​
Overview

LLMs excel at characterization. That’s why we use agents and continually refine their system prompts. This also creates new opportunities for agent teams: since our tasks are split among specialized agents, we must manage the process ourselves—assigning each task to the right agent and enabling effective collaboration. In Upsonic, we have a Team class for this purpose. It automatically performs these operations.

​
Creating an Team, Agents and Tasks

The Team class is a backend component that incorporates multiple automated systems. It facilitates matching and executing the right tasks with the appropriate agents and enables active collaboration. 
from upsonic import Agent, Task, Team

# My Agents
travel_agent = Agent(name="My Travel Agent")
history_agent = Agent(name="My History Agent")

# My Tasks
task = Task("Generate a plan to visit cities in Canada")
task2 = Task("Write historical information about the cities")

# My team 
team = Team(
	agents=[travel_agent, history_agent], # Adding Agents
	tasks=[task, task2] # Adding Tasks
)

​
Getting Final Answer

After instantiating the Team class, you can easily run tasks on it—just as you would with the Agent and Direct classes—by using the do and print_do functions. 
from upsonic import Agent, Task, Team

# My Agents
travel_agent = Agent(name="My Travel Agent")
history_agent = Agent(name="My History Agent")

# My Tasks
task = Task("Generate a plan to visit cities in Canada")
task2 = Task("Write historical information about the cities")

# My team 
team = Team(
	agents=[travel_agent, history_agent], # Adding Agents
	tasks=[task, task2] # Adding Tasks
)



# Run the task and get the answer
result = team.complete() 

print("Summerized result")
print(result)

​
Adding Response Format to The Team

Teams are well but sometimes we need some specific answers from them. To do that you can use response format params and a BaseModel. 
from upsonic import Agent, Task, Team
from pydantic import BaseModel

# My Agents
travel_agent = Agent(name="My Travel Agent")
history_agent = Agent(name="My History Agent")

# My Tasks
task = Task("Generate a plan to visit cities in Canada")
task2 = Task("Write historical information about the cities")

# Response Format
class City(BaseModel):
	name: str
	history: str
class TravelGuide(BaseModel):
	cities: list[City]

# My team 
team = Team(
	agents=[travel_agent, history_agent], # Adding Agents
	tasks=[task, task2], # Adding Tasks,
	response_format=TravelGuide # Adding Response Format
)



# Run the task and get the answer
result = team.complete() 

print("Summerized result")
print(result)

​
Automatic Task Assignment

By default, the Team class uses automatic task assignment. This means that the system intelligently analyzes each task and selects the most appropriate agent from your team to handle it. You don’t need to do anything special to enable this feature—it’s the default behavior that works automatically when you create tasks without specifying an agent. The automatic assignment process considers the task description, context, and available agents to make the best match, ensuring optimal task distribution across your team.

​
Manual Task Assignment

Sometimes you want to have direct control over which agent handles which task. You can achieve this by specifying the agent parameter when creating your tasks. When a task has a predefined agent, the team will skip the automatic assignment process and use your specified agent directly. 
from upsonic import Agent, Task, Team

# My Agents
travel_agent = Agent(name="My Travel Agent")
history_agent = Agent(name="My History Agent")

# My Tasks with manual agent assignment
task = Task("Generate a plan to visit cities in Canada", agent=travel_agent)
task2 = Task("Write historical information about the cities", agent=history_agent)

# My team 
team = Team(
	agents=[travel_agent, history_agent], # Adding Agents
	tasks=[task, task2] # Adding Tasks with predefined agents
)

# Run the task and get the answer
result = team.complete() 

print("Summerized result")
print(result)