Mitsubishi Research Institute Launches AI Agent-Powered Support for Quantitative Business Impact Assessment
Mitsubishi Research Institute (MRI) has launched a support function that utilizes an AI agent to quantitatively assess the business impact of sudden external environmental changes. This new function, added to MRI's 'Intelligence Platform,' executes the entire process from creating impact scenarios to calculation and providing evidence, enabling rapid decision-making.
📋 Article Processing Timeline
- 📰 Published: May 18, 2026 at 23:00
- 🔍 Collected: May 18, 2026 at 14:31
- 🤖 AI Analyzed: May 18, 2026 at 14:39 (8 min after Collected)
Mitsubishi Research Institute, Inc. (President and CEO: Kenji Yabuta, hereinafter MRI) will begin offering a support function from May 18 to quantitatively estimate the impact of sudden changes in the external environment on business, using indicators such as sales, profits, and production volume. This has been added as a new feature to the 'Intelligence Platform,' an information foundation for strategic decision-making that MRI provides to companies. This function utilizes an AI agent to execute the entire process from creating business impact scenarios based on external environmental changes to calculation and providing the rationale, enabling swift implementation of impact assessments necessary for decision-making. 1. Background In recent years, sudden external environmental changes such as the Middle East crisis and export regulations have had significant impacts on companies, both in terms of losses and opportunities. For management decision-making, it is necessary to quantitatively estimate the impact of these changes, but this is difficult for many companies to do quickly due to the need for advanced expertise and enormous man-hours. MRI provides companies with the 'Intelligence Platform,' which implements the knowledge of information gathering and analysis cultivated as a think tank and consulting firm into an AI agent. The Intelligence Platform supports the sophistication of management decision-making through functional enhancements in management risk assessment, mid- to long-term management plan formulation and promotion management evaluation, and business opportunity exploration. By adding this function for quantitative impact assessment, we can now support management decision-making based on concrete evidence. Figure 1: Overall view of the Intelligence Platform - Created by Mitsubishi Research Institute 2. Overview and Features of This Function This function defines the target of impact assessment based on an impact scenario considering changes in the external environment, and performs everything from proposing calculation methods to trial calculations and presenting the rationale. ① Examples of Impact Assessment Targets: - Financial Impact: Sales/Profit (e.g., impact on operating profit in the event of a Taiwan contingency) - Corporate Activities: Procurement/Supply Chain (e.g., impact on monthly production capacity due to China's export control measures on critical items) - Others, set according to purpose ② Features of This Function: (1) The AI agent assembles the estimation process itself and calculates the degree of impact in an end-to-end manner, from creating impact scenarios → setting coefficients → calculation (approximate estimate with assumptions) → presenting the rationale. (2) By incorporating quantitative evaluation as part of the AI workflow for intelligence operations, it becomes easier to continuously perform evaluations directly linked to decision-making. (3) Since it is possible to dig deeper and make corrections as needed in a chat format, it supports the process of incorporating findings into practical documents while adjusting assumptions and reinforcing explanations. ③ Output Content (Output): It presents the scale of the impact in three patterns—pessimistic, neutral, and optimistic—along with the rationale. The goal is not a high-precision point estimate, but to emphasize the explainability and persuasiveness necessary for decision-making. Figure 2: Sample output report image (partial excerpt, figures show impact coefficients against a business-as-usual scenario) - Created by Mitsubishi Research Institute ④ Accuracy Verification Results As a result of internal verification at MRI, we confirmed a 9.2-point improvement in accuracy compared to using a stand-alone chat-based generative AI (see appendix). 3. Future Plans The 'Intelligence Platform' will be gradually enhanced with functions for analyzing the internal environment, in addition to the analysis of the external environment. Concurrently, we will expand its capabilities to quantitatively grasp and evaluate the impact on corporate activities, including the supply chain, not just on management indicators like sales and profits. In the future, we aim to evolve it into a comprehensive management support platform that integrates the results of external and internal analysis to consistently support everything from strategic reviews to their implementation in annual plans. Appendix: Business Impact Quantitative Evaluation Function by Agent Pipeline