To move from local testing with

python agent.py dev

to a production environment on LiveKit Cloud, follow these steps: 1. Set up your production environment variables: ◦ Configure your LiveKit Cloud project credentials:

LIVEKIT_URL=<your livekit server url>
LIVEKIT_API_KEY=<your api key>
LIVEKIT_API_SECRET=<your api secret>

◦ Add any other required API keys for your agent (like Deepgram, OpenAI, etc.). 1. Use production mode: ◦ Switch from development mode to production mode by running:

python agent.py start

◦ This runs your agent with production-ready optimizations. 1. Consider deployment requirements: ◦ Memory: 8GB of memory for every 25 concurrent sessions (recommended). ◦ CPU: 4 cores for every 25 concurrent sessions (recommended). ◦ Storage: 10GB of ephemeral storage is typically sufficient. 1. Use separate environments: ◦ Create separate LiveKit Cloud projects for development, staging, and production. ◦ Each environment will have its own unique URL, API key, and secret. For more detailed deployment examples, you can refer to the agent-deployment repository, which contains example Dockerfiles and configuration files for various deployment platforms. Sources: Deploying to production | LiveKit Docs | Tree View: | dse-workbench