The Ultimate Guide to Creating AI Prompts on PromptBros.ai
Learn Prompt Engineering, the easy way.
Introduction
If you haven't heard of AI prompts, they are the key ingredients for AI to function.
They are set of instructions from users that turn AIs into useful assistants. From marketeers, developers, writers, or content creators, with a well-writen Prompt, AIs can take that role more effectively.
Whether you're looking to generate creative content, streamline coding tasks, or enhance user interactions, crafting effective AI prompts is the important and exciting step that leads to a worthwhile AI Interaction, and avoiding meaningless or, even worse, misleading conversations with a virtual ignorant.
// image with AI being a donkey vs an AI that knows what it's doing
This comprehensive guide will walk through:
- What types of AI prompts are there?
- How do I craft Prompts in PromptBros?
- What are the best practices to follow?
- Are there frameworks and ways to test and improve them?
- Do you have some examples of successful AI prompts?
What types of AI Prompts are there?
AI prompts are carefully crafted inputs or instructions that guide an artificial intelligence model to generate a desired output.
These prompts are usually separated in 2 types:
- User Prompts - these are asked as input to a conversation and can range from simple questions to complex scenarios that require detailed responses. They serve as the starting point for the AI’s generation process without any influence to their standard behaviour.
For instance, a simple prompt like "Write a poem about the ocean" directs the AI to focus on poetic elements and themes related to the ocean. On a more technical note, a prompt like "Generate Python code to sort a list" instructs the AI to produce a specific coding solution.
The User Prompts in these examples is robust enough to provide the quality and performance of the AI tool. Let's look at how to build a robust prompt that returns the depth you need in knowledge, and how to avoid simple mistakes together.
- System Prompts - these are usually structured sets and, although they can be passed as User Prompts, when set as System Prompts they carry a more permanent influence to the behaviour of the AI models' response throughout the conversation.
An example can be "Please act as a TV presenter host, and never break character. You are hosting a 'Guess the Phrase' game and should only interact in this way. IMPORTANT: Guess the Phrase is a game where the objective is to guess the missing letters on a phrase. Guess the Phrase Players take turns guessing letters in a word using clues. Each wrong guess lowers an anvil, and players lose if they make five wrong guesses."
Steps to Create an AI Prompt on PromptBros.ai
Creating an AI prompt on PromptBros.ai is straightforward but requires some strategic thinking to achieve the best results.
For basic info on how to create a prompt, please see the Getting-Started with prompts.
Let's go step by step and craft a great Prompt:
Step 1: Understand Your Objective
Before you start, clearly define what you want the AI to achieve. Are you looking for creative text, technical solutions, or user interaction scenarios?
If you are looking to brainstorm ideas for AI Prompts, you can use our Brainstorming Agent to help.
Knowing your end goal will help you frame a more effective prompt, and the next step will help you do this.
Step 2: Decide your AI Framework
Selecting a Prompt Framework helps to organise your thoughts and structure the key elements you should carefully provide.
We recommend trying out our Prompt Builder Agent that is instructed to help apply these to your objective.
Below are some of the most commonly used frameworks.
-
RACE
- Role: Specify role.
- Action: Mention action needed.
- Context: Provide background information.
- Explanation: Describe your outcome.
-
CARE
- Context: Give background information.
- Action: Mention action needed.
- Result: Mention your goal.
- Example: Give some example outputs.
-
APE
- Action: Define job to be done.
- Purpose: Mention your goal.
- Execution: Describe the outcome you want.
-
CREATE
- Character: Give a role.
- Request: Define job to be done.
- Examples: Give some example outputs.
- Adjustment: Provide betterment instructions.
- Type of output: Specify output format.
- Extras: Add more context.
-
TAG
- Task: Define task.
- Action: Define job to be done.
- Goal: Explain end goal.
-
CREO
- Context: Give background information.
- Request: Define job to be done.
- Explanation: Explain the task.
- Outcome: Describe the outcome.
-
RISE
- Role: Mention role.
