AI-Powered Service Desk: We Tested Best AI Prompt Frameworks

Posted on Wednesday 23 August 2023.

When it comes to ChatGPT prompts, finding the right framework is key!

It can totally transform your AI-generated responses, making them more accurate and on point. So, whether you’re a Service Desk Manager, a developer, a Marketer, a content creator, or just curious about the fascinating world of AI, exploring these commonly used ChatGPT prompt frameworks can be a game-changer.

In this article, we will provide you with a list of prompt frameworks to improve your ChatGPT communication. And share our experience with some of them using ChatGPT-3.5 and ChatGPT-4!

So, let’s get started.

What Is a ChatGPT or AI Prompt?

Before we dive into the popular ones, let’s understand what ChatGPT prompts, or AI prompts are and why they matter.

Simply put, a prompt is a question or set of instructions used to start or guide a task or conversation. Think of it as a starting point for the AI to create content. You give ChatGPT a prompt, and it uses its massive brain (trained on large datasets) to come up with cool and relevant stuff.

It’s an essential component of Generative AI. Generative AI is a field of artificial intelligence that has the capability to generate new content based on the input they receive.

And it’s rapidly gaining momentum due to its transformative potential across diverse sectors and its capacity to produce creative and valuable outputs.

Think about it as an AI artist that can paint pictures, write stories, and even have conversations! And guess what? Your prompt is the brushstroke that guides this AI artist.

When provided with a prompt or input, the Generative AI leverages this learned knowledge to create relevant and coherent outputs.

And that’s where things get exciting! When it comes to the service desk, utilising AI prompts can help the tools generate prompt and accurate responses to customer inquiries and technical problems.

This streamlines the support process, reduces response times, and ensures consistent and reliable assistance. Additionally, Generative AI can handle a wide range of queries, freeing up support agents to focus on more complex tasks.

And that can ultimately lead to improved customer satisfaction and a more efficient service desk operation.

The Magical World of AI Prompts

Prompt: “Once upon a time in a land far, far away…”

Generated Output: ” In a distant realm where magical wonders and mystical creatures roamed, a captivating tale begins to unfold…”

By using different prompt frameworks, you can shape how the AI responds. You get more control, more relevant answers, and an overall better experience.

The prompt can be a short phrase, a sentence, or even a paragraph in the example above.

And there are several different ways you can form the prompts. From question-answer prompt, task-oriented, opinion to storytelling prompt frameworks.  

The better the prompt, the more accurate and relevant the generated output tends to be.

Here are the Most Popular ChatGPT Prompt Frameworks

By incorporating some of the following frameworks into your prompts, you can effectively guide ChatGPT to generate more engaging, persuasive, and personalised outputs that resonate with your audience.  

In the next section, we’ll share our comparative analysis between the ChatGPT-3.5 and ChatGPT-4 models, particularly in the context of prompts and their resultant outputs.

Here’s a list of some popular prompt frameworks, which we tested in both tools.

#1 APE: Action, Purpose, Expectation


First on our list is a popular framework called APE.

The first step is to define the specific job or activity that needs to be done (Action).

Next, discuss the intention or goal behind the task (Purpose).

Finally, state the desired outcome that is expected to be achieved (Expectation).

⚡ Prompt example:

Could you help develop a content marketing strategy for our new research report “Worklife in IT Industry”? We aim to generate interest and increased awareness among our target audience of IT professionals who are passionate about the future of work and tech trends. The strategy should engage our audience and drive an increase in pdf downloads by at least 25%.

#2 RTF: Role, Task, Format

RTF AI prompt

To effectively utilise this prompt, you need to establish its function (Role), specify the job you want it to perform (Task), and clearly state the desired result or how you want the result to be displayed (Format).

⚡  Prompt example:

I want you to act as an interviewer. I want you to ask the competency-based interview questions for the position of Service Desk Analyst. All questions should require the candidate to demonstrate knowledge of the industry. Show this as a list of 5 questions.

#3 TAG: Task, Action, Goal

To use this framework, start by defining the specific task at hand (Task).

Begin by clearly stating what action you require (Action).

And provide more information to clarify the desired goal (Goal).

⚡ Prompt example:

The task is to evaluate the performance of team members. As a Service Desk Manager, you have to assess the strengths and weaknesses of team members, provide constructive feedback and define the metrics you would use to evaluate the performance. The end goal is to improve team performance so that the average user satisfaction score moves from 6 to 7.5 in the next quarter.

#4 ERA: Expectation, Role, Action

Unlike the previous framework, ERA starts by describing the desired outcome (Expectations).

Following that, the role of ChatGPT is specified (Role), and the necessary actions are outlined (Action).

⚡ Prompt example:

We expect to increase the number of knowledge base visits from promotional end-user email campaigns by 20% in the next quarter. Imagine you are the knowledge manager, and your responsibility includes defining and successfully executing a knowledge usage email promotion strategy. To achieve this goal, please provide a detailed plan and strategy for subject line optimisation and segmentation.

