Three NTT Group Companies Conduct Trial of 'AI Constellation' to Enhance and Optimize Building Management Operations

NEWS RELEASE May 19, 2026 NTT Urban Solutions Inc. NTT Docomo Solutions Inc. NTT Docomo Business Inc. NTT Urban Solutions Inc. (Headquarters: Chiyoda-ku, Tokyo; President: Yasushi Ikeda), NTT Docomo Solutions Inc. (formerly NTT Comware Corporation, Headquarters: Minato-ku, Tokyo; President: Tetsuya Mikajiri), and NTT Docomo Business Inc. (formerly NTT Communications Corporation, Headquarters: Chiyoda-ku, Tokyo; President: Katsushige Kojima) have conducted a trial (hereinafter 'this trial') utilizing 'AI Constellation®*2'. This large-scale AI collaboration technology promotes discussion and coordination among multiple 'Specialized AIs*1' using urban district data to generate optimal solutions from diverse perspectives, aiming to elevate and streamline building operations and management. In this trial, effectiveness was verified across two operational tasks: 1) Examination of event planning proposals, and 2) Optimization of signage content display. The trial confirmed a reduction in planning time, quality improvements resulting from multifaceted perspectives, and behavioral changes following the display of optimized signage content. *1: Refers to individual AI agents possessing specialized information in respective domains to analyze various events within a district. *2: AI Constellation® is a technology currently under research and development at NTT Computer & Data Science Laboratories. AI Constellation® is a registered trademark of NTT Corporation. https://www.rd.ntt/cds/ai-constellation/ ■ Background of the Trial Building operations and management face numerous challenges, including increased workloads due to expanding operational scopes, labor shortages, and the personalization or rendering tacit of operational knowledge. Consequently, there is a strong demand for further enhancement and efficiency. Additionally, these operations involve highly complex elements such as 'cross-sectional utilization of specialized knowledge and diverse data accumulated through daily operations to make situational judgments and optimizations.' Therefore, a system capable of handling such complex operational structures is required. ■ About AI Constellation® AI Constellation® is an AI utilization approach that combines AI agents with various specialties and roles, supporting the entire examination process through generation, evaluation, and supplementation. Beyond answer generation by a single generative AI, it features the ability to support the processes necessary for decision-making—such as generating proposals from diverse viewpoints using multiple AIs, organizing discussion points, conducting comparative analyses, and clarifying the rationale behind judgments. ■ Trial Overview and Results A trial system utilizing AI Constellation® (hereinafter 'this system') and district data collected from the NTT Urban Solutions Group's district operations was constructed. Demonstrations were conducted on the following two themes to verify the effectiveness and reproducibility (potential for deployment to other properties) in practical building operations and management. Theme Name 1) Examination of event planning proposals Implementation Period Event planning period: January 2026 – February 2026 Event execution date/time: February 27, 2026, 18:30–20:00 Implementation Location Urbannet Nagoya Nexta Building (Workers' Lounge exclusively for tenants) Target Operation Event planning aimed at improving customer satisfaction (planning formulation, creation of explanatory materials, and presentations) Trial Procedure - Using this system, three specialized AIs (Event Information AI, Worker Needs AI, and Building Facility Information AI) engaged in discussions. A summarizing AI consolidated their opinions to generate an event planning proposal for office workers, and the event was subsequently held. - Evaluations were conducted on the workload of the staff in charge and the quality of the event planning proposals. - Satisfaction evaluations from the perspective of participants were conducted for the AI-generated event. Verification Viewpoints (1) Reduction in operational time (2) Improvement in planning quality Results of 1) Examination of event planning proposals Utilizing this system enabled the generation of plans that included multifaceted perspectives with minimal omissions. The time required for formulating plans and explanations was reduced by approximately 50% compared to when the system was not used. This indicates an estimated surplus capacity creation of approximately 30 man-days annually for the targeted operations. Furthermore, regarding planning quality, by inputting property information, tacit knowledge in building management, and past event data into the specialized AIs, it became possible to draft plans based on accumulated expertise. This progress represents the successful verbalization and structuring of decision-making processes that were previously personalized and tacit. Positive feedback was received from operational staff, including comments such as, 'We obtained a planning proposal incorporating new perspectives and content,' and 'The process where AIs with different specialties engage in discussion to derive conclusions significantly shortens the time required for planning, and we felt it was highly meaningful.' Theme Name 2) Optimization of signage content display