System Message
The system message is a crucial component of an AI Agent, serving as the hidden instruction set that defines the agent’s behavior.
The system message is arguably the most important part of an AI Agent, as it forms the foundation of the agent’s functionality and decision-making process.
The system message:
- Defines the agent’s role and behavior
- Specifies when and how to use tools
- Instructs on invoking AI Teammates
- Is included in every request sent to the agent
Key Features
Behavior Definition
The system message sets the tone, personality, and operational boundaries for the AI Agent. It’s where you define how the agent should interact, what tasks it should perform, and any specific guidelines it should follow.
Tool and Teammate Integration
When tools or AI Teammates are added to an agent, their IDs and descriptions are automatically inserted into the system message. This enables the agent to understand and utilize these resources effectively.
For more details on tools and AI Teammates, refer to their respective documentation pages.
Context Integration
The system message automatically incorporates relevant context from the knowledge base when documents are uploaded and selected for the AI Agent to use.
Retrieved context appears at the end of the system message in the following format:
[System Message]
Used Context and Memories: [Context Chunks]
[Chat History]
[Your current request]
Customization
You can modify the agent’s system message in the builder interface. This allows for fine-tuning of the agent’s behavior and capabilities to suit specific use cases or requirements.
Changes to the system message can significantly impact the agent’s performance and behavior. Always test thoroughly after making modifications.
Best Practices
When crafting or modifying a system message:
- Be clear and specific about the agent’s role and limitations
- Provide guidelines for tool usage and teammate interactions
- Include any necessary ethical constraints or operational boundaries
Consider creating templates for common agent roles to streamline the process of setting up new agents.