Virtualex Achieves "Business Process Optimization" Using Generative AI in Email Response Operations for TEPCO Energy Partner
Virtualex Consulting assisted TEPCO Energy Partner in implementing the generative AI chatbot "KotoznaTPG" for email responses, resulting in a 1.7x increase in productivity and an 11.1-point improvement in quality.
📋 Article Processing Timeline
- 📰 Published: April 14, 2026 at 00:30
- 🔍 Collected: April 13, 2026 at 16:35
- 🤖 AI Analyzed: April 19, 2026 at 20:36 (148h 0m after Collected)
Virtualex Consulting, Inc. (Headquarters: Minato-ku, Tokyo, President and Representative Director: Hayato Maruyama, hereinafter "Virtualex"), a member of the Virtualex Group, supported the implementation of the generative AI chatbot system "KotoznaTPG" at TEPCO Energy Partner, Inc. (Headquarters: Chuo-ku, Tokyo, President and Representative Director: Momoko Nagasaki, hereinafter "TEPCO EP").
At TEPCO EP, operators were meticulously drafting responses to email inquiries from corporate clients tailored to individual circumstances while using templates as a base. To maintain this detailed response quality while evolving into a system capable of executing tasks more quickly and efficiently, they were undergoing a review of their business processes.
To solve these challenges, Virtualex utilized a business alliance scheme with Kotozna, Inc. (hereinafter "Kotozna") to introduce the generative AI chatbot "KotoznaTPG". This project goes beyond simply introducing an AI chatbot; Virtualex's operations team, which has been engaged in contact center operations (BPO) for many years, is leading the "prompt engineering" to implement business knowledge into the AI. As a result, they achieved high standards in email response operations, improving productivity by 1.7 times and response quality (Good rate) by 11.1 points. Significant improvements were confirmed in both operational efficiency and customer satisfaction.
Challenges Before System Implementation: Increased Man-hours for Manual Drafting and Developing Knowledgeable Staff
Lack of Contextual Flexibility: With pre-prepared templates, not only did man-hours increase, but it also took time to train personnel capable of handling diverse inquiries. Furthermore, it was difficult to draft responses that fully captured the circumstances and context of each customer's inquiry.
Bloated Man-hours Due to Manual Corrections: To bridge the gap between the template and the actual response content required, operators had to make many additions and corrections, prolonging the response time per case.
Inconsistent Response Quality: The expressions and politeness of the response text depended on the individual writing skills of each operator, leading to variations in response quality.
Effects After System Implementation (PoC Results)
Email Response Productivity Increased by 1.7 Times: By having generative AI automatically generate drafts of responses, the time required for writing and correcting texts in email responses was drastically reduced (the number of cases processed per hour improved from 2.4 to 4.1).
Achieved an 84.9% Response Quality Good Rate: While accurately referencing internal regulations and proprietary data using "RAG (Retrieval-Augmented Generation)" technology, highly accurate response generation was achieved through prompt engineering. The quality score improved by 11.1 points from 73.8% before implementation to 84.9%.
Suppression of Hallucination Risks: By generating responses based on a proprietary database, the risk of misinformation (hallucinations) specific to generative AI was minimized, ensuring reliability capable of withstanding actual business operations.
Success Factor: Advanced Fusion of High Value-Added BPO and AI Tools
The success of this project is not simply due to the introduction of a high-performance AI tool. Its essence lies in the "high value-added BPO" provided by Virtualex. BPO personnel well-versed in Kotozna's prompt engineering skills are engaged in contact center operations, systematizing the tacit knowledge of the frontlines into the control of AI (prompts).
Furthermore, to realize this advanced operational know-how on the system, the characteristics of "KotoznaTPG"—which makes prompt engineering and tuning easy—were indispensable. Because it is a platform that allows intuitive and rapid adjustments aligned with frontline improvements, the "operations team that knows the business inside out (people)" and "KotoznaTPG that maximizes that knowledge (technology)" functioned together as two wheels of a car. This was the greatest factor producing these dramatic results.
