

→ Create powerful apps with just a prompt.

Softr AI App Generator breaks the rules by letting you build full-stack apps effortlessly. From databases to user authentication, it's all there!

No more settling for simple static websites - go big with Softr AI 💻🧠

We’ve all been there.

You open up ChatGPT and have an idea about what you want to ask it.

But you’re struggling with putting your thoughts into words and making sure that ChatGPT actually understands what you want it to do.

This is where prompt frameworks come in.

When used right, they help you maintain a structure in your prompts and deliver the right information to your AI chatbot.

Today I’ll be talking about my 5 favorite prompt frameworks.

When you use them correctly, you’re going to improve your AI outputs massively.

Here’s what you’ll learn today:

• The RTF framework

• Chain of thought

• RISEN

• RODES

• Chain of density

The RTF framework

R - Role
T - Task
F - Format

The "R-T-F" prompt framework categorizes requests into Roles, Tasks, and Formats for clearer communication.

This is the jack-of-all-trades prompt and can be used for most use cases, even non-work related ones.

Prompt Template:

"Act like a [insert the role you want AI to take]. Give me a [insert task] in [insert format] format."

Example:

"Act like a life coach with 30 years of experience in mentoring. Give me a plan to improve my work-life balance in table format."

And here’s the output:

As you can see, ChatGPT gives a pretty comprehensive breakdown of ways to improve my work-life balance.

And like I said above, this framework has lots of use cases.

I’ve also got great results by telling AI to act like a chef or a personal trainer.

But the possibilities are pretty endless.

Chain of thought

The chain of thought prompt framework is one that improves LLM’s reasoning by telling AI to go through a problem step-by-step.

This makes it especially good for problem-solving or complex analytical tasks.

All you have to do is add the simple phrase “Let’s think step-by-step” at the end of your prompt.

Prompt Template:

"[insert your prompt instructions].

Let's think through it step-by-step."

Example:

"How do I improve my sales calls? I've only got a 15% close rate right now, and I think it's because I'm not selling the dream enough.

Let's think through it step-by-step."

And here is the output from ChatGPT:

As requested, the AI goes through sales calls step-by-step giving great tips on how to improve your sales call.

This prompt can be especially useful when you have some math problem to solve and ChatGPT keeps getting it wrong.

Try adding “Let’s think step-by-step” at the end and watch how the results improve.

RISEN

R - Role
I - Instructions
S - Steps
E - End goal
N - Narrowing

The RISEN framework provides a structured approach to dissect complex or constrained tasks, such as blog posts or research projects, into actionable components for better execution and focus.

Prompt Template:

"Role: [insert the role you want AI to take.] 

Main Task: [Insert the task you want AI to complete.]

Steps to complete task: [Insert numbered list of steps to follow.]

Goal: [Insert goal of the output]

Constraints: [Enter constraints]."

Example:

"Role: You are an expert digital course builder who has sold millions in online courses.

Main Task: Please give me a list of EVERYTHING important that I should include in my AI course and tell me all the different methods of growth I can implement to maximise revenue.

Steps to complete the task:

1. First start by covering all the things that ANY digital course should include. 

2. Then proceed by giving your thoughts on what AI courses should include. 

3. End with covering the best growth marketing tactics and strategies for digital courses.

Goal: The goal is to give me a concise list of everything I should include within the course, as well as give me ideas on how I can maximize the revenue from my course. 

Constraints: Maximum of 500 words. - Avoid technical jargon. - Make it actionable - Make it clear"

Here’s the output…

ChatGPT gives us exactly what we ask for. It’s obviously a much longer prompt but that means it can be tailored to your exact needs, thus improving your output.

RODES

R - Role
O - Objective
D - Details
E - Examples
S - Sense Check

While this is similar to the above prompt, this one is better used whenever you have good examples of your desired output.

Prompt template:

R - Role: [Insert desired role you want AI to take.]

O - Objective: [Insert objective you want AI to do.]

