Skip to content
All posts

AI Mini Course Part 1 - Prompting

By  Jorrit Posor  ·  3 minute read

Welcome to Our AI Mini Course

Hello, and welcome to our AI mini course! The goal of this course is to provide you with a foundational understanding of this exciting new technology called AI. Drawing upon our experiences and insights gathered while developing Zen AI—a tool for AI knowledge work assistance and workflow automation—we aim to share some hopefully valuable knowledge with you.

Each week, we will post a new video on our YouTube channel, so be sure to check in regularly to gather fresh insights and see how we leverage this innovative technology. Let's dive into today's topic.

What is Prompting?

Prompting essentially means instructing AI to perform specific tasks for you. You can think of it as giving instructions to a new, slightly junior colleague who has just joined your team. Whenever you provide an instruction, it's crucial to ensure that this new 'AI colleague' understands you well. I'll explain this further in a moment.

Let's imagine our goal: creating a new job description and posting it on our company's website. Usually, we already have a number of existing job descriptions on our website, and our aim is to use our AI colleague to facilitate this process.

When communicating or prompting AI, there are several key points to keep in mind.

Key Considerations for Effective AI Prompting

Provide a Starting Point

Firstly, it's essential to give your AI a starting point. Explain where you are coming from and what your goal is. For example, you might say, "Hey, I have an existing job description here. Please create a new job description based on this one." Providing context helps the AI understand your requirements better.

Instruct Clearly

Instead of giving vague instructions like "Please create a new job description," it's more effective to be specific. You might say, "Please create a new job description with the following details on salary, responsibility, and using sections similar to the existing job description." The more detailed and clear your instructions, the better the results will be.

Stay Focused

It's often better to generate one artifact at a time. For instance, create the job description first and then move on to creating the LinkedIn post, rather than attempting to do both simultaneously. Breaking down your task into smaller subtasks and allowing your AI to handle each one sequentially can significantly improve the quality of the output.

Practical Example: Creating a Job Description with AI

To illustrate these points, let's look at a practical example. We'll showcase how this can be done using Zen AI in our workflow interface. For now, let's forget about chat and focus on the workflow aspect.

We start with an existing job description from Finn—an e-commerce company offering car subscriptions. These job descriptions usually follow a specific structure: they include sections like "About the Job," "Your Role," "Responsibilities," "Your Profile," and so on. We want our new job description to follow the same format.

job_description

We have also already gathered bullet points on qualifications, key responsibilities, and salary details for the new role.

additional_job_details

Now, let's instruct our AI to create a new job description.

Step-by-Step Instructions

  1. Add a Node for Context: We first add a node explaining the starting point and what we expect. For instance, we could write, "We have an existing job description from Finn. Please create a new job description for the following role using the same structure and similar sections."
  2. Include Specific Instructions: Make the prompt more specific. For example, we might add, "Include sections on salary, responsibilities, and qualifications using the existing job description's format."
  3. Check the Output: Once the AI generates the job description, we review it to ensure it follows the required structure, such as "About the Job," "Your Role," "Responsibilities," and so forth, even if the order might be slightly different.
  4. Stay Focused and Sequential: Remember our guideline about staying focused. After the job description is complete, we can then prompt the AI to create a LinkedIn post to promote this job opening, ensuring each task is handled independently for higher quality results.

When you try this yourself. you probably see that the AI-generated job description maintains the necessary sections and content accuracy.

job_desc_node

Ensuring Quality Through Focused Task Management

Following our structured approach, we achieved a detailed and well-formatted job description. The generated content adhered closely to the templates set by the existing job descriptions, ensuring consistency in presentation and style.

Creating a LinkedIn Post

After successfully generating the job description, we turn to our next task—creating a LinkedIn post to promote the new job opening. Again, we follow our guideline of staying focused by tackling one task at a time. We use the newly created job description as a reference point for the LinkedIn post.

  1. New Prompt for LinkedIn Post: We prompt the AI, "Please create a LinkedIn post for the new job description to help promote this job opening." Ensuring clarity and specific instructions here is equally critical.
  2. Review and Adjust: The AI quickly generates a LinkedIn post, which we review. It included the job title, key responsibilities, and an engaging call-to-action, making it an effective piece of promotional content.
  3. Consistency in Messaging: By structuring our instructions and focusing on one task at a time, we maintained consistency and quality across different content forms—whether it be a detailed job description or a succinct LinkedIn post.

Conclusion

By adhering to clear guidelines and focusing on specific, sequential tasks, we can effectively leverage AI to streamline and enhance our workflow. As an example, we looked at creating detailed job descriptions to drafting engaging LinkedIn posts. Our prompting best practices ensure high-quality outputs and maximize the utility of AI-driven tools like Zen AI.

We hope you found this session useful. Stay tuned for next week's course, where we'll continue to explore the vast possibilities of AI in our mini course.

Thank you for joining us, and see you next time!