This demo showcases how LiveKit Agents can power a lightweight call center workflow: a dispatcher dashboard lists every active survey room, while a voice agent interviews callers and pushes their answers back into LiveKit metadata in real time.

Each folder includes a README with setup tips and architecture details.

1. Prep LiveKit credentials

   • Create (or reuse) a LiveKit Cloud project.
   • Copy the LIVEKIT_URL, LIVEKIT_API_KEY, and LIVEKIT_API_SECRET into both the agent and frontend .env.local files.

2. Run the survey worker

   cd call-queue/survey-agent
   uv sync
   cp .env.example .env.local
   uv run python src/agent.py download-files
   uv run python src/agent.py dev

   The agent registers as devday-survey-agent by default; keep this name aligned with the frontend config.

3. Launch the dashboard

   cd ../frontend
   pnpm install
   cp .env.example .env.local  # reuse the same LIVEKIT_* values
   pnpm dev

   Visit http://localhost:3000 to load the queue. When new rooms appear, click the headset icon to join the call and continue the survey.

1. The survey agent greets callers, follows the script in SURVEY_QUESTIONS, and calls record_survey_response after every answer. That tool updates room metadata with totals and timestamps.
2. The frontend polls /api/rooms, which wraps LiveKit’s Room Service API and JSON-decodes the metadata into survey.total / survey.answered.
3. Queue cards surface caller info, SIP participants, duration, and progress. If no rooms exist, the UI invites dispatchers to place outbound calls instead.
4. When a dispatcher starts a call, the frontend POSTs to /api/connection-details, embeds agentName="devday-survey-agent" in the token, and hands the room to the LiveKit Agent session view with chat + mic controls.

• Update SURVEY_QUESTIONS or load copy from an external CMS to support different campaigns!