@calm-oyster-84545 you can try out cekura.ai which is a whole product specifically for voice agent testing. You can test any kind of flows pretty easily and it comes with a native livekit integration. Feel free to reach out in case of any doubts : )

Thanks — great pointer to cekura.ai, and thanks for sharing that with the team. I’ll summarize how this fits with LiveKit and offer next steps if you want help wiring it up. Acknowledgement • You want to simulate conversation between two AI agents by routing Agent A’s audio into Agent B. cekura.ai is a helpful product for voice-agent testing and already offers a native LiveKit integration, so it can speed up setup and testing. How cekura.ai relates to LiveKit testing • cekura.ai can handle flow definition, test orchestration, and reporting while using LiveKit for media transport and room management. • With a LiveKit-native integration, cekura.ai can publish synthesized audio into a room and subscribe to tracks, letting you run automated agent-to-agent scenarios without building the orchestration layer yourself. If you’d rather use LiveKit directly (quick recap) • Put both agents in the same room. • Configure the listening agent to accept audio from other agents by setting

RoomInputOptions

with a participant filter that includes agents (for example,

participant_filter=lambda p: True

). • Use VAD → STT on the listener to transcribe the speaker agent’s TTS audio, process it, generate a reply, synthesize it with TTS, and publish it back to the room. • Repeat to create full conversational turns. Want help with next steps? • I can provide a concrete code example for connecting two agents in LiveKit (Python or TypeScript). • I can outline an evaluation script for Agent B (scoring, latency measurement, or automated checks). • Or I can walk through how to integrate cekura.ai with your LiveKit setup if you want to use that tool. Which would you like me to do next?