AI Case Study

"The product was beta-launched during last year’s Alibaba 11.11 Global Shopping Festival to help nine brands, including Apple, Xiaomi, Huawei Honor, Vivo, Semir and Nike, to serve some one million customers.

Through AI, the chatbot can “learn” to improve its responses by using algorithms to interpret millions of customer-service interactions. The bot is capable of understanding context and can converse naturally, unlike customer-service chatbots that are programmed to provide answers only in response to specific questions. In addition to text messages, Dian Xiaomi is also able to respond to questions with images, such as product photos and emojis, to add depth to its answers. It can even make customer-specific recommendations for shoe and clothing sizes and answer technical questions about consumer electronics, such as the data-storage capacity of particular smartphone models.

Merchants can customize Dian Xiaomi by building on the bot’s already extensive product and customer knowledge base, adding information on their own stores, products and promotions. When Dian Xiaomi has difficulty understanding certain types of questions, a system called AI Boost collects and categorizes these questions so that a human “coach” can intervene by either directing the bot to relevant answers in its knowledge base or by adding appropriate answers to the base. In this way, Dian Xiaomi becomes continuously better at understanding customers’ intents and what they are looking for. Merchants can adjust the workload allocated between the chatbot and customer-service staff according the time of day to optimize efficiency. Questions that cannot be settled by Dian Xiaomi are automatically routed to human staff."