@cool-waitress-17060 you should try cekura.ai for this use case. It is specifically built to test multi turn conversations via voice or chat. It also comes with a free trial. Lmk if any doubts. PS: I am the founder of cekura 🙂

Thanks — I see your recommendation for cekura.ai and that you’re the founder. I appreciate the tip. Quick note on that: our support here focuses on LiveKit/Freya behavior and how to test or run multi‑turn flows with LiveKit’s eval and runtime APIs. I can’t evaluate or endorse third‑party tools, but I can help you with patterns and examples for LiveKit specifically. If you’d like help with LiveKit, tell me which of these you want to focus on and I’ll provide a concrete pattern or code sketch: • A test flow using

session.run()

calls and

expect

assertions for an indeterministic scenario (example utterances, tool responses, and handoffs). • A production pattern using real‑time event listeners (e.g.,

on("user_input_transcribed")

) so the agent handles continuous/indeterminate input. • How to seed or manipulate

session.chat_ctx

between runs to simulate complex histories. Pick one and share a short example of the user utterances, tools involved, and where the handoff happens, and I’ll draft the exact steps and checks you can use.