#5 RACE: Role, Action, Context, Expectation

RACE AI prompt

The RAC framework has four components.

The first is Role, which involves specifying the role of ChatGPT.

The second is Action, which details what action is necessary.

The third is Context, which provides relevant details about the situation.

The fourth is Expectation, which describes the expected outcome.

⚡ Prompt example:

Imagine you are an IT Helpdesk Analyst, explain a troubleshooting process and suggest a set of actions, this is for a customer who has a troubleshooting issue with their slow computer. The expected outcome is to select a set of practical and effective tactics to solve this problem quickly with a high level of customer satisfaction.

How Accurate Are the Generated Responses

Now that we fed the same prompts into both of these powerful tools let’s do a quick analysis.

What did we find? Well, it’s like comparing a trusty old map to a futuristic GPS navigation system. The responses were coherent, contextually on-point, and pretty darn impressive. 

ChatGPT-3.5 Results

✅ Pros:

Coherence and Contextual Comprehension: The answers showed a strong level of consistency and comprehension of the context, reflecting that ChatGPT-3.5 can understand input prompts accurately.

Informative Precision: The generated answers were mostly correct and based on facts, making them suitable for providing brief information.

Structured Responses: The outputs often followed a structured format, mirroring a well-informed conversation that delivers data.

❌ Cons:

Limited Nuance: While contextually relevant, responses occasionally lacked important details like tones or subtle changes in what was said.

Lack of Creativity: Outputs were primarily informative in nature, with minimal creative flair or inventive language use.

ChatGPT-4 Results

✅ Pros:

Contextual Nuance: ChatGPT-4 demonstrated an advanced ability to grasp the nuances of prompts, resulting in responses that were finely tailored to the context and intent.

Enhanced Creativity: Generated outputs displayed more creativity with interesting language and innovative expressions.

Semantic Depth The responses showed a thorough comprehension of the input, incorporating both factual accuracy and nuanced semantic details.

❌ Cons:

Complexity Management: The responses, while sophisticated, occasionally veered into intricate structures that might not align with straightforward communication needs.

Potential Over-Contextualization: The nuanced responses might overemphasise certain aspects of the prompt, potentially leading to outputs that are overly specific.

Explore Other Popular ChatGPT Prompts

COAST AI prompt

COAST: Context, Objective, Actions, Scenario, Task

  • Context: Start by setting the foundation for the conversation.
  • Objective: Describe the desired goal.
  • Actions: Explain in detail what actions are required.
  • Scenario: Provide a description of the scenario.
  • Task: Describe the task.

GRADE: Goal, Request, Action, Details, Example

  • Goal: Define the overall objective or purpose.
  • Request: Outline the specific task or inquiry that needs to be fulfilled.
  • Action: Describe the steps or actions that should be taken.
  • Details: Provide additional information or context that is relevant to understanding and executing the action.
  • Example: Provide a scenario that demonstrates how the goal, request, action, and details components come together.

RISE: Role, Input, Steps, Expectation

  • Role: Defines the specific role or function that a person or entity plays in the scenario.
  • Input: Specify the necessary inputs or factors that need to be considered.
  • Steps: Ask for detailed steps that need to be taken to achieve the desired outcome.
  • Expectation: Describe the desired result, outcome, or expectation.

RASCEF: Role, Action, Steps, Context, Examples, Format

  • Role: Define the specific function.
  • Action: Describe the main action that needs to be performed.
  • Steps: Outline the set of steps or tasks that need to be performed.
  • Context: Provide more information about the setting in which the scenario takes place.
  • Examples: Ask to provide concrete examples.
  • Format: Define how you want to present the output.

SPARK: Situation, Problem, Aspiration, Results, Kismet

  • Situation: Describe the context in which you’re operating
  • Problem: Identify the specific problem
  • Aspiration: State the desired vision for success
  • Results: Define the specific and measurable results
  • Kismet: Add an element of surprise.

Final Thoughts

Just as a tiny spark starts a fire, well-crafted prompts ignite AI’s creative brilliance! By providing concise instructions, you guide ChatGPT’s focus and help it avoid potential pitfalls that might lead to irrelevant or nonsensical answers.

These prompt frameworks provide valuable insights on how to use Generative AI tools to enhance your thought process without replacing it. Always remember that the quality of your prompt shapes the quality of the AI-generated output.

📢 We would love to hear about your experience with Generative AI tools!

Have you had a chance to try any of the prompts? If you have, we’d love to know which ones were your favourites.

Get in touch and share your thoughts!

Antonija Bozickovic

Antonija Bozickovic

Content creator at SDI



close slider
  • Feel free to get in touch. To ask a question directly please submit your message using the form below.
  • Can we send you updates and offers to help you be brilliant?
  • This field is for validation purposes and should be left unchanged.