Report on the Results of the NEDO Project 'Promotion of R&D and Verification for Ensuring AI Safety / R&D on Strengthening AI Safety'
Citadel AI has published the 'Generative AI Practical Guide and Enterprise Case Studies' as a result of a NEDO-commissioned project, providing practical frameworks for safe AI production deployment based on interviews with over 20 companies.
📋 Article Processing Timeline
- 📰 Published: April 3, 2026 at 19:00
- 🔍 Collected: April 3, 2026 at 10:30
- 🤖 AI Analyzed: April 21, 2026 at 04:36 (426h 6m after Collected)
Citadel AI Inc. (Representative Director: Hironori Kobayashi, hereinafter "Citadel AI") has been commissioned by the New Energy and Industrial Technology Development Organization (NEDO) to work on the "Promotion of R&D and Verification for Ensuring AI Safety / R&D on Strengthening AI Safety" project starting in FY2025.
This project promotes R&D and verification related to ensuring AI safety, aiming to integrally formulate and disseminate AI safety standards and develop AI safety evaluation and management technologies necessary for properly managing and utilizing generative AI.
We are pleased to announce that as one of the results of this project, we have compiled the "Generative AI Practical Guide and Enterprise Case Studies: A Framework for Integrating Quality, Safety, and Governance Leading to Production Deployment."
After conducting interviews with over 20 companies, we systemized the challenges and responses related to the utilization and governance of generative AI in a practical manner, taking into account existing concepts and frameworks. This document is presented as an implementation guide for enterprises, featuring specific cases that can be reproduced in other organizations.
1. Background
Generative AI has become a familiar technology, and instances of AI agents being used in actual businesses have increased. On the other hand, risks related to security and reputation, such as hallucinations and information leaks, have also become apparent.
Many organizations are currently proceeding by trial and error, facing challenges such as, "We don't know where to start," "Considering the risks, we hesitate to put it into production," and "Guidelines exist, but we don't know exactly how to apply them to our own services."
The lack of established "specific methods for quality assurance and risk management," which are essential for responsibly providing services as a company, has become a major barrier to full-scale introduction, and there are many cases where organizations cannot proceed from Proof of Concept (PoC) to actual production deployment.
2. Overview of Results
The report "Generative AI Practical Guide and Enterprise Case Studies: A Framework for Integrating Quality, Safety, and Governance Leading to Production Deployment" aims to provide organizational guidelines for AI utilization by sharing the challenges companies face in the field and concrete solutions and operational methods for them.
Based on interviews with over 20 companies that are actually developing and operating AI agents, the specific insights and practical know-how gained from these efforts regarding the utilization of generative AI, including AI agents, were organized from the perspectives of "technology," "process," and "organizational culture," systemized into a framework, and reflected in this report.
Citadel AI itself also engaged in the development of an AI agent, comprehensively performing everything from setting the issues to be solved to selecting technologies, collecting necessary data, identifying anticipated risks, and evaluating the system, thereby verifying the effectiveness of the framework.
The Appendix features numerous case studies showing for what purposes the interviewed companies use AI agents, and what countermeasures they take against specific challenges. Additionally, it details how Citadel AI actually implemented a "chatbot designed to support foreigners".
This project promotes R&D and verification related to ensuring AI safety, aiming to integrally formulate and disseminate AI safety standards and develop AI safety evaluation and management technologies necessary for properly managing and utilizing generative AI.
We are pleased to announce that as one of the results of this project, we have compiled the "Generative AI Practical Guide and Enterprise Case Studies: A Framework for Integrating Quality, Safety, and Governance Leading to Production Deployment."
After conducting interviews with over 20 companies, we systemized the challenges and responses related to the utilization and governance of generative AI in a practical manner, taking into account existing concepts and frameworks. This document is presented as an implementation guide for enterprises, featuring specific cases that can be reproduced in other organizations.
1. Background
Generative AI has become a familiar technology, and instances of AI agents being used in actual businesses have increased. On the other hand, risks related to security and reputation, such as hallucinations and information leaks, have also become apparent.
Many organizations are currently proceeding by trial and error, facing challenges such as, "We don't know where to start," "Considering the risks, we hesitate to put it into production," and "Guidelines exist, but we don't know exactly how to apply them to our own services."
The lack of established "specific methods for quality assurance and risk management," which are essential for responsibly providing services as a company, has become a major barrier to full-scale introduction, and there are many cases where organizations cannot proceed from Proof of Concept (PoC) to actual production deployment.
2. Overview of Results
The report "Generative AI Practical Guide and Enterprise Case Studies: A Framework for Integrating Quality, Safety, and Governance Leading to Production Deployment" aims to provide organizational guidelines for AI utilization by sharing the challenges companies face in the field and concrete solutions and operational methods for them.
Based on interviews with over 20 companies that are actually developing and operating AI agents, the specific insights and practical know-how gained from these efforts regarding the utilization of generative AI, including AI agents, were organized from the perspectives of "technology," "process," and "organizational culture," systemized into a framework, and reflected in this report.
Citadel AI itself also engaged in the development of an AI agent, comprehensively performing everything from setting the issues to be solved to selecting technologies, collecting necessary data, identifying anticipated risks, and evaluating the system, thereby verifying the effectiveness of the framework.
The Appendix features numerous case studies showing for what purposes the interviewed companies use AI agents, and what countermeasures they take against specific challenges. Additionally, it details how Citadel AI actually implemented a "chatbot designed to support foreigners".