XGEN · Enterprise Agentic AI Platform
Enterprise AI that keeps going after adoption — into operation
XGEN is an AX Evolution Platform that grows enterprise AI step by step. Agents, knowledge and tools you build become operational assets you extend to the next workload.
GS certification, Grade 1
National quality certification · TTA
AI-MASTER
AI reliability certification in progress
On-premise · air-gapped
Air-gap deployment supported
Deployed at Jeju Bank and others
Finance · public sector · commerce
The Challenge
A great many AI projects finish at technical validation and never reach the actual work.
Documents, knowledge, and legacy sit apart, so the AI never grasps the organization's context.
Data sovereignty, audit, and approval requirements make external AI hard to adopt.
The Decisive Difference
XGEN absorbs change in your work and environment and turns it into evolution, building the AI you deploy into an operational asset.
Deployment, monitoring and improvement continue on the same platform after the PoC.
Agents, knowledge and tools are shared and reused in the next workload.
Execution records and evaluations are used to refine how work gets done.
AX Evolution Model
Enterprise AI grows in stages: understand → execute → collaborate → evolve. XGEN provides the capabilities each stage needs.
01UnderstandAssisted
Answers questions and understands internal documents.
02ExecuteAugmented
Connects to systems and carries out tasks.
03CollaborateAutonomous
Multiple agents split roles and work together.
04EvolveSelf-Evolving
Refines how work is done from execution records and evaluations.
6 Capabilities
Purple marks technology developed in-house by Plateer AI Labs.
Knowledge
Re-reads documents by meaning and relationship. A hybrid Vector + Graph structure finds the evidence and the relations behind an answer.
Workflow
Design workflows on the canvas and turn existing APIs into tools with PathFinder. Vibe Coding in DexOn builds agents and apps through conversation.
Model
Picks the model that fits the task and raises consistency with multi-step verification and judge evaluation.
Orchestration
Agents collaborate by role, and the MCP Compiler manages execution logic to support standalone runs.
Operations
Real-time monitoring and evaluation-feedback loops manage performance and trace execution history. Geny reviews its approach after each conversation and applies it next time.
Governance
Applies RBAC/ABAC access control, multi-layer guardrails, approval routes and AI risk review in on-premise environments.
Scope and behavior depend on the deployment environment and configuration.
XGEN · Agentic AI Platform
It doesn't stop at the pilot — the proof is in numbers validated in real on-premise operation.
Local
On-premise deployment support
Network separation · data sovereignty
110+
Built-in tools and plugins
Extensible with your own
Grade 1
National GS certification
Verified by third-party testing
20
Public libraries
Install with pip or npm
4 Core Values
Building the AI is not where it ends. Connecting it, operating it, governing it, and continuing to extend it — XGEN provides the whole path on one platform.
(Agentic AI)
Proactive automation through multi-agent workflows and task routing
(Grounded Answers)
Hybrid RAG and five-stage precision retrieval keep answers grounded and hallucination low
(Multi-LLM Routing)
GPT, Claude, private LLMs — dynamic model selection per requirement, with no vendor lock-in
(High Availability)
Supports k3s high availability and GitOps deployment. Service continuity depends on redundancy, health checks, and application configuration.
Build for Everyone
Without development knowledge, business users design agents themselves and connect them to existing systems. Cutting out the complicated development path turns an idea directly into AI that can be used at work.
Vibe Coding in DexOn
Build agents and apps with vibe coding
Inside DexOn, move from vibe coding—describing the agent or app you want in natural language—to hands-on code editing. Project context, implementation, execution, testing, and error correction stay in one continuous flow.
DexOn
Run the agents you built on your team's devices
DexOn is XGEN's Device Interface workspace for connecting agents to the user's PC, phone, and tablet working environments. Within the scope you allow, it uses files and applications on each device to produce real deliverables.
PathFinder
Connect existing systems as tools the AI can use
PathFinder connects web systems and APIs as Agent Tools an AI agent can use. Sign-in, connection, testing, and registration are automated without code, so existing systems extend quickly into an AI working environment.
EasyMode
No coding — an agent in about 90 seconds
A guided drag-and-drop builder lets non-developers combine knowledge, prompts, and tools into an agent. The moment it exists, you validate it in chat.
Business Value
Enterprises have moved past simply adopting an AI model. They are choosing an operating platform that connects existing systems to AI safely and applies it to real work on a foundation of security and governance. XGEN is not a platform for adopting AI — it is the Enterprise AI platform for continuously extending how you run it.
Past feature advantage — five outcomes that become organizational strength on day one: efficiency, risk, trust, flexibility, and regulatory readiness, all on one platform.
Fast, accurate RAG-based retrieval cuts the time repetitive internal work takes
Five categories of risk — model, data, security — controlled up front for a safe AI environment
Minimal hallucination and transparent evidence raise internal users' confidence
Multi-LLM operation through the Model Router secures the best cost-performance structure without lock-in
Transparent governance aligned with AI-MASTER criteria and data security law
The same AI, operated differently. Moving past build-style SI and PoC-bound approaches, XGEN shifts to a model where business teams run AI on a finished platform.
