This research project focuses on creating a system architecture of a audio sales bot for small medium businesses using Natural Language Processing and Neural Networks in Deep Learning. The proposed application and its design is suggested to complete 40% tasks of a Human Sales Agent for a target market of United Kingdom. After careful preliminary tests of basic input and output delivery via chat interface and processing through neural network the expected results are able to achieve 86% accuracy in controlled environment with limited dataset. The two core modules, Speech To Text and Text to speech methods has been identified and tested separately. This research also highlights and distinguishes the core features of such application along with additional essential features and features to operate as close as a Human Sales Agent with respect to traits of a Human Sales Agent. A proposition has also been made in regards to applicability of such system in various scenarios and its versatility to handle inquiries and responses in real time.