We realized that by adjusting training models we could incorporate autonomous recognition of not only images but intent and behavior into our application suite.