ResultShorter build time — months of building down to weeks to operational readiness
Built anew for every project, so build time and maintenance cost keep climbing
Services composed on a finished platform by combining agents, RAG, and workflows
ResultPast the PoC, in use in production straight away
AI that passes technical validation but never reaches the actual work
Enterprise AI services usable immediately in an operating environment that includes permissions, security, and audit
ResultBusiness teams improve things themselves, without waiting on development
AI service launches wait on the development organization's schedule
With the no-code canvas and PathFinder, business users build agents and apply them to work immediately
ResultOperating tools consolidated into one platform
Development, deployment, monitoring, and quality management sit apart, raising operational complexity
Design, knowledge management, deployment, evaluation, and operation managed on a single platform
Features
These are not separate tools but a single operating pipeline. Building an agent, deploying it, running it, and improving it all connect inside one platform.
Inside DexOn, move between vibe coding—describing what you want in natural language—and full coding with direct code control. Build, run, test, and refine agents and apps, then continue work started on a PC from a tablet or phone.
Upload documents into a collection and chunking and embedding make them vector-searchable. Rerankers and ontology raise accuracy, and responses carry their citations.
Register external APIs and functions as tools an agent can call, and connect models to external tools through the MCP standard. Credential profiles keep keys and tokens managed safely.
Assign collaborating departments, governance officers, or information-security managers as approvers for an AI service risk review. Configure approval routes by organization and risk level, then manage every review, decision, and audit record consistently.
01 · Approval routing
02 · AI risk grades
Two deployment paths serve different purposes, and the deployment environment can be customized to your infrastructure.
Users share agents built in their own development environment with colleagues. External system calls, web embedding, and scheduled runs are also supported.
Apps built with vibe coding are provided in a per-user development environment. For public release, users request deployment from a system administrator, and the app is served through a registered deployment server.
01 · Agent deployment
02 · App deployment server
Role-based widgets put status and statistics on one screen. Execution history, token usage, and system health, seen from an administrator's perspective.
01 · Operations dashboard
02 · System monitoring
More product screens
Core Technology · 6 Layers
From infrastructure up to AI agents, the core technology enterprise AI operation needs, integrated into one platform.
XGEN is not bound to a particular AI model or cloud. A standards-based six-layer architecture connects and operates on-premise GPUs, commercial AI APIs, and internal systems in a consistent way.
Agent Runtime · Geny
Geny is the runtime inside XGEN that defines how agents work. With Memory, Workspace, Tools, and Schedule built in, it helps an agent pick up earlier context and build on past results in the next task. The existing way of building agents in XGEN Canvas stays the same.
Collects information from conversations and tasks, and finds what later work needs.
Keeps documents, materials, and tools in a separate workspace per agent. Shared agents and tools build up as organizational assets.
Finds the tools the current task needs instead of loading every connected tool at once.
When a needed tool does not exist, the agent writes and tests the code and saves it to the Workspace.
Runs recurring work, such as regular reports and data checks, at set times.
After each conversation, reviews how the work went and adjusts its instructions and tool setup for the next one.
Geny defines how an agent works; DexOn defines how far it can reach. Together, an agent can use the work environment on user devices within the allowed scope.
Geny is generally available in production. Availability can vary by environment and configuration.
Learn more about GenyIsolated Execution
Agents run code and reach external systems. XGEN separates where decisions and execution happen, and designs the execution environment on the assumption that a malicious document or prompt injection can lead an agent to run the wrong command.
Isolation level and permission scope can vary by deployment environment and configuration.
For Every Role
Planners, developers and administrators each use, build and govern AI from where they sit.
On-Premise
XGEN supports processing within the enterprise security boundary when configured with internal models. External integrations require review of data transfer scope and provider training and retention policies.
On-premise deployment with internal models supports processing within your security boundary. External integrations require separate transfer policies.
Deploys and runs in network-separated and air-gapped environments, meeting the closed-network requirements of finance and the public sector.
Controls help restrict the use of customer data for external model training. External integrations require review of provider training and retention policies and contract terms.
Deployment
From closed-network intake through proof, build, and operation, support is staged around the on-premise environment. Once intake completes, proof starts quickly, supporting pre-production validation and early results.
Installed into your infrastructure (on-premise or air-gapped) through the closed-network intake process.
Validated quickly against real data, settling the domain requirements and review rules.
Domain tools, knowledge, and agents configured and integrated around the organization's work.
Controlled deployment, monitoring, and improvement continue through the GitOps pipeline.
It doesn't end at the build. Training that lands the capability inside your team, and technical support that keeps it running steadily.
