We generate AInote (supplementary information) symbolto improve customer center operations and the quality of customer service.
In the first stage of the verification experiment, the AI will summarize the content of the response, in the second stage, it will analyze customer sentiment and similarities with past response history, and in the third stage, it will generate FAQs. Operators of customer centers for electric power, which is indispensable for daily life, are required to have a wealth of expertise and experience, and we are working to reduce operator labor and improve the quality of customer service.
In addition to our service customer centers, we would like to use AI in the customer centers of 10 other electric power and gas-related companies to which we are entrusted to provide operation services.
*About the generated AI we utilize: For security and privacy protection reasons, OpenAI API, the company that provides ChatGPT, states that it does not use input text for training.

■ Background of Generated AI Demonstration
MHI operates customer centers for its own condominium bulk power receiving service (approximately 2,200 buildings, or 175,000 households) and electric power retailing business (approximately 8,000 contracts). In addition, we are contracted by 10 other electric power retail and bulk electric power receiving service companies and major LP gas-related companies to provide operation and maintenance services, including customer support.
The customer center strives to provide prompt and accurate customer service through the use of response manuals based on specialized knowledge of the energy industry and a follow-up system by leaders, etc. However, through the use of AI that learns from the responses of operators with a wealth of experience and knowledge, the customer center as a whole is expected to further raise the level of customer service quality. Through the use of AI that learns from the responses of experienced and knowledgeable operators, it is expected that the overall customer center will be able to further raise the quality level of customer service and continue to operate at a high level even with changes in staff.

■ Demonstration Phase I
The first step in utilizing generative AI is to summarize the textualized content of the call. Customer center operators have the task of registering the history of customer service after the completion of a customer service call. Using AI summarization is expected to reduce the time required for operators to register the history and standardize the level of detail in the history.

■ Demonstration Phase II
The second step is to analyze the customer's emotions based on the content of the response and its similarity to past response histories. Based on the analysis results, we can gain insight into latent needs that customers do not express in words, and take actions that lead to improved customer satisfaction.

■ Demonstration Phase III
In the third stage, a generative AI, which has learned from past call-handling history, generates FAQs to support operators in answering calls. When an operator types in a query during a call, the generated AI provides an appropriate response script. This reduces the number of cases where operators are followed up by a leader, enabling all operators to respond quickly and accurately.

■Future
Based on the results of the demonstrations at each stage, we will promote the introduction of generated AI into the operation of the customer centers of our services and the customer centers that we are entrusted with. Furthermore, we will continue our efforts to operate high-quality customer centers that can provide a better customer experience while making full use of AI technology.

■Comment from Riku Iyota, Executive Officer and General Manager, Information Systems Department, in charge of development
We are positioning AI as a strategic tool that is being demonstrated. We will use large-scale language models and generative AI to improve operational efficiency and customer experience, as well as explore new business opportunities.