Transcript is showing 1 turn per agent word. Feels broken

Linked the image, call id is 019b0ac1-4bc9-7225-b0a2-63d46f4d7e4e. But the agent transcript is 1 message for the whole sentence, then 1 message per word. We're unable to filter out the single word duplicate messages, which makes us unable to show the whole conversation in our UI
Screenshot_2025-12-10_at_7.28.51_PM.png
Was this page helpful?