Future Outlook
Moving forward, Virtualex will continue to support the improvement of client companies' CX (Customer Experience) and operational efficiency by fusing its three capabilities: "Consulting," "Technology," and "Operations." Particularly in the utilization of generative AI, we believe that "building and optimizing prompts" by professionals intimately familiar with contact center and BPO operations is the crucial element that determines the success or failure of practical application. Virtualex will not stop at mere AI implementation, but will continue to push forward with this "CC/BPO Knowledge x AI" approach.
At TEPCO EP, operators were meticulously drafting responses to email inquiries from corporate clients tailored to individual circumstances while using templates as a base. To maintain this detailed response quality while evolving into a system capable of executing tasks more quickly and efficiently, they were undergoing a review of their business processes.
To solve these challenges, Virtualex utilized a business alliance scheme with Kotozna, Inc. (hereinafter "Kotozna") to introduce the generative AI chatbot "KotoznaTPG". This project goes beyond simply introducing an AI chatbot; Virtualex's operations team, which has been engaged in contact center operations (BPO) for many years, is leading the "prompt engineering" to implement business knowledge into the AI. As a result, they achieved high standards in email response operations, improving productivity by 1.7 times and response quality (Good rate) by 11.1 points. Significant improvements were confirmed in both operational efficiency and customer satisfaction.
Challenges Before System Implementation: Increased Man-hours for Manual Drafting and Developing Knowledgeable Staff
Lack of Contextual Flexibility: With pre-prepared templates, not only did man-hours increase, but it also took time to train personnel capable of handling diverse inquiries. Furthermore, it was difficult to draft responses that fully captured the circumstances and context of each customer's inquiry.
Bloated Man-hours Due to Manual Corrections: To bridge the gap between the template and the actual response content required, operators had to make many additions and corrections, prolonging the response time per case.
Inconsistent Response Quality: The expressions and politeness of the response text depended on the individual writing skills of each operator, leading to variations in response quality.
Effects After System Implementation (PoC Results)
Email Response Productivity Increased by 1.7 Times: By having generative AI automatically generate drafts of responses, the time required for writing and correcting texts in email responses was drastically reduced (the number of cases processed per hour improved from 2.4 to 4.1).
Achieved an 84.9% Response Quality Good Rate: While accurately referencing internal regulations and proprietary data using "RAG (Retrieval-Augmented Generation)" technology, highly accurate response generation was achieved through prompt engineering. The quality score improved by 11.1 points from 73.8% before implementation to 84.9%.
Suppression of Hallucination Risks: By generating responses based on a proprietary database, the risk of misinformation (hallucinations) specific to generative AI was minimized, ensuring reliability capable of withstanding actual business operations.
Success Factor: Advanced Fusion of High Value-Added BPO and AI Tools
The success of this project is not simply due to the introduction of a high-performance AI tool. Its essence lies in the "high value-added BPO" provided by Virtualex. BPO personnel well-versed in Kotozna's prompt engineering skills are engaged in contact center operations, systematizing the tacit knowledge of the frontlines into the control of AI (prompts).
Furthermore, to realize this advanced operational know-how on the system, the characteristics of "KotoznaTPG"—which makes prompt engineering and tuning easy—were indispensable. Because it is a platform that allows intuitive and rapid adjustments aligned with frontline improvements, the "operations team that knows the business inside out (people)" and "KotoznaTPG that maximizes that knowledge (technology)" functioned together as two wheels of a car. This was the greatest factor producing these dramatic results.
Future Outlook
Moving forward, Virtualex will continue to support the improvement of client companies' CX (Customer Experience) and operational efficiency by fusing its three capabilities: "Consulting," "Technology," and "Operations." Particularly in the utilization of generative AI, we believe that "building and optimizing prompts" by professionals intimately familiar with contact center and BPO operations is the crucial element that determines the success or failure of practical application. Virtualex will not stop at mere AI implementation, but will continue to push forward with this "CC/BPO Knowledge x AI" approach.