Create Custom AI Agents of Type Supervisor from Scratch

1. Go to Navigator > Tools > AI Agent Studio.

2. From the Tools tab, add the required tools.

   For example, to create an HR benefits administrator agent that can answer questions related to medical, vision, dental, retirement, and stock plans, these are some of the tools needed.

   • Document tools, with the organization's health and financial benefits documents.
     Note: Before you add any document tools to an agent, make sure you’ve done these things:

     • Set the status of the document tool as Ready to Publish.

     • Run the Process Agent Documents scheduled process from the Scheduled Processes work area.

     • After the scheduled process completes, set the status of the document tool as Published.

   • Business object tools, to fetch the employee enrollments data.
   • Calculator tool, to check balance amounts and percentages.
3. From the Topics tab, add the required topics. Make sure you include instructions about these key areas.

   • What the topic is about
   • Tools to use with the topic
   • Examples of possible questions
   • Any guidelines and guardrails
4. From the Agents tab, add an agent.

   • In Maximum Interactions field, specify the number of times the agent can interact with the topics and tools assigned to it.

   • Describe the persona and role of the agent, including the tone to use.

   • Add the prompt for the agent, and select any variables from Insert Expression.

5. Assign the required tools and topics to the agent, and create it.

6. From the Agent Teams tab, add an agent team and provide details for the new agent team.

   • Details Tab

     Field	Description
     Family	Select the family to which this agent team belongs.
     Product	Select the product within the family to which this agent team belongs.
     Type	Select Supervisor to create an agent team where one agent acts as supervisor and manages other agents and artifacts.
     Maximum Interactions	Indicate the number of times an agent within this agent team can interact with the topics and tools assigned to it.

   • LLM Tab

     Field	Description
     Provider	Select the LLM for your agent team.

   • Security Tab

     Field	Description
     Add	Select the roles which will have access to this agent team.

   • Questions Tab

     Field	Description
     Starter Questions	Enter initial questions for the agent team.
     Follow-up Questions	Enable this to indicate that the agent team can ask follow-up questions based on the user's conversation history.
     Prompt	Enter the prompt to be used for the follow-up questions. For example: using {chatHistory} generate 3 follow up questions in json format. JSON Schema format mentioned below. Remove the ```json markdown from the output. Here's the JSON Schema format the output should adhere to: [{"question": "<put first question generated here>"}, {"question": "<put second question generated here>"}, {"question": "<put third question generated here>"}]
     Insert Expression	Add additional variables to the prompt. For example, to add the current system date to your prompt, select the Current Date Time option.

   • Output Tab

     Field	Description
     Specification Mode	Select this mode to directly modify the JSON schema for the output.
     Simple Mode	Select this mode to define the output values and types. The corresponding JSON schema will be generated automatically and displayed in the specification mode for any further changes.

7. Select Create.
8. Select and click New Supervisor Agent.
9. Enter details for the supervisor agent and create it. This supervisor agent is specific only to the agent team in which it's created, and can't be reused.
10. Select and add any existing agents as worker agents, or create worker agents.

    When you create worker agents from within the agent team, those agents are available only for this agent team.

11. Add any needed artifacts such as tools and topics.
12. If needed, use to test the agent team. For any required fine-tuning, you can edit the agent team using .
13. Publish your agent team.

Users can view the published agents from the AI Agents page. To open this page, add agent-explore to the end of the URL for AI Agent Studio. For example, https://example.com/myApp/redwood/human-resources/ai-studio/agent-explore.