Edit Copilot Prompt

Copilot Name

This is where you can edit the name of your copilot.

Copilot Description

This description is for you so that you may specify the use of your copilot for yourself and your team.

Objective

This tells the copilot what its primary objective is.

Audience

This tells the copilot who will primarily be using the copilot.

Personality

This tells the copilot how to act and carry itself in conversation. Some examples: friendly and helpful, motivational, pirate (ARG!), silly, etc.

Assistant Name

This is the name that appears in the chat screen for the copilot.

Greeting

This is how the copilot will initially greet the user.

Inputs

This is what you would expect a user to input into the chat window. Some examples: General questions, error codes, concerns, feedback, etc.

Outputs

This is what you want the copilot to output to the user. Examples: step by step instructions, general guidance, product support, etc.

Additional Rules

The "additional rules" section allows you to customize your copilot's behavior to better meet your specific needs by providing tailored responses, enhanced context, and improved efficiency. Here, you can define clear and specific rules, ensure consistent outputs, save time on repetitive tasks, and refine prompts through experimentation. This section also supports particular use cases, making the copilot a more valuable and effective tool for various scenarios. There is more information regarding this section and best practices here.

Advanced Settings

• Choose an LLM:

  Copilot Builder stands out because it offers the flexibility to choose from multiple Large Language Models (LLMs). This is significant because different LLMs excel in various tasks and contexts. For instance, some models may be better suited for conversational interactions, while others might excel in generating creative content or processing technical information. By selecting the right LLM for your specific needs, you can enhance your copilot's performance, ensuring it delivers more accurate and relevant responses. There is more information regarding our available LLMs and their strengths/weaknesses here.

• Creativity:

  This is how consistent you want the copilot responses to be. If 100 users ask the same question, would the answers be the same (low creativity) or very different (high creativity).

• Log Chat:

  We recommend you log your chat history so that you may go back and view how your copilot has responded in the past.

• Show Title of Copilot in Chat Window:

  This is for hosted and framed deployed copilots. If you want the title of your copilot at the top of the chat window, toggle this on.

• Get real-time data:

  Toggle this on to allow your copilot to do a web search to answer user’s questions.

  WARNING: This can drastically affect the outputs of your copilot and make it hard to control the accuracy and consistency of your copilot's responses. Only use this feature if you are sure you need it, like in the case of a copilot that needs access to the current weather of a specific zip code.

• Allow image generation requests:

  Allows your copilot to respond to users with images.