Conduct dry runs

Run test calls internally till you are fully satisfied that the voice AI agents speak exactly the way you want them to and can handle pretty much anything