Regular customer assistance chatbots are generally based on dialogues delivered by a human. It faces symbolic issues usually related to data scaling and privacy of one’s information. In this paper, we present AgentG, an intelligent chatbot used for customer assistance. It is built using the deep neural network architecture. It clouts huge-scale and free publicly accessible e-commerce data. Different from the existing counterparts, AgentG takes a great data advantage from in-pages that contain product descriptions along with user-generated data content from these online e-commerce websites. It results in more efficient from a practical point of view as well as cost-effective while answering questions that are repetitive, which helps in providing a freedom to people who work as customer service in order to answer questions with highly accurate answers. We have demonstrated how AgentG acts as an additional extension to the actual stream Web browsers and how it is useful to users in having a better experience who are doing online shopping. © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021