If you want, we can take this off here I can walk you through the approach or even help you set it up properly so you get consistent, replayable tests without all the manual calling. Have you already built your voice stack on something like Twilio/Deepgram, or is it fully custom?