Cooperating in the Development of Generative AI for Government "iJAMP RAG" ~Promoting Efficiency and Advancement of Government Operations through PoC with Jiji Press and Shift Plus~

Miyakonojo City has cooperated as the implementing body for a Proof of Concept (PoC) in the development of "iJAMP RAG (iJAMP Rag)," a data utilization option for the government-向け news service "iJAMP," jointly developed by Jiji Press Ltd. (Chuo-ku, Tokyo, President and CEO: Katsuhiko Sakai, hereinafter "Jiji Press") and Shift Plus Co., Ltd. (Miyakonojo City, Miyazaki Prefecture, President and CEO: Yuya Nakao, hereinafter "Shift Plus"). This option is scheduled to begin trial provision in July 2026.

[Regarding the Business Partnership]

This city is actively promoting the use of generative AI in government operations. As part of this initiative, we have been verifying the effectiveness of RAG (Retrieval-Augmented Generation) functions that utilize the news data of "Kancho Sokuhou" held by "iJAMP" as reference information on the government-向け generative AI platform "Jichitai AI zevo (zevo)" provided by Shift Plus.

We conducted a PoC from April 2025 to verify the effectiveness of RAG utilization in government operations. Effects such as improved operational efficiency and increased response accuracy have been confirmed in document generation and information organization using "Kancho Sokuhou" data, and we are realizing that generating practical answers based on policy background and local government case studies contributes to improving staff operational efficiency and supporting decision-making.

[Background and Objectives]

In recent years, while the use of generative AI in government operations has been advancing, challenges have become apparent, such as insufficient response to job-specific information with general-purpose generative AI alone, difficulty in effectively utilizing dispersed knowledge within government agencies, and the time required for information gathering, organization, and comparison. "iJAMP RAG," in which this city cooperated, realizes the generation of practical-level outputs based on policy trends and advanced cases by combining it with the news data of "iJAMP Kancho Sokuhou," in addition to a design that assumes the sharing of prompt templates and the accumulation and reuse of local government knowledge.

Furthermore, it is expected to significantly reduce the time required for information gathering, organization, and comparison by suppressing hallucinations (generation of misinformation) through answer generation based on reference data.

[Future Outlook]

This city plans to promote its use for purposes such as creating proposals and business proposals, and drafting answers for assembly sessions. We also have expectations for future implementations such as organizing discussion points in policy deliberations and creating explanatory materials for management, which are planned to be implemented. We will continue to work on improving citizen services and increasing the efficiency of government operations through the utilization of advanced digital technologies.