Skip to main content
Build a voice agent with Gemini Realtime on Stream’s edge network. You’ll need a free Stream account, a Google AI Studio API key, and 18 lines of Python.

Copy this prompt into Claude Code, Cursor, Windsurf, or any coding agent to scaffold your project.

CursorOpen in Cursor

​
1. Create a project

Requires uv and Python 3.12 with CPython. 
mkdir my-agent && cd my-agent
uv init && uv add "vision-agents[getstream,gemini]" python-dotenv

​
2. Add your API keys

Create a .env file in your project root. Vision Agents auto-loads these for each plugin: 
# Stream — getstream.io/try-for-free
STREAM_API_KEY=your_stream_api_key
STREAM_API_SECRET=your_stream_api_secret

# Google AI Studio — aistudio.google.com
GOOGLE_API_KEY=your_google_api_key

​
3. Write the agent

Create main.py: 
from dotenv import load_dotenv

from vision_agents.core import Agent, AgentLauncher, User, Runner
from vision_agents.plugins import getstream, gemini

load_dotenv()


async def create_agent(**kwargs) -> Agent:
    return Agent(
        edge=getstream.Edge(),
        agent_user=User(name="Assistant", id="agent"),
        instructions="You're a helpful voice assistant. Be concise.",
        llm=gemini.Realtime(),
    )


async def join_call(agent: Agent, call_type: str, call_id: str, **kwargs) -> None:
    call = await agent.create_call(call_type, call_id)
    async with agent.join(call):
        await agent.simple_response("Greet the user")
        await agent.finish()


if __name__ == "__main__":
    Runner(AgentLauncher(create_agent=create_agent, join_call=join_call)).cli()

​
4. Run it

uv run main.py run
Swap to OpenAI Realtime in one line: 
from vision_agents.plugins import openai

llm=openai.Realtime()

​
Next Steps

Voice Agents

Custom STT/LLM/TTS pipelines, function calling, provider options

Video Agents

VLMs, YOLO processors, real-time video analysis

Deploy to Production

Docker, Kubernetes, and monitoring

Browse Integrations

25+ AI providers to mix and match