AI Convergence Center

 LEE Gapheon, JO Yeonhwa, LIM Taewoo, JANG Hwasup

  1. Introduction

The field of artificial intelligence is undergoing a rapid paradigm shift toward AI agents that go beyond simple conversational models to interpret user intent and perform actual tasks by linking appropriate tools. There is a growing demand for AI agents that can identify data, execute relevant tools, and carry out practical business operations, moving beyond the level of simple Q&A.

In response to these global AI trends and internal demands for business innovation, Korean Register (KR) has developed MariNote.AI, a next-generation AI agent platform that provides integrated support for various internal and external tasks.

  2. Next-Generation Integrated KR AI Agent Architecture

This system creates a practical, interconnected work environment based on a three-axis structure in which the platform, engine, and data are closely integrated.

- MariNote.AI (Integrated AI Agent Platform): The interface through which users experience AI services and an integrated workspace where tasks are performed.

- AI Agent Hub (Business Agent Process Engine): The central engine of the system that manages AI agents based on workflows and links them with practical business tools.

- AI Data Lake (Business Knowledge Repository): A knowledge asset hub that standardizes fragmented business knowledge, such as manuals and checklists, into a format that AI systems can understand.

KR AI Agent Architecture

  3. Workflow and Tool Design for Practical Task Execution

The core innovation of MariNote.AI lies in the ability for users to directly design execution procedures and deploy specialized tools through the AI Agent Hub.

- Structured Step-by-Step Tasks: Users define execution steps tailored to the nature of the work, implementing complex tasks as systematic processes.

- Dynamic Attachment of Professional Tools (MCP & Task Tool): Necessary tools for actual business operations can be freely attached to each workflow stage. Through this, the corresponding AI agent performs advanced specialized tasks by executing professional tools sequentially.

[AI Agent Hub] Create a new Agent and configure workflow

  4. AI-Dedicated Data Environment and Data Bundle Supply

For AI agents to perform tasks accurately, a data environment that reorganizes legacy system data into an AI-compatible format is essential.

- Implementation of an AI-Optimized Data Environment: The AI Data Lake structures and standardizes fragmented legacy data into AI-ready knowledge assets for immediate deployment.

- Legacy Integration and Automatic Updates: It supports maintaining the context of information by retrieving and managing only the specific data required for practical AI agents from legacy systems.

- Provision of Task-Specific Data Bundles: Information required for specific tasks is grouped into “data bundles” and supplied to the AI Agent Hub to be linked to each workflow step.

[AI Data Lake] Data Bundle Management Screen

  5. User-Centric Integrated AI Agent Platform Service

MariNote.AI is the final integrated workspace that connects the advanced engine (AI Agent Hub) and refined data (AI Data Lake) to user’s actual work environments.

- Flexible Selection and Deployment of Specialized AI Agents: Users can identify and invoke agents optimized for specific tasks through the integrated panel.

- Autonomous Task Execution: Once invoked, the AI agent performs tasks according to the predefined workflow.

- Practical Results: AI agents go beyond providing answers to generate tangible outputs by combining linked tools and knowledge.

  6. Conclusion and Future Outlook

MariNote.AI is an integrated AI agent platform that seamlessly embeds the process design capabilities of the AI Agent Hub and the specialized knowledge framework of the AI Data Lake into the practical work environment. This structure, in which AI dynamically combines the most suitable tools and data bundles at every workflow step, is driving substantial innovation in KR’s business operations.

Moving forward, KR plans to further advance integration with legacy systems and continuously expand the scope of the MCP & Task Tool ecosystem, as well as data bundle support, to accelerate the AI-driven transformation of the shipbuilding and maritime industry.