Introduction to Agents
AI agents in HALO are autonomous programs designed to understand context, make decisions, and utilize tools to accomplish tasks efficiently. This documentation provides an overview of creating, managing, and optimizing these agents within the HALO platform.
Accessing Your Agents
You can find your agents by clicking the 'Agent' button located in the left-side menu. This will take you to the Agents Overview section, where you can access information about all your agents.

The Root Agent
In the Agents Overview section, you'll see a catalog of all your agents, including the designated "root agent," marked by a blue toggle. The root agent acts as the central coordinator, managing the distribution of tasks among other agents. It is designed to streamline operations by overseeing tasks for each audience segment. You can have one root agent per audience, such as one for internal and one for external use cases. If no root agent is set, HALO defaults to using the Knowledge Agent, which accesses information from the Knowledge sources you've added to HALO.
Overview Features
Within the Agents Overview, you can see:
Agent Details: Which agents have been created and their respective audiences.
Root Designation: Whether an agent is designated as the root agent.
Tools and Handovers: The number of tools each agent can use and how many handovers they can perform.
By clicking on the three dots next to each agent, additional options become available. Here, you can edit the agent, duplicate it, or copy the agent to another profile within your account. You also have the option to delete an agent.

Top Panel Insights
In the top panel of your Agents Overview, you can review critical performance metrics:
Total Agent Triggers: Displays how often agents are activated.
Average Response Time: Shows the average time it takes for an agent to respond.
Completion Rate: Indicates the percentage of tasks successfully completed.
By using these insights and tools, you can effectively manage and optimize your AI agents within HALO, ensuring they perform at peak efficiency for various use cases.