Our mission at Reply.ai is to remove frustration and save people's time when interacting with a business. One of the ways that we can impact customer experience is by making customer support agents happier and more efficient. We would love to share with the ML community how we used Deep Learning to make a real-time Assistant to increase the efficiency of Customer Service teams. Support agents do a lot of repetitive work when answering tickets and chats in customer service platforms like Zendesk, Kustomer and Khoros. Instead of writing unique, personalized responses, they have to classify tickets, add specific tags, route tickets to other teams, and select templates, among other responsibilities.