Skip to main content
AgentOn Security · AI Security Domain Service

Out of countless security issues,
identify the risks to act on first

AI connects assets, vulnerabilities, information systems, organizations and security policies, and analyzes security risks according to each user's role and scope.
It provides remediation priorities and rationale to make security work more efficient.

KEY SUMMARY Security Domain Package · On-Premises LLM · Air-gapped operation · Connects to existing security workflows via Inference API
Field Issues

6 recurring problems in security operations

The same task poses different challenges for the users who carry out security work and the administrators who review and approve it.

Field IssueUsersAdministrators
01Submitting and reviewing security evidence
Users

Is the right evidence being submitted for each manual inspection item?

Users must find and submit the right evidence for each of many manual inspection items, but the criteria are hard to interpret, and incorrect submissions lead to re-review and resubmission.

Administrators

Can the time spent reviewing manual inspection evidence be reduced?

Each piece of submitted evidence must be checked against security criteria item by item, which takes significant time and staff; when reviews pile up, other security work is delayed as well.

02Deciding on security exceptions and deferrals
Users

When can a security exception or deferral be requested?

It is hard to tell whether a case qualifies for an exception or deferral, and writing a justification consistent with security policy takes considerable time and effort.

Administrators

What criteria should be used to approve security exception and deferral requests?

Approvers must review a wide range of request reasons and security risks, but repetitive similar cases and complex criteria increase workload and make consistent decisions difficult.

03Processing security exceptions and deferrals
Users

Should even simple exception and deferral requests wait this long for approval?

Even standardized low-risk requests must pass through multiple review and approval steps, so approval takes time and follow-up work can be delayed.

Administrators

Does every low-risk exception or deferral request need a full review?

The same review and approval process applies even to repetitive low-risk requests, increasing workload and making it hard to focus on critical security risks.

04Prioritizing vulnerabilities
Users

With so many vulnerabilities, where do you start?

Many vulnerabilities must be remediated in limited time, and misjudged priorities can delay or miss critical fixes.

Administrators

Are critical security risks slipping through your prioritization?

It is difficult to weigh vulnerability severity, asset criticality and external threat intelligence together, so high-risk vulnerabilities can be missed or priorities set incorrectly.

05Security management blind spots
Users

Do you have an accurate picture of the assets and inspection targets you are responsible for?

If asset and owner information is inconsistent or out of date, inspection targets or vulnerabilities that need action can be missed.

Administrators

Are asset data inconsistencies causing you to miss critical security risks?

Inconsistent or missing information across assets, information systems and owners can lead to misidentified inspection targets and blind spots in risk management.

06Handling security inquiries
Users

How long must you wait for answers to security questions?

To check inspection criteria, remediation methods or exception and deferral procedures, users must ask the person in charge, and delayed answers can hold up their work.

Administrators

Do repetitive security inquiries have to be answered one by one?

With limited security staff handling a wide range of inquiries, repetitive guidance consumes time and makes it hard to focus on critical security management.
Why AgentOn Security

Faster security work, more accurate risk decisions

AI simplifies repetitive tasks and complex security procedures.
By connecting asset, vulnerability and threat information, it helps you assess critical security risks more accurately and consistently.

AI-POWERED WORK EFFICIENCY Work Efficiency

Simplifies repetitive tasks and complex procedures.

  • 01 · Streamlined Evidence Review

    AI checks repetitive evidence reviews first.

    Compares and analyzes submitted security evidence against inspection criteria and provides a first-pass verdict, reducing resubmissions for users and review time for administrators.

  • 02 · Approval Decision Support

    AI helps review complex exception and deferral decisions.

    Analyzes request reasons and security risk information to provide the grounds needed for approval review, helping staff decide by consistent criteria.

  • 03 · Optimized Approval Workflow

    Fewer unnecessary approval steps for low-risk exception and deferral requests.

    AI classifies standardized, low-risk request reasons and optimizes approval routing based on predefined business rules to improve processing efficiency.

INTELLIGENT RISK DECISIONS Risk Assessment

Identify critical risks with connected data.

  • 04 · Security Risk Analysis

    Across all your vulnerabilities, see the critical risks first.

    Analyzes asset criticality, vulnerability severity and external threat intelligence together to surface the security risks to check and address first.

  • 05 · Security Data Visibility

    Connect scattered security data, and the risks become visible.

    Connects assets, information systems, owners and vulnerabilities in an ontology so AI understands the context of security data and analyzes critical risks more accurately.

AI ASSISTANT

Have a security question? Ask the AI directly.

  • 06 · Conversational Work Support

    Security questions? Don't wait for the person in charge.

    Ask the AI Assistant about security inspection criteria, procedures, and asset and vulnerability information in natural language, reducing the burden of repetitive inquiries.

