Yeah, if it’s still steamrolling even with split assistant messages, then the issue is how Vapi batches the outputs it treats them as one continuous turn unless the flow is structured a certain way. I’ve had to tweak the flow/node structure itself to force the TTS engine to stop between messages, not just split the text. If you want, we can jump into a quick private chat and I can look at how your flow is set up sometimes it’s just one small configuration causing the whole thing.