Goal-oriented conversational systems are becoming ubiquitous in daily life for tasks ranging from personal assistants to customer support systems. For these systems to engage users and achieve their goals, they need to 1) not just provide informative replies and guide users through the tasks but also 2) interact with them in a natural manner.  This presentation will cover how Uber applies and studies conversational AI techniques to address these two research questions and provide delightful user experiences especially regarding customer care services.  To answer the first question, I will describe COTA, a productionized system we built to improve the speed and reliability of customer support for end users through automated ticket classification and answer selection for support representatives.  After that, in order to build conversational agents which can generate more natural responses, we applied statistical modeling techniques to understand whether and how social language is related to user engagement and task completion in goal-oriented conversations, which motivated us to propose a conversational agent model capable of incoporating social language into agent responses.