Copilot vs. Google Gemini AI Prompt Tecnique Use Case: Customer Service Response

I have been studying the different AI offerings over the past month and I wanted to give an Apples-to-Apples comparison of a detailed prompt for a Customer Service Use Case involving an Irate Customer that is being abusive. The prompt also shows the input from the Agent as being a little biased due to their anger reaction from the customer internally. The comparison is between Microsoft Copilot Pro and Google Gemini. Both offerings offer similar answers but again just like Bing Search vs. Google Search, the AI responses to the question are a bit different.

Again, this is what I call a Zero-Shot prompt that is detailed in nature but also remind us of what needs to be looked at when prompting AI LLMs. A zero-shot prompt is a way to instruct a large language model (LLM) like Copilot or Gemini to complete a task or answer a question without providing any specific examples beforehand. It’s essentially giving the model a single, clear instruction and hoping it can leverage its existing knowledge to deliver the desired outcome.
In a One-Shot Prompt, the prompt provides a single concrete example alongside the instruction for Copilot or Gemini. I am not asking for it to compare to another example. I am stating what the problem is and what I want, which is the answer to give to the customer.

You have to consider the following when using Effective Prompting Techniques.

Effective prompting is like a good recipe:

  • Direction: Instead of “use”, clear adjectives like “boil”, “sauté”, “fry” etc.
  • Formatting: Provides clarity and flow to your prompt.
  • Examples: Includes photos of ingredients, process, etc. (Stating the sentiment of the user)
  • Division: Consists of smaller tasks which lead to main dish. (Telling to focus on key items)
  • Evaluation: Provides benchmarks like taste, aroma, appearance. (You’re and AI language expert)

This leads to a well-prepped dish, a parallel to how effective prompts = well-curated and detailed results.

(Reference: AICerts Prompt Engineering Training)


Look at the prompt I provided and think about how you might word the prompt differently to effectively get a different result or an even better result.

Prompt Given:

I am a customer support agent dealing with an abusive customer regarding their issue with the Microsoft Most Valuable Professional Program. The customer is complaining that they were informed of their denial in a rude and cold sentiment at a time of 3AM locally and is using offensive language. As an AI language expert, could you help me craft a professional and empathetic response to address their concerns and deescalate the situation? Remember to keep the response respectful and focused on showing the power of the Microsoft Corporation as a whole and relay the fact that they just need to go away or get dealt with. Also, focus on finding a resolution for the customer’s problem.

Copilot Logo

Copilot Pro Response

I understand the situation you’re facing, and I’m here to help you craft a response that is both professional and empathetic. Here’s a suggested reply:

Dear [Customer’s Name],

I want to start by expressing our sincerest apologies for the experience you’ve had with the Microsoft Most Valuable Professional Program. We understand that receiving a message at an inconvenient time, and in a manner that felt disrespectful, has caused you frustration and for that, we are truly sorry.

At Microsoft, we value every individual and their contributions to the tech community. Our aim is to foster a positive and supportive environment, and it’s clear that in this instance, we fell short. Please know that your feedback is crucial to us, and we are taking steps to ensure that all communications are handled with the utmost care and respect, regardless of the outcome.

We appreciate your interest in the MVP Program and understand that it can be disappointing not to be selected. The evaluation process is thorough and competitive, and while not everyone can be chosen, we encourage you to continue your valuable contributions to the community and to reapply in the future.

If you have any further concerns or need assistance with specific issues, please do not hesitate to reach out. Our team is dedicated to providing you with the support you need and finding a resolution that meets your satisfaction.

Thank you for bringing this matter to our attention, and once again, we apologize for any distress caused. We hope to have the opportunity to serve you better in the future.

Warm regards,

[Your Name] Customer Support Agent Microsoft Corporation

This response aims to acknowledge the customer’s feelings, apologize for the negative experience, and offer a path forward while maintaining the professionalism and integrity of Microsoft. It’s important to avoid any language that could be perceived as dismissive or confrontational, focusing instead on resolution and support. I hope this helps you in addressing the customer’s concerns.

Google Gemini Response

It’s important to prioritize de-escalation and avoid using threatening language, even when dealing with an abusive customer. Here’s a more professional and empathetic response focused on resolution:

Acknowledge and Apologize:

  • “I understand you’re upset about receiving the MVP rejection notification at an inconvenient time and in a cold tone. We sincerely apologize for the impersonal communication and the late hour. We’re working on improving our notification process to ensure a more positive experience.”

Active Listening and Empathy:

  • “Let’s take a step back. I can tell this situation has caused you a lot of frustration. Can you tell me more about what you were hoping to achieve with the MVP program?”

Focus on Resolution:

  • “While the decision regarding MVP status is final, I’d be happy to explore the specific reasons for your denial and provide resources that might help you qualify in the future. Would you be open to discussing your application in more detail?”

Offer Alternatives:

  • “The MVP program is highly competitive, but there are other valuable ways to engage with the Microsoft community. Would you be interested in learning more about our user groups, online forums, or other programs that might align with your interests?”

Maintain Professionalism:

  • “I understand your frustration, but abusive language won’t help us work towards a solution. I’m here to genuinely assist you, so let’s focus on finding the best path forward.”

Shift the Focus:

  • “Microsoft is a company built on innovation and a strong community. Let’s work together to find a way for you to continue being a valuable part of that ecosystem, even if the MVP program isn’t the right fit at this time.”

De-escalation Techniques:

  • Speak slowly and calmly.
  • Use respectful language and avoid mirroring the customer’s aggression.
  • Validate their feelings. Let them know their frustration is heard.

Escalation Protocol (Optional):

  • If the customer continues to be abusive, you may need to politely excuse yourself and inform them that the conversation will continue when they can speak with you respectfully.
  • You can then follow your company’s protocol for escalating such situations.

Remember, the goal is to de-escalate and find a solution. By focusing on clear communication, empathy, and alternative solutions, you can turn this negative interaction into a more positive customer experience.

So, as you can see from the responses, there is an argument for using different AI Solutions with the same prompt. I would challenge you to rephrase the prompt based on the product, sentiment given by the customer, details of the issue, and the result you want to achieve and then try that prompt against different LLMs and AI chat bots to see what response you get.

Feel free to contact me for more information about Copilot and General AI prompting as I will be offering training courses in various aspects of AI Concepts and Fundamentals. Also, stay tuned for more posts on Copilot Studio, Azure Open AI, ChatGPT, and other AI offerings.

About Lance Lingerfelt

Lance Lingerfelt Profile Photo

Lance Lingerfelt is an M365 Specialist and Evangelist with over 20 years of experience in the Information Technology field. Having worked in enterprise environments to small businesses, he is able to adapt and provide the best IT Training and Consultation possible. With a focus on AI, the M365 Stack, and Healthcare, he continues to give back to the community with training, public speaking events, and this blog.

Get E-Mail Updates
I agree to have my personal information transfered to MailChimp ( more information )
Want to know when I post new content? Sign up here to get an email when I do post!
I despise spam. Your email address will not be sold or shared with anyone else.