Step-by-Step Solution

1. Choose a Scenario

• Think of a practical use case relevant to you or your organization:

  1. Content Creation (e.g., generating product descriptions, blog posts, marketing copy).

  2. Customer Support (e.g., drafting answers for FAQs).

  3. Text Analysis (e.g., summarizing a policy document).

For this example, we’ll pick “Content creation for a new product launch.”

2. Formulate a Sample Prompt

• The prompt should be specific enough to see how the LLM responds. For instance:
  “Write a short, engaging product description for a new eco-friendly water bottle aimed at young professionals. Emphasize its sustainable materials, sleek design, and convenience for daily use.”

1. The prompt includes:

   • Target audience: young professionals

   • Key features: eco-friendly materials, sleek design, daily convenience

   • Desired style: short, engaging

3. Input the Prompt into Your Chosen LLM

• Examples:

  • ChatGPT (GPT-3.5): https://chat.openai.com/

  • Bard: https://bard.google.com/

Paste the prompt above and let the model generate a response.

4. Observe the Output

• Check how the model addresses your request:

  1. Does it mention the eco-friendly aspect?

  2. Does it keep it short and engaging for young professionals?

  3. Is the tone formal, casual, or brand-aligned?

Take notes on any strong points or missed details.

5. Identify Improvements or Limitations

1. Improvements:

   • Maybe the LLM forgot to mention “BPA-free” or “recyclable packaging,” so you might refine your prompt.

   • Or the text is too long, requiring you to ask for fewer words or a bullet-list format.
     
     

2. Limitations:

   • The LLM might produce generic phrases (“ultimate companion,” “revolutionary design”) that lack uniqueness.

   • If you need strict brand voice guidelines, it might not capture them unless you specify deeper style instructions.
     
     

6. Refine Prompt (Optional)

• 1. If time permits, do an iterative approach:

     • Re-run the prompt with added constraints:

       1. “Keep it under 70 words,” or

       2. “Use a friendly, informal style,” or

       3. “Mention it’s made from 100% recycled stainless steel.”

     • See if the output improves or addresses prior gaps.
       
       

Example Demonstration

Scenario: “Content creation for a new product launch: an eco-friendly water bottle for young professionals.”

Sample Prompt:

“Write a short, engaging product description for a new eco-friendly water bottle aimed at young professionals. Emphasize its sustainable materials, sleek design, and convenience for daily use.”

GPT-3.5 (Example Output)

“Stay stylish and sustainable with our new eco-friendly water bottle, crafted from 100% recycled stainless steel. Its sleek, modern design easily fits into your daily routine—whether you’re commuting to the office or hitting the gym. Enjoy leak-proof convenience and a guilt-free sip every time, knowing you’re reducing plastic waste and protecting the planet.”

Observations:

• Strengths: mention of “recycled stainless steel,” references daily routine, short + engaging tone.

• Limitations: might be too broad. If you need a brand-specific voice or exact word count, that’s missing.
  Potential Refinement:

“Can you make it 50 words or fewer, and highlight that it comes in multiple stylish colors?”

Conclusion

By following these steps:

1. You chose a real scenario: “content creation for a product launch.”

2. You prepared a targeted prompt.

3. You observed how the LLM responded and identified strengths + limitations.

4. (Optionally) you iterated to refine the output.

Result: A deeper understanding of how LLMs handle domain-specific prompts, providing you with practical insight on using AI for everyday tasks.