GPT Prompting Frameworks - What they are and how they can be more effective in prompt engineering

The way you phrase your questions or commands can greatly impact the quality of the response you receive. Prompting frameworks are structured approaches that help users get the best possible results from AI tools like ChatGPT.

These frameworks can transform your interactions - making them clearer, more effective, and more efficient. In this guide, we’ll explore several popular frameworks and provide examples for both novice and expert users.

Why Prompting Frameworks Matter

Prompting frameworks are simple yet powerful tools for improving your interactions with AI. They provide a structured way to communicate clearly and concisely, which leads to better answers and less back-and-forth. By specifying roles, actions, and expectations, frameworks can help even the most complex queries become more manageable - whether you’re using AI to generate ideas, solve problems, or write content.

Both beginners and experts can benefit from these frameworks: beginners gain a clear path for structuring requests, while experts can harness them to achieve specific, nuanced outcomes. Let’s take a closer look at nine useful prompting frameworks and how to use them.

Overview of the Frameworks

We’ll be exploring the following frameworks:

• ERA (Expectation, Role, Action)
• APE (Action, Purpose, Expectation)
• TAE (Task, Action, Goal)
• CARE (Context, Action, Result, Example)
• RACE (Role, Action, Context, Expectation)
• RISE (Role, Input, Steps, Expectation)
• ROSES (Role, Objective, Scenario, Expected Solution, Steps)
• COAST (Context, Objective, Actions, Scenario, Task)
• TRACE (Task, Request, Action, Context, Example)

Each of these frameworks provides a slightly different approach depending on the type of response you need and the level of detail you require.

Detailed Exploration of Frameworks for Novices and Experts

Let's explore each framework in detail, providing specific examples tailored for both novices and experts.

ERA (Expectation, Role, Action)

Definition: This framework helps set clear expectations, roles, and actions for precise responses.

• Novice Example 1: "You are a teacher. Explain what photosynthesis is to a 10-year-old student."
  • Here, the role is defined (teacher), and the expectation is that the explanation is age-appropriate.
• Novice Example 2: "You are a customer service representative. Describe how to reset a password to a non-technical user."
  • This prompt defines the role (customer service representative) and ensures the action (resetting a password) is explained in simple terms.
• Expert Example 1: "As an SEO analyst, provide an audit of my blog post. List improvements and associated strategies for each improvement."
  • The role (SEO analyst) and action (blog audit) are clearly defined, which helps to get actionable, targeted advice.
• Expert Example 2: "You are a data scientist. Analyse the provided dataset and identify three key insights, focusing on customer engagement."
  • This prompt clearly defines the role and focuses on extracting targeted insights, useful for complex data analysis.

APE (Action, Purpose, Expectation)

Definition: This framework focuses on defining the purpose behind the action.

• Novice Example 1: "Write a friendly email inviting someone to an event, with a clear call to action at the end."
  • This prompt ensures that both the action (writing) and the goal (inviting someone) are clear.
• Novice Example 2: "Create a to-do list for a student preparing for exams, making sure the tasks are manageable."
  • This prompt helps in defining clear actions with a purposeful breakdown of tasks.
• Expert Example 1: "Compose a formal request for research data, ensuring the tone is respectful and purpose aligns with the organisation’s goals."
  • This prompt layers additional context to ensure alignment with organisational standards.
• Expert Example 2: "Draft a proposal for a collaborative research project, clearly outlining the benefits for each partner and expected outcomes."
  • This prompt ensures the action (drafting a proposal) is done with a focus on mutual benefits and specific results.

TAE (Task, Action, Goal)

Definition: Clearly describes a task, the actions to complete it, and the desired goal.

• Novice Example 1: "Tell me the steps to bake a simple cake, and what the goal of each step is."
  • This prompt ensures that each action and its corresponding purpose are clear.
• Novice Example 2: "Explain how to organise a bookshelf, detailing each action and its purpose."
  • This prompt helps the user understand each task involved in organising a bookshelf, making it more manageable.
• Expert Example 1: "Create a project plan for developing a software feature, detailing coding tasks and end objectives."
  • The clear articulation of tasks and goals helps generate a comprehensive project plan.
• Expert Example 2: "Outline the steps required to deploy a machine learning model, specifying the goal of each phase."
  • This prompt helps experts understand the complex actions involved in deploying a model, with clear objectives for each phase.

CARE (Context, Action, Result, Example)

Definition: This framework helps provide context, specify actions, define expected results, and give examples.

• Expert Example 1: "As a marketing manager, analyse the recent campaign's performance. Provide insights into what worked and what didn't, and suggest improvements. Include examples from the data."
  • This prompt ensures the marketing manager uses contextual data to generate actionable insights and specific examples.
• Expert Example 2: "You are a software architect. Review the proposed system design, identify potential bottlenecks, and suggest improvements, providing real-world examples of similar challenges."
  • This prompt gives a clear role and action, focusing on contextual analysis and real-life applicability.

