SoftBank Corp. Launches Development Project in Collaboration with Microsoft Japan to Optimize its Call Center Operations with Generative AI

April 9, 2024

45 1 minute read

In March 2024, SoftBank Corp. (TOKYO: 9434) announced it is aiming to further enhance its call centers for mobile, fixed-line and Internet communication services in Japan by using generative AI. To achieve this goal, it is launching a joint development project with Microsoft Japan. The project, which will start implementation in phases from July 2024, is expected to reduce customer wait times and standardize responses, thereby contributing to greater customer convenience.

SoftBank first started utilizing generative AI across its business in February 2023, and used generative AI at its call centers on a trial basis. SoftBank decided that by further utilizing generative AI to increase automation at its call centers–where routine tasks are numerous and various types of information are referenced for customer support purposes–it could improve the customer experience. By using Microsoft Japan’s “Azure OpenAI Service,” SoftBank is now aiming to build a state-of-the-art call center that maximizes the utilization of generative AI.

SoftBank’s call centers handle a wide range of inquiries for the various communications services it provides, with over 10,000 business processes in existence. Among these processes, the use of generative AI will be focused on tasks that provide guidance in response to customer inquiries, price plan details, and procedures for making changes to contracts. Large language models (LLMs) will provide optimal responses based on data sources.

In the joint development project with Microsoft Japan, SoftBank will aim to provide flexible and accurate customer support by developing an LLM-based autonomous thinking system. This system contrasts with traditional ones that classify the intentions (search the intentions) of customer inquiries to respond with a predetermined procedural order and fixed script. LLMs will query the necessary functions and data sources according to the conversation with the customer.

To enhance response accuracy, prompting will be conducted with large amounts of information, including SoftBank’s service content details and operator response patterns. And by utilizing Microsoft Japan’s “Azure AI Search” to integrate retrieval augmented generation (RAG) with SoftBank’s internal databases, customers are expected to quickly receive optimal answers to their questions.