- Input: Give context and instructions.
- Steps: Ask for stepwise output.
- Execution: Describe the outcome.
-
PAIN
- Problem: Describe problem.
- Action: Mention job to be done.
- Information: Ask for details.
- Next Steps: Ask for resources.
-
COAST
- Context: Give background information.
- Objective: Mention your goal.
- Actions: Explain all actions needed.
- Scenario: Mention your problem.
- Task: Mention job to be done.
-
ROSES
- Role: Mention role.
- Objective: Mention result you need.
- Scenario: Give background information.
- Expected solution: Describe outcome.
- Steps: Ask for steps for the outcome.
Step 3: Craft your Prompt
Using the framework style of your choice, add your prompt being as clear and specific as possible to ensure the AI understands your requirements.
For example, using the RISE methodology, instead of just typing
Role: Software Developer. Input: You need to write software code. Steps: Generate Code. Execution: Output will be working software code.
The above will not return anything useful without several conversations with the AI to provide more details, and it will be quite inefficient.
You should be more specific. See here a revised better example:
Role: You are a Python Back-end Software Developer. Input: You are building a Back-end Python 3.12 application using Flask. Steps: We should take this step by step. I will give you the functionality that needs development. I want you to ask me for anything you want to know about the feature such as data needs, API parameters, or other requirements that you need to build the software code. You will use TDD approach so that the code comes with relevant test setup and cases. Execution: You will provide the python code using Flask, recommend libraries, provide comments and explanation to the code, and use best practices for Python development and code maintenance.
The above will return a better interaction for a complete requirement much faster.
Step 4: Test and Refine
Once your prompt is configured, test it by generating a few outputs. Review these outputs to see if they meet your expectations. If not, tweak your prompt and settings accordingly.
Step 5: Save and Deploy
Once you are happy with your prompt, you can save it and deploy it directly on any AI Provider.
Best Practices for Effective AI Prompts
Creating an effective AI prompt involves more than just typing a sentence. Here are some best practices to ensure your prompts yield high-quality outputs:
Be Specific
The more specific your prompt, the better the AI can understand what you’re asking for. For example, rather than saying "Write a story," specify the genre, characters, and setting.
Use Clear Language
Avoid ambiguity in your prompts. Clear, concise language helps the AI understand your intent without misinterpretation.
Provide Context
Giving context can significantly improve the quality of the output. For instance, if you're asking for a piece of coding, provide a brief description of its intended use.
Experiment with Settings
Don’t be afraid to adjust the AI model settings such as temperature.
//picture for parameters or explanation
These adjustments can greatly influence the creativity and length of the generated output.
Iterate and Refine
Creating the perfect prompt often requires some trial and error. Continuously refine your prompt based on the outputs you receive.
Examples of Successful AI Prompts
To give you a clearer idea, here are some examples of successful AI prompts across different domains:
Content Creation
Prompt: "Write a blog post about the benefits of a plant-based diet, highlighting health, environmental, and ethical considerations." Outcome: The AI generates a comprehensive blog post covering all specified areas, suitable for publication with minimal edits.
Coding
Prompt: "Generate a JavaScript function that takes an array of numbers and returns a new array with each number doubled." Outcome: The AI produces a functional JavaScript code snippet that meets the given requirement.
Customer Support
Prompt: "Provide a polite and informative response to a customer asking for a refund due to a delayed shipment." Outcome: The AI crafts a diplomatic and empathetic reply, addressing the customer’s concerns effectively.
Creative Writing
Prompt: "Compose a short story about a time traveler who accidentally disrupts an important historical event." Outcome: The AI creates an engaging narrative with a clear beginning, middle, and end, incorporating the time travel theme effectively.
Conclusion
AI prompts are powerful tools that can unlock the full potential of artificial intelligence for various applications. By understanding what AI prompts are, following the steps to create them on PromptBros.ai, adhering to best practices, and learning from successful examples, you can harness AI to achieve your specific goals. Start experimenting with your own prompts today and explore the endless possibilities that AI can offer.