Hi We are building an AI platform and want to integrate LiveKit for real-time voice telephony. Our goal is to support: - AI voice conversations (real-time) - Telephony (inbound/outbound calls) - Integration with ServiceNow for ticketing/workflows - Using ElevenLabs for TTS (and STT/LLM in our pipeline) We’d like guidance on the overall architecture and best practices. Specifically: 1. What is the recommended architecture for: - Telephony (SIP/PSTN) → LiveKit → AI Agent → response (TTS)? 2. Should we use LiveKit Agents for orchestration, or manage our own pipeline (STT → LLM → TTS)? 3. How do we: - Stream audio from telephony into LiveKit? - Send AI-generated audio back into the call? 4. What is the best way to handle real-time events like: - Call start/end - Transcripts - AI responses → so we can push them into ServiceNow? 5. Are there any reference architectures or examples for: - Voice bots / call center automation using LiveKit? Any pointers, diagrams, or sample repos would be super helpful .

Hi Kevin M, ⚠️ This channel will be deleted soon ⚠️ Please use our new community at https://community.livekit.io for technical discussions and to post technical questions. The Slack community is not going away, but we are migrating technical questions to our new forum. Please see full details here.

With Livekit you can connect to inbound calls only in the US atm, you'd need other platform for outbound telephony. It is exactly the platform for orchestration where you can manage STT/LLM/TTS. For real-time events like call start/end, transcripts, LiveKit emits those via callbacks, you can push them to ServiceNow via webhooks. For reference architectures, check the docs/Git - they have voice bot templates you can ask ai in here as well