RACE (Role, Action, Context, Expectation)

Definition: This framework assigns a role, defines an action, provides context, and sets an expectation.

• Expert Example 1: "As a customer support manager, draft a response to a customer complaint about delayed delivery, including context about recent logistical issues and setting expectations for future deliveries."
  • This prompt ensures the support manager acknowledges the complaint and provides an informed response that sets realistic expectations.
• Expert Example 2: "You are a financial analyst. Create a summary of the quarterly performance for the executive team, incorporating the market context and setting expectations for next quarter's growth."
  • This prompt requires the analyst to include market trends and context, providing a summary tailored for an executive audience.

RISE (Role, Input, Steps, Expectation)

Definition: This framework specifies a role, provides inputs, defines steps, and sets expectations.

• Expert Example 1: "As a business consultant, evaluate the client’s current growth strategy using the provided market data. Outline steps to improve their market positioning and set clear success metrics."
  • The consultant role, combined with specific steps and metrics, ensures a well-rounded response.
• Expert Example 2: "You are an HR specialist. Review the provided employee engagement survey results and create an action plan with defined steps to address key concerns, setting clear outcomes for improvement."
  • This prompt ensures the HR specialist takes concrete actions with well-defined outcomes to improve engagement.

ROSES (Role, Objective, Scenario, Expected Solution, Steps)

Definition: This framework details a role, objective, scenario, expected solution, and steps to achieve it.

• Expert Example 1: "As a cybersecurity analyst, your objective is to enhance the organisation's security posture. Given the scenario of a recent phishing attempt, propose a solution and outline the steps needed to implement it."
  • The framework helps in developing a thorough response, addressing the scenario with clear solutions and actions.
• Expert Example 2: "You are a product manager. Your objective is to improve the onboarding experience for new users. Given customer feedback about difficulties in setup, propose a solution and detail the steps needed for implementation."
  • This prompt leads to a practical solution-driven approach that aims to improve user experience.

COAST (Context, Objective, Actions, Scenario, Task)

Definition: This framework uses context, defines objectives, specifies actions, describes scenarios, and outlines tasks.

• Expert Example 1: "As a project manager, use the context of limited resources to plan the next sprint. Outline the objectives, actions required, and tasks for each team member."
  • This ensures that the project manager effectively navigates resource constraints while achieving set objectives.
• Expert Example 2: "You are a sustainability consultant. Given the context of rising energy costs, develop an action plan for reducing energy consumption in a manufacturing facility. Outline specific tasks for the operations team."
  • This prompt defines a clear context and provides actionable steps, ensuring a practical plan.

TRACE (Task, Request, Action, Context, Example)

Definition: This framework provides a task, makes a request, specifies an action, gives context, and includes an example.

• Expert Example 1: "As a logistics coordinator, your task is to optimise delivery routes. Request data on current routes, analyse the efficiency, and propose improvements based on traffic data, including a specific example of a high-traffic route."
  • This ensures that the logistics coordinator provides detailed analysis and improvement strategies.
• Expert Example 2: "You are a curriculum developer. Your task is to enhance the science module for high school students. Request feedback from current teachers, propose changes, and include an example of an improved lesson plan."
  • The prompt guides the curriculum developer to gather feedback and implement specific improvements.

Practical Framework Comparisons

To illustrate how different frameworks work, let’s consider a practical example—writing a product review.

• ERA: "As a consumer, write a review of this product highlighting its pros and cons."
• TRACE: "Write a review for this product, including the key features, your experience using it, and an example of when it was most useful."
• COAST: "Describe the context of using this product, the objective it helped achieve, and the steps taken to get the most out of it."

By comparing outcomes from different frameworks, it becomes clear that each one helps focus the review differently, giving both novices and experts a variety of tools to achieve specific results.

Tips for Novices and Experts

• Novices:

  • Start simple! Choose a framework that feels intuitive and helps you organize your thoughts.
  • CARE and TRACE are great for beginners who need a bit of extra structure and context.

• Experts:

  • Stack frameworks for deeper, layered prompts. For example, use COAST to establish context and RACE to detail the actions.
  • Experiment to get nuanced responses—the more details and clarity you provide, the better the output will be.

Creating Effective Prompts with GPT Frameworks

No matter your skill level, crafting effective prompts takes a bit of practice. Here are a few strategies:

• Be Specific: The clearer your request, the better the response. Use precise language.
• Iterate: Don’t be afraid to refine your prompt if the first response isn’t quite right.
• Framework Selection: Choose the framework that best matches the complexity and detail of your request—for example, RISE is great when you need step-by-step guidance.

Summary and Further Notes

Prompting frameworks can make your experience with AI far more productive, whether you’re a complete beginner or a seasoned user. Each framework is designed to help you communicate clearly and get the most relevant and precise response. Take time to experiment with different approaches, use these frameworks to refine your interactions, and see how they improve the quality of your results.