How to do the second order or more using Crew AI agents?

[PremKumar135]

I tried to do the Hierarchical crew like the director agent at the top level, manager agents at the middle , employee agents at the second level and so on. Right now, i can only do till the middle but cant do second level and so on. Does anybody know how to do that?

[dosubot[bot]]

To implement a hierarchical structure with Crew AI agents, including multiple levels such as a director agent at the top, manager agents in the middle, and employee agents at the second level and beyond, you can use the HierarchicalCrewComponent class from the crewai module. This class allows you to define a group of agents and specify how they should collaborate and perform tasks hierarchically.

Here's a brief outline of how you can set this up:

1. Define Agents and Tasks: Use components like CrewAIAgentComponent and HierarchicalTaskComponent to specify the agents and tasks. You can list multiple agents and tasks to form the hierarchical structure.

2. Manager LLM and Manager Agent: Use OpenAIModelComponent to define language models for both the manager and other agents. Set up a manager_agent using CrewAIAgentComponent to handle the management level logic.

3. Build the Crew: Use the build_crew method to create a Crew instance with the specified agents, tasks, and hierarchical process. You can set up prompts for roles, goals, and backstories to define the behavior and responsibilities of each agent.

Here's a code snippet to illustrate the setup:

from langflow.components.crewai.crewai import CrewAIAgentComponent
from langflow.components.crewai.hierarchical_crew import HierarchicalCrewComponent
from langflow.components.crewai.hierarchical_task import HierarchicalTaskComponent
from langflow.components.inputs import ChatInput
from langflow.components.models import OpenAIModelComponent
from langflow.components.prompts import PromptComponent

# Define language models
llm = OpenAIModelComponent(model_name="gpt-4o-mini")
manager_llm = OpenAIModelComponent(model_name="gpt-4o")

# Define prompts for role, goal, and backstory
role_prompt = PromptComponent(_display_name="Role Prompt")
role_prompt.set(template="Define a Role that could execute or answer well the user's query.")

goal_prompt = PromptComponent(_display_name="Goal Prompt")
goal_prompt.set(template="Define the Goal of this Role, given the User's Query.")

backstory_prompt = PromptComponent(_display_name="Backstory Prompt")
backstory_prompt.set(template="Define a Backstory of this Role and Goal, given the User's Query.")

# Define dynamic agent
dynamic_agent = CrewAIAgentComponent()
dynamic_agent.set(
    llm=llm.build_model,
    role=role_prompt.build_prompt,
    goal=goal_prompt.build_prompt,
    backstory=backstory_prompt.build_prompt,
)

# Define manager agent
manager_agent = CrewAIAgentComponent()
manager_agent.set(
    llm=manager_llm.build_model,
    role="Manager",
    goal="You can answer general questions from the User and may call others for help if needed.",
    backstory="You are polite and helpful. You've always been a beacon of politeness.",
)

# Define task
task = HierarchicalTaskComponent()
task.set(
    task_description="Respond to the user's query with as much information as possible.",
    expected_output="Succinct response that answers the User's query.",
)

# Build hierarchical crew
crew_component = HierarchicalCrewComponent()
crew_component.set(
    tasks=task.build_task,
    agents=[dynamic_agent.build_output],
    manager_agent=manager_agent.build_output
)

# Example usage
crew = crew_component.build_crew()

This setup will allow you to create a hierarchical structure where the director agent can oversee manager agents, who in turn manage employee agents. Adjust the agents and tasks inputs to fit your specific hierarchy and task requirements ^[1][2][3].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Other}