Today, customer service is unscalable, unautomated and costly for companies. Support agents have to answer the same questions and processes every day. This is not only bad for companies but also bad for their clients since they often have to wait minutes or even hours in a hotline or have to wait days for an email reply. For this reason, Customer Service is one of the first departments to be outsourced. We believe this presents a missed opportunity as it is a direct link to the customers and their concerns.
The customer demand for fast service on more and more channels is steadily increasing, however, it is extremely costly for companies to meet this demand. This challenge is only compounded by the fact that the distribution of requests is highly-volatile, so that support capacity is either not permanently used or processing times are unsatisfactory for customers at peaks.
We bring practical applications of Deep Learning and AI to customer service and help companies achieve greater customer service efficiency. The system analyses inbound messages, sends them to the right department, and provides agents with accurate response suggestions. This reduces average handling time up to 80% and automates more and more responses and processes. The e-bot7 system is based on complex NLP algorithms (Convolutional Neural Networks, LSTMs) trained on historical customer service scripts. The NLP is constantly optimised during operation so that more and more requests can be processed more efficiently.