Ontology — a model that represents the meaning and structure of data by explicitly defining a domain's concepts, attributes and the relationships between them

Service Architecture

Connect your existing security data and add AI analysis and judgment

AgentOn Security service architecture — existing security systems and external security data are connected through MCP, and the results analyzed by Data Fabric · Ontology · AI Agent are delivered to security operations and LLMs SECURITY OPERATIONS Security Workflows Remediation Exceptions Risk Reporting Policy Checks Audit Response LEGACY SYSTEMS IT Assets Service Security Vuln. Scanning Infrastructure Security HR · Org Management Security Policy Compliance EXTERNAL SECURITY DATA NVD (CVE) CISA KEV EPSS ONTOLOGY AGENT PLATFORM AgentOn Security AI AGENT Risk analysis · Priority · Rationale ONTOLOGY Relationship · Context model DATA FABRIC Connect · Sync · Map · Automate INTEGRATION MCP Server Tools Security data integration API · DB · File Analysis · inference results Priority · Rationale · Actions Reuse existing data · functions Use analysis · inference results Collected data is mapped to Ontology objects and relationships, independent of source system data structures. Inference API LLM Reference LLM Customer LLM

NVD — CVE-based national vulnerability database operated by the U.S. NIST · CISA KEV — catalog of known exploited vulnerabilities maintained by the U.S. CISA · EPSS (Exploit Prediction Scoring System) — metric that estimates the probability that a CVE will be exploited within the next 30 days · MCP (Model Context Protocol) — protocol that lets AI agents connect to and call external tools in a standard way

Key Capabilities

Key Services and Features

Delivers AI-based security analysis through OAP Console and Inference API.

01 OAP CONSOLE

OAP Console

Web console for security operators

  • 01

    Data Integration

    Collects, links and integrates security data and automates data processing and synchronization

  • 02

    Ontology Modeling

    Models and visualizes security objects and their relationships as an ontology

  • 03

    Vulnerability Topology

    Visualizes the impact relationships between discovered vulnerabilities and information systems, operating environments and infrastructure

  • 04

    AI Analysis

    Analyzes and infers security risks using an ontology and provides priorities and rationale

02 INFERENCE API

Inference API

Inference API for legacy systems and expert users

  • 01

    System Inference API

    AI inference API for external system integration

    System integration API that lets legacy and external systems call AgentOn Security's AI analysis and inference functions

  • 02

    User Inference API

    AI inference API for authorized users

    User API that lets authorized users call AgentOn Security's AI analysis and inference functions directly from development and analysis environments on their local PCs

HUMAN REVIEW Review of AI Decisions

Configure AI decision criteria to match your organization's security policies and operating standards,
and have security experts review critical decisions directly.

Human-in-the-Loop control over AI decisions

  1. 01

    AI Configuration

    Decision Criteria Settings

    Sets the criteria AI applies to its decisions, such as security policies, risk criteria and priorities

  2. 02

    AI Analysis

    Analysis and Inference

    Presents risk level, priority and rationale based on Ontology Context and configured criteria

  3. 03

    Human Review

    Expert Review

    Staff verify, correct and approve AI decisions that are critical or need review

Specifications

Product Components and Supported Specifications

Configured to match your workload, from Standard GPUs to high-performance Data Center GPUs.

  • OAP CoreData Fabric · Ontology · Agent · LLM Runtime · Validated Open-weight LLM
  • Security Domain PackageRisk Monitor · Topology · AI Assistant · Priorities · Asset Integrity
  • Inference ServiceOAP Console · System Inference API · User Inference API
  • Additional ServicesDeployment and installation · Customer system integration · Ontology development/consulting · LLM and GPU Optimization

LLM Support

ItemSupported Specifications
LLMReference Model based on open-weight LLMs — deployed and operated directly on the customer's own servers and GPUs
Customer LLMIntegration with customer-owned Custom LLMs
Deployment ModelOn-Premises LLM · Air-Gapped Network

Inference Capacity

GPU ConfigurationRecommended Concurrent Inference RequestsRecommended Use
A100 80GB × 1Approx. 4–6Legacy / Entry
RTX PRO 6000 Blackwell × 1Approx. 8Standard
H100 × 1Approx. 12Performance
H200 × 1Approx. 24High Performance

Recommended concurrent inference requests are reference figures based on the same LLM and inference environment with a 32K context; actual performance varies with the model and inference settings. Validated open-weight LLMs are supported by default; applying new models and GPU optimization fall under separate technical support.

Related Products
Broadcom Private AI
Private AI Infrastructure Platform

A private AI environment in the same AI product family. We design it together with the VCF-based platform that hosts on-premises LLMs and AI workloads.
GenWorks
GPU Cluster Management & AIOps Platform

We deploy and operate the GPU cluster that hosts on-premises LLMs and allocate its resources.
Get Started

AI-Powered Security Intelligence

AI handles the routine. Your people handle the real risks.

Keep the security management system you use today. We show you what changes when you connect the data.

Request a Demo →Request a PoC →