In my current role as a Data Scientist at DataLytics, we worked on a project in the domain of customer service automation for a major retail company. The challenge was to construct a system capable of understanding customer inquiries and generating appropriate responses.
We tackled this using a two-pronged approach: BERT and GPT. We used BERT for extracting the semantic meaning of the customer's inquiry, considering both preceding and following text. For generating eloquent responses, we utilized GPT models. The data used for this project majorly constituted customer interactions from the past two years which were cleaned and processed for the training purpose.
The deployment of this AI-driven customer service solution drastically improved the company's service efficiency. It managed to answer over 70% of customer inquiries without human intervention, resulting in faster response times and higher customer satisfaction rates.