On-Premises Autonomous Defense

Purple AgentOn-premises-only autonomous defense

Vulnerability assessment and forensics, both with a local LLM. Completed within a closed network

A fully on-premises AI security product in which an analysis platform equipped with a purely domestic local LLM is installed directly in your organization's environment. No internet connection is required, and the data being analyzed never leaves for the outside. Available since December 2025.

Request Purple Agent materials View features

Exterior of the Purple Agent appliance enclosure

NEWS2025.12.16Press release (in Japanese): Launch of "Purple Agent", a purely domestic AI security product that runs entirely on-premisesLearn more

Issues

Do any of these challenges sound familiar?

CASE 01

Cloud-based tools cannot be used because of the nature of the data we handle

We handle data that cannot be sent to external services, such as investigation materials and confidential information, and closed networks or strict standards rule out cloud-based options.

CASE 02

We cannot hand analysis data to foreign AI

Even if we want to use AI for analysis, we cannot entrust data to overseas AI platforms because of data handling and governing-law concerns.

CASE 03

We want to complete analysis work inside a closed network

Data cannot be taken out and outsourced, so we need a setup in which forensics and vulnerability assessment can be completed within the organization.

Capability

Red work and Blue work, all within a closed network

Both offensive (red team) and defensive (blue team) work can be run as-is inside a completely closed environment.

RED

Vulnerability assessment & penetration testing

Run attacker-perspective assessments and penetration tests without sending any information about the target system outside. Even if the target sits inside a closed network, it can be assessed as-is.

BLUE

Forensics & malware analysis

Complete everything from analysis to report creation inside your environment, without sending evidence or samples outside. Investigation materials and confidential data can be analyzed on the spot.

All inference is executed by the built-in local LLM. Analysis data never leaves for the internet.

Features

Purple Agent Features

Analysis time cut to 1/10
AI automates forensics and malware analysis, greatly speeding them up.
Fully on-premises operation
Both the analysis platform and the AI are installed on your premises. No data leaves for the outside.
Purely domestic AI (CoWorker AI Agent) built in
Inference by our in-house developed, purely domestic LLM runs advanced analysis while staying in a closed network.
Covers both assessment and analysis
Supports vulnerability assessment and penetration testing, and forensics and malware analysis (static and dynamic).
Threat intelligence integration
Cross-references threat information to quickly identify the nature of an attack.
Legal compliance support
Helps you meet reporting obligations, such as the initial report within 72 hours.

Flow

How It Works

01
Always-on local monitoring
Fully closed operation with a purely domestic LLM
02
Hybrid of assessment and detection
Vulnerability assessment and real-time threat detection within a closed network
03
Autonomous learning
Keeps up with the latest threats through continuous machine learning
04
On-site response
Supports the initial report within 72 hours

Structure

A configuration completed in a fully closed network

Both the terminal that runs the analysis and the AI (LLM) are installed on the user’s premises. The configuration assumes operation in an environment that does not use the cloud and sends no data outside.

ON-PREMISE

Everything installed on your premises

Both the analysis platform and the AI run inside the user’s environment.

Front terminal and LLM both installed on premises
Configuration that does not use the cloud
Assumes operation in a closed network

DOMESTIC AI

Equipped with purely domestic AI

Equipped with our in-house developed, purely domestic AI, “CoWorker AI Agent”.

Inference by a purely domestic LLM
No data sent overseas
Supports analysis of highly confidential data

USE CASE

Expected usage environments

Intended for use by organizations that cannot send data outside.

Government agencies and investigative bodies
Critical infrastructure operators and large enterprises
Security vendors and research institutions handling confidential data

FAQ

Frequently Asked Questions

Q.Is there really no communication with the outside?

A.The configuration assumes that both the terminal that runs the analysis and the AI (LLM) are installed on the user’s premises, with no use of the cloud. Operation can be completed within a closed network.

Q.What kinds of work can it handle?

A.It supports vulnerability assessment and penetration testing, as well as forensic analysis and malware analysis (static and dynamic). All of it is completed inside the closed environment, and all inference is executed by the built-in local LLM.

Q.What is the process up to adoption?

A.After advance meetings and hearings, we propose a configuration that fits your environment. At delivery, we handle terminal kitting, initial setup, and onboarding.