Agents

Agents are the core concept in Archestra Platform. Each agent represents a distinct AI application or workflow that you want to secure and monitor. Think of an agent as a logical grouping for:

• Interaction history - All LLM requests and responses
• Tool configurations - Which tools the agent has access to
• Security policies - Tool invocation and trusted data policies specific to this agent
• Analytics - Performance metrics and security events

Why Use Agents?

Using agents provides several benefits:

1. Isolation - Keep different AI applications separate with their own policies
2. Monitoring - Track interactions and security events per application
3. Flexibility - Apply different security rules to different use cases

Agent Labels

Agent labels are a powerful feature that can be used to organize and categorize your agents.

Beyond organization, labels also play a crucial role in observability. Read more about how to use labels for observability here.