There are 3 parts in the instructions screen: Identity, Tasks and Guardrails.

1 - Identity: Defines who the AI agent is, including its role, tone, and personality.

To set up an AI agent with the right personality and communication style, clearly define its identity by specifying its role, tone, and approach to interactions. Defines who the AI agent is, including its role, tone, and personality.

Example: “You are a professional customer support agent for a tech company, responding in a friendly and concise manner.”

2 - Tasks: Specifies what the AI agent is designed to do, such as answering customer inquiries, booking appointments, or collecting user feedback.

Outline the specific functions the AI agent will perform to ensure it delivers relevant and useful responses to users. Specifies what the AI agent is designed to do, such as answering customer inquiries, booking appointments, or collecting user feedback.

Example: “Assist users with troubleshooting issues, provide product recommendations, and schedule support calls.”

3 - Guardrails: Defines the boundaries of the AI agent’s behavior to ensure it stays on-topic and avoids sensitive or restricted subjects.

Establish clear boundaries to keep the AI agent focused, ensuring it does not provide misinformation or engage in topics beyond its scope. Defines the boundaries of the AI agent’s behavior to ensure it stays on-topic and avoids sensitive or restricted subjects.

Example: “Do not provide medical, legal, or financial advice. Redirect users to human support if a question falls outside predefined topics.”

You can generate this prompt automatically using our prompt generator, available on the top right of the screen.

Was this page helpful?