Generative AI "FUJIRAG" Officially Released, Enabling Secure AI Utilization in On-Premises Environments
KTGS LLC has officially launched "FUJIRAG," a generative AI that can be used securely in an on-premises environment. This is significant for security-conscious companies as it allows them to leverage AI without exposing confidential corporate data externally.
📋 Article Processing Timeline
- 📰 Published: April 9, 2026 at 18:00
- 🔍 Collected: April 9, 2026 at 09:30
- 🤖 AI Analyzed: April 20, 2026 at 13:42 (268h 12m after Collected)
KTGS LLC (Headquarters: Nakano-ku, Tokyo; Managing Partner: Toshio Katagase) officially launched its new service "FUJIRAG," which integrates generative AI and knowledge search technology (RAG: Retrieval-Augmented Generation), on April 9, 2026. This service supports on-premises deployment and enables AI utilization without exposing confidential data by linking with a company's internal file server.
## Development Background
Companies accumulate a vast amount of knowledge daily in the form of manuals, FAQs, meeting minutes, and reports. A common challenge is that necessary information cannot be found, leading to a dependency on specific individuals. Furthermore, while the use of generative AI is gaining attention, many companies abandon its adoption, citing that "confidential information cannot be sent to the cloud." To solve these issues, KTGS developed FUJIRAG based on the concept of "generative AI that can be used without sending data externally."
## What is FUJIRAG?
FUJIRAG is a knowledge search platform where an AI understands documents existing within a company and generates optimal answers to questions in natural language.
Its greatest feature is its support for on-premises use, allowing for AI utilization while keeping internal data secure.
## Key Features
**1. On-Premises Support (AI that keeps data internal)**
FUJIRAG supports on-premises deployment, enabling AI utilization that is completed within the corporate network. Highly sensitive information can be used safely without being sent to an external cloud.
**2. File Server Integration**
It can link with the company's internal file server (NAS), allowing existing document assets to be used as is. Smooth implementation is possible without new data migration.
**3. High-Precision RAG Engine**
Based on internal documents and FAQs, the AI extracts relevant information and generates answers based on evidence.
**4. Natural Language Search**
Users can search using question formats like "What are the steps for XX?"
## Development Background
Companies accumulate a vast amount of knowledge daily in the form of manuals, FAQs, meeting minutes, and reports. A common challenge is that necessary information cannot be found, leading to a dependency on specific individuals. Furthermore, while the use of generative AI is gaining attention, many companies abandon its adoption, citing that "confidential information cannot be sent to the cloud." To solve these issues, KTGS developed FUJIRAG based on the concept of "generative AI that can be used without sending data externally."
## What is FUJIRAG?
FUJIRAG is a knowledge search platform where an AI understands documents existing within a company and generates optimal answers to questions in natural language.
Its greatest feature is its support for on-premises use, allowing for AI utilization while keeping internal data secure.
## Key Features
**1. On-Premises Support (AI that keeps data internal)**
FUJIRAG supports on-premises deployment, enabling AI utilization that is completed within the corporate network. Highly sensitive information can be used safely without being sent to an external cloud.
**2. File Server Integration**
It can link with the company's internal file server (NAS), allowing existing document assets to be used as is. Smooth implementation is possible without new data migration.
**3. High-Precision RAG Engine**
Based on internal documents and FAQs, the AI extracts relevant information and generates answers based on evidence.
**4. Natural Language Search**
Users can search using question formats like "What are the steps for XX?"