Prompt Engineering for Technical Writing 101
Chapter 2: The 5 Step Prompt Engineering Framework

Step 3 – Define the Project Scope, Goals, and Type of Content

Complete & Continue Next Lesson Learn More

Outcomes

  • Define the primary goals, scope, and type of content for a technical writing project to create focused and compelling educational materials.

  • Develop a level-based content rubric to structure technical writing projects, enhancing the depth and progression of educational content.

In This Lesson

  • Outcomes

  • Step 3 – Define the Project Scope, Goals, and Type of Content

    • Watch out for Hallucinations

Step 3 – Define the Project Scope, Goals, and Type of Content

We need to define the primary goals of your project’s output clearly. This could include providing a comprehensive guide, explaining a specific concept, or offering troubleshooting assistance. A well-defined purpose will help you create focused and compelling content.  

I write technical learning plans for various topics. Below is a level-based content rubric I use to define the technical flow of multiple projects.  This can be integrated into the prompting sequence as a guide.

Technical
Level
Topic
Description
000
Prerequisites
This level introduces integration, optimizations, and deeper concepts, including deployment and operations.
100
Introduction
Provides basic information on the learning topic. Review solution architectures and case studies that put the technology into context
200
Basic Concepts
Provides basic information on the learning topic. Review solution architectures and case studies that put the technology into context
300
Optimization and Deep Technical
This level introduces integration, optimizations, and deeper concepts, including deployment and operations.
400
Best Practices and Use Cases
Use Cases, recommended practices, and deeper dives are reviewed to help bridge the theory into practice while showing the technology in a real-world context.
500
Resources
Other resources that can be included that provide value.

My Content Guide on Creating Learning Guides and Workshops

See Agile Learning for Tech Professionals: Embrace Guided Learning Paths - AI TechED Insights for a deeper discussion on the rubrics.

To use ChatGPT for technical writing, start by crafting a seed phrase that clearly conveys the topic. The more specific you can be is the key to ensuring accurate and relevant results. For example, if you are looking for an outline or writing to create a lesson plan, your seed phrase could be something like the following;

Prompt:

Keeping with the inputs so far, please re-do the summary and outline based on the following.

The goal of Content = I need to teach my students from beginner level to experts so that at the end of the lessons, they will be competent self-service data analysts.

Scope of Content - The content needs to have the following;

- a Learning Plan with numbered levels

- The learning plan is based on levels where the content gets deeper into the topic at each level. The difficulty level is, Level 100 = Introduction, Level 200 = Basic Concepts, Level 300 = Optimization and Deep Technical content, Level 400 = Best Practices and Use Cases, Level 500 = Resources and next steps

The following is the resulting output.

Chat Response from adding in the numbered levels

The content has now been modified based on the level structure. This allows us to reference portions that need to be changed.

Let’s split up Level 100 Point 3, Setting Up and Navigating Power BI. Also, respond on the accuracy of what was produced.

Prompt:
That was great.
Can you split up Level 100, Point 3 into 2? Point 3 becomes "Setting Up Power BI", and the new Point 4 becomes "How to Navigate Power BI" and provide the list again without any further changes?

The following is the resulting output.

Tip: Use the Copy Content to Clipboard. This can retain more of the formatting of the content.

Watch out for Hallucinations

As we saw in the previous module, The Generative AI Revolution, AI tools can experience hallucinations when their predicted responses are inaccurate or lack factual consistency. These tools, for the most part, function by predicting word strings that match the query. Remember that they are next-word predictors and cannot apply logic or fact-check. Such instances are not considered malfunctions but result from the model’s design philosophy and decision-making process.

See the following resources:

Remember to Validate and Fact Check!! Rely on your expertise to drive the results !! Don’t take anything at face value.

The key point to remember is this is the step in which you provide the prompt with your voice for the content you want. Your brand voice is essential in your educational content, as your lessons need a consistent voice, format, and flow.