Building Teams

Learn how to build Teams with Agno.

To build effective teams, start simple — just a model, members, and instructions. Once that works, layer in more functionality as needed.

Here’s the simplest possible team with specialized agents:

news_weather_team.py

from agno.team import Team
from agno.agent import Agent
from agno.models.openai import OpenAIChat
from agno.tools.duckduckgo import DuckDuckGoTools

# Create specialized agents
news_agent = Agent(
    id="news-agent",
    name="News Agent", 
    role="Get the latest news and provide summaries",
    tools=[DuckDuckGoTools()]
)

weather_agent = Agent(
    id="weather-agent",
    name="Weather Agent", 
    role="Get weather information and forecasts",
    tools=[DuckDuckGoTools()]
)

# Create the team
team = Team(
    name="News and Weather Team",
    members=[news_agent, weather_agent],
    model=OpenAIChat(id="gpt-4o"),
    instructions="Coordinate with team members to provide comprehensive information. Delegate tasks based on the user's request."
)

team.print_response("What's the latest news and weather in Tokyo?", stream=True)

It is recommended to specify the id, name and the role fields of each team member, for better identification by the team leader. The id is used to identify the team member in the team and in the team leader’s context.

Team members inherit their model from their parent team if not specified. Members with an explicitly assigned model retain their own. In nested team structures, inheritance always happens from the direct parent. Teams without a defined model default to OpenAI gpt-4o.

The reasoning_model, parser_model, and output_model must be explicitly defined for each team or team member.

See the model inheritance example.

Run your Team

When running your team, use the Team.print_response() method to print the response in the terminal.

For example:

team.print_response("What's the latest news and weather in Tokyo?")

This is only for development purposes and not recommended for production use. In production, use the Team.run() or Team.arun() methods. For example:

from typing import Iterator
from agno.team import Team
from agno.agent import Agent
from agno.run.team import TeamRunOutputEvent
from agno.models.openai import OpenAIChat
from agno.utils.pprint import pprint_run_response

news_agent = Agent(name="News Agent", role="Get the latest news")
weather_agent = Agent(name="Weather Agent", role="Get the weather for the next 7 days")

team = Team(
    name="News and Weather Team", 
    members=[news_agent, weather_agent],
    model=OpenAIChat(id="gpt-4o")
)

# Run team and return the response as a variable
response = team.run("What is the weather in Tokyo?")
# Print the response
print(response.content)

################ STREAM RESPONSE #################
stream: Iterator[TeamRunOutputEvent] = team.run("What is the weather in Tokyo?", stream=True)
for chunk in stream:
    if chunk.event == "TeamRunContent":
        print(chunk.content)

# ################ STREAM AND PRETTY PRINT #################
stream: Iterator[TeamRunOutputEvent] = team.run("What is the weather in Tokyo?", stream=True)
pprint_run_response(stream, markdown=True)

Modify what is show on the terminal

When using print_response, only the team tool calls (typically all of the delegation to members) are printed. If you want to print the responses from the members, you can use the show_members_responses parameter.

team.print_response("What is the weather in Tokyo?", show_members_responses=True)

Next Steps

Next, continue building your team by adding functionality as needed. Common questions:

• How do I run my team? -> See the running teams documentation.
• How do I manage sessions? -> See the team sessions documentation.
• How do I manage input and capture output? -> See the input and output documentation.
• How do I give the team context? -> See the context engineering documentation.
• How do I add knowledge? -> See the knowledge documentation.
• How do I add guardrails? -> See the guardrails documentation.
• How do I cache responses during development? -> See the response caching documentation.

Developer Resources

• View the Team reference
• View Team Cookbook