Prompting Copilot

Overview

Learn how to prompt Copilot.

How to prompt Copilot

A useful prompt gives Copilot enough direction to understand what you want, what information it should consider, what a successful response should include, and which source material it should use.

  • Goal: What do you want Copilot to do?
  • Context: What information or background does Copilot need?
  • Expectations: What should the response include or look like?
  • Source: What information or material should Copilot use?

Specify Your Audience

Tell Copilot who the response is for so it can adjust the tone, level of detail, word choice, and examples for that audience.

Examples 

  • Bad example: Write a blog post about financial planning.
  • Good example: Write a blog post about financial planning from the perspective of a financial advisor speaking to recent college graduates. Use a friendly, approachable tone and focus on practical steps they can take to start saving and investing.

Refine Your Prompt

Few-shot prompting helps Copilot by providing examples of the style or structure you want it to follow.

Example

Prompt

Write a summary of this article. Use active voice and present tense. Here are two examples of summaries we have used before.

“This article discusses the importance of cybersecurity for small businesses, emphasizing best practices for protecting sensitive data.”

“This article explores the benefits of remote work, highlighting key statistics and offering tips for maintaining productivity.”

Now summarize this article in a similar style.

https://www.fakeexample.com/paper.doc

A Few Points to Keep in Mind

  • Review and verify responses you get from Copilot. Copilot is built upon Large Language Models (LLMs), advanced tools designed to predict and generate text. Occasionally, Copilot responses can include incorrect content, due to the vast and diverse nature of LLMs. Evaluate Copilot's responses and cross-reference with trusted sources when needed.
  • Using the same prompt multiple times can result in different responses. LLMs are built upon neural network, which introduces some randomness. Even with the same input prompt, most likely, you will get slightly different results each time.
  • Use Copilot in a respectful, ethical, and lawful manner. Avoid using Copilot for any purpose that might cause harm to yourself or others. See Responsible Use Guidelines for Microsoft Copilot.
Tip: Copilot can support your work, but you remain responsible for reviewing its output and deciding how to use it.

Additional information

Need help?

For more information, see Microsoft Office 365 Support Request or select Request Help from this page.

Print Article