Direct Response Mode

Route requests to specialized agents who respond directly.

Use mode=TeamMode.route to route requests to the appropriate agent and return the member response directly. The legacy respond_directly=True flag still works, but mode is preferred.

This example creates a language router with three agents:

English Agent - Responds in English
Japanese Agent - Responds in Japanese
Spanish Agent - Responds in Spanish

Create a Python file

1from kern.agent import Agent
2from kern.models.openai import OpenAIResponses
3from kern.team.team import Team
4from kern.team.mode import TeamMode
5
6english_agent = Agent(
7    name="English Agent",
8    role="You only answer in English",
9    model=OpenAIResponses(id="gpt-5.2"),
10)
11japanese_agent = Agent(
12    name="Japanese Agent",
13    role="You only answer in Japanese",
14    model=OpenAIResponses(id="gpt-5.2"),
15)
16spanish_agent = Agent(
17    name="Spanish Agent",
18    role="You only answer in Spanish",
19    model=OpenAIResponses(id="gpt-5.2"),
20)
21
22language_router = Team(
23    name="Language Router",
24    model=OpenAIResponses(id="gpt-5.2"),
25    mode=TeamMode.route,
26    members=[english_agent, japanese_agent, spanish_agent],
27    instructions=[
28        "Route questions to the appropriate language agent.",
29        "If the language is not supported, respond in English.",
30    ],
31    markdown=True,
32    show_members_responses=True,
33)
34
35# English
36language_router.print_response("How are you?", stream=True)
37
38# Japanese
39language_router.print_response("お元気ですか?", stream=True)
40
41# Spanish
42language_router.print_response("¿Cómo estás?", stream=True)

Set up your virtual environment

1uv venv --python 3.12
2source .venv/bin/activate

1uv venv --python 3.12
2.venv\Scripts\activate

Install dependencies

1uv pip install -U kern-ai openai

Export your OpenAI API key

1export OPENAI_API_KEY="your_openai_api_key_here"

1$Env:OPENAI_API_KEY="your_openai_api_key_here"

Run Team

1python respond_directly.py