Module 5 · Project 4: Production Voice-Agent Platform (your own 'mini-Vapi')

OBJECTIVE: turn your LiveKit agent into a production-level platform that serves several businesses: each client gets its own agent settings and phone number, inbound calls and outbound campaigns run reliably at scale, every call is logged, measured, tested and secured, and you know your cost per minute — and learn what it takes to run voice agents in production. WHAT WE COVER (in project order): • Platform design: what is configured per client vs shared • Agent configuration stored in a database and loaded at call time • A backend API for agents, numbers, campaigns and call logs • Campaign engine: queue, concurrency, calling windows, retries • Post-call pipeline: transcript, summary, structured data, webhooks to client CRMs • Dashboard for clients • Automated testing with simulated callers and regression tests • Observability, reliability and provider fall-backs • Deployment and scaling of agent workers • Security and compliance • Cost model and pricing AFTER THIS MODULE YOU CAN: • Design and run a multi-client voice-agent platform • Keep quality, uptime, cost and compliance under control • Price a voice-agent service with confidence TOOLS: LiveKit Agents, FastAPI, PostgreSQL, a job queue (Redis-based), Streamlit or a web dashboard, Docker, Langfuse or similar, cloud hosting.