Enterprise Trust
XGEN traces who did what and when, verifies risky changes before deployment, and is designed so that only approved agents reach production. It provides governance you can rely on in regulated industries and on-premise environments.
Explore the security and governance architectureTier (Standard/SuperUser), role, and permission (ABAC keys) as independent layers, gating access down to individual screens and buttons.
Turning something into a service requires passing two stages: deployment approval from a system administrator and approval from a governance officer.
Personally identifiable information is masked automatically, and the risk level of requests and responses is classified and controlled.
User activity and system events are retained, and deployed agents are reviewed on a schedule.
MFA (OTP), IP allowlisting, and session timeouts are applied together, so only permitted users get through.
OWASP ZAP is embedded in the development and CI/CD pipeline to check vulnerabilities automatically before release — shift-left security.
Certifications & Quality
XGEN has been awarded Grade 1 GS certification and is undergoing AI-MASTER, the AI reliability certification. Accredited third-party laboratories verify the product's quality and reliability.
Good Software Certification · highest grade
XGEN has been awarded Grade 1 in GS (Good Software), Korea's national software quality certification. TTA, an accredited third-party laboratory, tested functionality, reliability, usability, and overall quality, and verified it at the highest grade.
Certificate No.: 26-0296
AI reliability certification · Korea AI Industry Association (AIIA)
XGEN is undergoing AI-MASTER, an AI reliability certification in which a third party verifies the reliability, transparency, and robustness of the AI.
AI Quality Policy
Beyond certification — we publish the standards for trustworthy enterprise AI, defining everything from safety, accountability, and controllability to where our responsibility ends and yours begins.
Reviewing security, regulatory, or procurement requirements? We will work through the architecture and controls against your requirements with you.
Discuss security requirementsCustomer Cases
Customer cases where XGEN was actually built and operated in finance, commerce, public sector, and IT.
Plateer is building an XGEN-based, on-premise enterprise AI platform that enables Company I to apply AI agents and RAG safely to internal capital-finance work. The engagement also establishes one operating and governance framework across agent development, validation, approval, deployment, operations, and audit.
Company I · Capital finance
Building an internal generative-AI platform on XGEN with Jeju Bank, then stabilizing it through on-site operations and widening its use across the bank
Jeju Bank
An XGEN-based AI coding assistant deployed on-premise into an air-gapped development environment with strict security requirements
A defense and systems company
FAQ
XGEN is an on-premise Enterprise AI platform — an agent development environment for designing, deploying, operating, and governing Agentic AI services on the LLMs and infrastructure you choose. It is not a finished service delivered to you; you build and run the agents your organization's work needs on top of XGEN. It was designed and developed by the Plateer AI Labs AI research team.
You are not tied to a particular model. Choose among open-weight models such as Qwen and our own sLLM (Polar) by purpose, cost, and accuracy, and serve them on the GPUs and infrastructure you already own. The management center configures the LLM, vector DB, and the rest of the system in detail.
Past simple text extraction (OCR), a VLM (vision-language model) reads the context of images, tables, charts, and graphs. OCR is one capability within the VLM; where needed, an open-weight VLM is served directly to turn charts and graphs into text, vectorize them, and use them in answers. This is already applied in commerce and finance settings.
Yes. Even without tidy data, accumulating a few months of it and constructing an ontology links the different expressions of the same entity (for example US, USA, United States) into one. The richer the data the better the quality gets, and domain terminology mapping and review-rule design happen alongside. We have introduced knowledge graphs in commerce and financial projects.
Quality is scored on RAGAS from preprocessing through answer verification and evaluated across several dimensions. Preprocessing strategy is also selected automatically by document type — an adaptive pipeline where the LLM judges the document type and adjusts whether OCR or VLM applies, along with chunking and overlap parameters.
You don't need to build new screens. LLM and RAG integration is composed as nodes and provided as snippet code, so adding a single button to an existing system to call it is entirely workable. XGEN is plugin-structured throughout — around 80 built-in tools combined with quickly-built custom tools keep integration and customization effort to a minimum.
A first filter applies at creation through risk assessment and administrator approval, and if something goes wrong in operation an admin kill switch stops the service immediately. Where accuracy is decisive we apply human-in-the-loop, designing the AI to raise throughput through pre- and post-processing rather than replacing the work entirely, with a person confirming the result before the next step.
On-premise deployments using internal models can process data within the enterprise security boundary and support network-separated and air-gapped environments. RBAC/ABAC manages access. External model and tool integrations require separate review of data transfer scope and provider training and retention policies.
Yes. Agentflow's drag-based visual canvas lets you design a workflow without code, validate it in live chat, and deploy it as an API or workflow.
We check and provide the obligations attached to each license type of the open source that makes up XGEN (copyright notices, NOTICE files). Components and obligations can vary with the delivered version and configuration, so we prepare a component-by-component license inventory during evaluation.
Get Started
Start with a 15-day trial in your browser, or see the full build–deploy–govern flow in a demo.
