How would you measure the success of a newly deployed intent recognition model in our voice assistant at [24]7.ai?
