Skillup NeXt Builds Obayashi Corporation's "Concierge-Type AI Agent," Aiming for Annual Workload Reduction of Approximately 19,000 Hours Across the Entire Design Department

Skillup NeXt Inc. (Head Office: Chiyoda-ku, Tokyo; President: Shinichi Tahara), a holder of Microsoft's highest partner qualification, the "Microsoft Copilot" Specialization, has developed a Minimum Viable Product (MVP) for an "AI Concierge Agent for BIM" for Obayashi Corporation (Head Office: Minato-ku, Tokyo; President: Toshiyoshi Sato) using Microsoft Copilot Studio to automate inquiries related to the BIM software "Revit." The deployment to approximately 300 Revit users is expected to reduce concierge response time by about 1,000 hours annually and employee self-resolution time by 18,000 hours annually, resulting in a total reduction of approximately 19,000 work hours per year and a decrease of 200 rework cases in subsequent project stages.

Interview Article ▶ https://www.skillupai.com/private-training/success_stories/obayashi3/

Background: Strain of Inquiry Support and Psychological Hurdles for Novice Users

Obayashi Corporation's design department has been handling inquiries related to the BIM software "Revit" with a team of three specialized concierges as the software gained wider adoption. However, with approximately 40 inquiries per month, each requiring an average of 180 minutes for communication and meetings to identify the cause, the strain of 120 hours of support work per month became a significant challenge. Furthermore, response history was only recorded in a simplified format in Excel, lacking images and detailed resolution processes, making it difficult to systematically reuse past knowledge.

On the other hand, for users asking questions, especially new employees and those with limited BIM experience, asking questions on public channels presented a high psychological barrier. It was found that these users spent an average of 5 hours per month on self-directed searches and trial-and-error to resolve doubts. Therefore, there was an urgent need to build a practical AI agent that could simultaneously address the "strain on concierge operations" and the "high barrier to asking questions for users," thereby accelerating BIM adoption.

Support and Construction Details: MVP Development Leveraging Existing Microsoft 365 Infrastructure

To improve the efficiency and automate BIM inquiry support at Obayashi Corporation, "Microsoft Copilot" was selected, considering its integration with the existing Microsoft environment. By minimizing processes such as new infrastructure construction and initial verification procedures, development began immediately, leading to the creation of the concierge-type AI agent in 1.5 months.

Ultra-Short-Term Speed Development: Completed from requirements definition to construction in an ultra-short period of 1.5 months. With Skillup NeXt's support, an initial AI agent with an approximate 80% response accuracy was realized using Copilot Studio.

Practical Data Verification Operations and Flow Definition: Designed an operational flow for inquiry history and evaluation storage using SharePoint. Created benchmark Q&A to establish a testing and evaluation process for practical response accuracy.

Visualization of Organizational Challenges: Identified challenges in "AI-Ready Data Structure," such as the need for manual development to achieve an AI-native organization.

Results and Future Outlook: Towards "Data Structure Reform" to Unleash AI's True Potential

Comparison before and after the introduction of this project and the resulting outcomes:

Item

Before Introduction

After Introduction

Concierge Support Time

120 hours/month (approx. 180 minutes/inquiry)

40 hours/month (80 hours/month reduction, 960 hours/year reduction)

User Self-Resolution Time

Average 1,500 hours/month of trial-and-error for the entire design department (5 hours/month per Revit user)

Instant responses via AI

Average 1,500 hours/month (18,000 hours/year) reduction

Knowledge Management

Simple recording in Excel (individualistic, no images)

Automatic storage and generalization of history/evaluations via SharePoint

Annual Reduction in Hours

-

18,960 hours (equivalent to the labor of approximately 10 people)

Added Value to Operations

Engaged in identifying causes and manual individual responses

Shift to developing standards, updating manuals, etc.

Future Outlook:

Utilize the "importance of data structuring" visualized through this MVP construction as organizational learning, and aim to create a mechanism for knowledge management and accumulation for improved AI accuracy under our own initiative.

Spillover to the entire company through horizontal expansion to other departments and software:

Using the chat-bot format foundation and response control methods built this time as a model case, rapidly expand horizontally in conjunction with soliciting business improvement ideas within the design department.

Advanced Automation and Stabilization of Transformation:

In the future, aim for the stabilization of more advanced business transformation, such as enabling the AI agent to perform the same actions as humans to make operations self-sustaining and automated, rather than being limited to text responses.

Comment from Mr. Sakakibara, Design Solutions Department, Design Headquarters, Obayashi Corporation

Even though we communicated internally about how "AI cannot be fully utilized without organizing data," it was difficult to convey the message. However, through the agent built this time, the importance of data structuring has been understood by the members, making internal communication much easier. This is a significant gain in addition to the quantitative outcome of reduced work hours. Moving forward, we will use this foundation as a model case for horizontal expansion to other departments, and in the future, aim for advanced utilization where the agent performs actions similar to humans, supporting operations. We want to continue challenging ourselves with Skillup NeXt as a strong partner for stabilizing company-wide business transformation in the construction industry.

About Copilot and Power Platform Stabilization Support / AI Agent Construction Support

We provide end-to-end support from employee skill improvement to organizational transformation through Copilot-driven operational efficiency and AI agent development. We build agents in a short period and create a system for utilizing AI across the entire company.

▼ Service Details Here https://www.skillupai.com/private-training/copilot/

About Skillup NeXt Inc. Skillup NeXt Inc. is an AI solutions company that realizes corporate transformation through the "implementation" of cutting-edge technologies such as AI agents, their "stabilization" within the organization, and "training" to master them. We have supported over 1,000 companies and local governments.

With the mission of "Promoting the transformation of people and organizations and supporting corporate evolution," we are expanding our business into next-generation fields such as GX (Green Transformation) and quantum, in addition to AI/DX.

【Company Overview】 Established: 2018 President: Shinichi Tahara Location: 2-40-5 Kanda Jinbocho, Chiyoda-ku, Tokyo 101-0051 URL: https://skillup-next.co.jp/ Business Activities:

AI/DX Business (AI development/implementation, human resource development/organizational development, stabilization/operation support)

GX Business (Strategy formulation, human resource development, GX certification planning/operation)

Quantum Business (Quantum computing related services, etc.)