D - Details: [Insert any context or constraints AI needs to create a good output]
  
E - Examples: Here are good examples you can use to model your answer.

[Insert examples of good outputs]

S - Sense Check: Do you understand the objective and the specific guidelines for this task?

Example:

R - Role: You are a seasoned copywriter specialized in crafting viral tweets.

O - Objective: Write a twitter hook that aims to go viral on the topic of Marketing. 

D - Details: - The tweet should be no longer than 280 characters.
  - Use powerful and persuasive language.
  - Do not include hashtags or emojis.

E - Examples: Here are some examples you should use to model your answer (note - these are not on my desired topic. I just want you to understand the frameworks and styles that work")

1."Credit card debt is killing your financial progress.

Here's how you get rid of it ASAP:"

2. "The most valuable Twitter feature you aren't using: 

Advanced search.

Knowing how to use it will help you find the hidden gems of the Twitter archives and 10x your Twitter experience.

Here's the step-by-step guide:"

3. "99% of business advice is terrible.

I’ve consulted hundreds of CEOs and made a list of widely accepted wisdom that you should ignore. 

Here are 7 “smart” sounding pieces of advice that are actually stupid:"



S - Sense Check: Do you understand the objective and the specific guidelines for this task?

Here is the output (using Claude this time as it’s a copywriting task):

Claude outputs multiple options, all of which have a good quality.

Chain of Density

I covered Chain of Density in a recent post, and there’s no doubt it’s one of the best prompt frameworks to add to your arsenal.

It’s amazing for article summaries, improving longer-form content via recursion or even improving your prompts, which you’ll see in this example.

Prompt template:

"Instructions: Here is [insert content you want to improve]. You will generate increasingly better versions of this content. 

Recursion: Repeat the following 2 steps 5 times for a total of 5 iterations. 

Step 1. Identify 1-3 points from the initial output which are missing.

Step 2. Write a new improved output of identical length which includes the missing points. 

Benchmark: Here is more information on what makes a good XYZ: [insert info] 

Additional guidelines: Follow these specific guidelines [insert guidelines]."

Example:

"Instructions: Here is a prompt I use to generate cold emails for my landing page creation agency: "I need a cold email idea that will engage my [ideal customer] with a unique perspective on [subject] and persuade them to take [desired action]"

You will generate increasingly improved versions of this prompt.

Recursion: Repeat the following 2 steps 5 times.

Step 1. Identify 1-3 points from the initial output which are missing.

Step 2. Write a new, improved output of identical length which includes the missing points.

Benchmark: Here is more information on what makes a good prompt:

Clarity and Specificity

Be Clear: The prompt should be easy to understand.

Be Specific: Vague questions will often yield vague answers. The more specific you are, the better the AI can generate a targeted response.

Open-Ended vs Closed-Ended

Open-Ended: Use open-ended questions for comprehensive answers or when you want the AI to generate multiple ideas.

Closed-Ended: Use yes/no or either/or questions when you need a straightforward answer.

Context

Provide Context: The more context you give, the more the AI understands the scenario you're asking about.

Time-Frame: If the question is dependent on a particular time or sequence, make sure to include it.

Purpose

Define Your Objective: Whether it's to gather information, generate text, solve a problem, or something else, make your objective clear.

Question Type: Tailor your prompt based on what you're looking for—explanations, summaries, creative writing, etc.

Simplicity

Avoid Jargon: Unless it's necessary, steer clear of complex language or industry-specific terms.

K.I.S.S. (Keep It Simple, Stupid): Sometimes, simpler prompts are more effective in getting the desired response.

Additional guidelines: Remember to repeat the process 5 times."

And the final iteration from the output (using Claude again):

I think that’s a considerable improvement on the original prompt. It’s more specific in who to target and adds in more variables to improve the persuasiveness.

Thanks for reading!

If you found this helpful, please support me by forwarding this newsletter to a friend or colleague.

See you in the next one!

Moritz