en.Wedoany.com Reported - South Korea's MIDAS Group will introduce company-wide artificial intelligence agents capable of understanding business context and autonomously planning and executing tasks. The group plans to launch "inAX," an AI office platform connecting internal data with business relationships, in October this year, followed by specialized agents for global customer sales, technical support, and training by the end of the year.

Shin Mi-young, CEO of MIDAS In, stated at the "2026 MIDAS Week, Honors Day" held at the DoubleTree by Hilton Hotel in Pangyo on the 16th, "Now, AI agents handle the work, and humans make the judgments. The future where humans are not replaced by AI but have their capabilities enhanced through AI is exactly 'The MAX' that MIDAS Group envisions." MIDAS Week, held from the 13th to the 17th, is an event where MIDAS Group members from home and abroad share business achievements and visions. Honors Day is a core program for sharing future visions across the entire group. This year, approximately 240 people participated, including 111 members from overseas subsidiaries in nine countries such as China, Japan, India, Russia, the United States, Europe, and Australia, along with leaders and key talent from the Pangyo headquarters.

The core of the AI Transformation (AX) promoted by MIDAS Group is inAX, which redesigns the group's office operation system around AI agents. Unlike existing generative AI that remains at the level of answering user questions or automating individual tasks, inAX proactively discovers and proposes tasks that need to be executed by connecting organizational goals with people, business, customers, and financial data. inAX consists of three agents: Guide Agent (GA), Operation Agent (OA), and System Agent (SA). The GA communicates directly with users, acting as a guide for business direction; the OA detects data changes and proactively suggests next steps; the SA verifies business execution and records, connects relationships between information, and supports continuous system learning. Named "Polar," meaning the North Star, the GA can analyze customer data, contract status, product usage, and meeting records to prioritize customers requiring attention and propose sales strategies. During video conferences, it analyzes customer statements and existing relationships in real-time, guides response strategies, and automatically generates meeting minutes, follow-up emails, and draft quotations after the meeting.

Polar also provides organizational operation functions for team leaders, allowing them to compare sales targets with progress, identify bottlenecks, and recommend necessary intervention strategies. Additionally, it can accumulate records of plans, execution, checks, and reviews performed by members as organizational assets, enabling leaders' experiences to be automatically reflected in subsequent work. Kim Byeong-seok, CP of the Product Strategy Office at MIDAS In, stated, "To make AI truly useful, it requires not only understanding the on-site context but also deep insight into the nature and attributes of the work. Instead of waiting for users to ask questions before responding, the team created a system that predicts outcomes and acts preemptively, which led to the launch of inAX."

The foundation of inAX is a "knowledge graph" that represents the entire enterprise as a relational map. It defines organizational goals, members, business, meeting records, decision records, customer responses, and financial information in an AI-understandable format and interconnects them. Since all agents share the same knowledge graph, they can share the same context even across different business domains. The GA delivers the required answers to users based on the relationships and context accumulated in the knowledge graph; the OA detects changes in data and relationships and proposes next steps; the SA verifies the data and records generated in actual business, semanticizes them, and stores them back into the knowledge graph. As agents are used, data and relationships continuously accumulate, improving the accuracy of answers and predictions.

MIDAS Group emphasizes that AI agents must not only provide accurate answers but also be capable of continuous interaction with humans. To this end, the group developed its own quality standard, the "Samanda Index," which evaluates agents across 27 items, including whether they understand emotional context, remember past conversations, and connect dialogues to subsequent judgments and actions.

Furthermore, inAX incorporates the "Zain Engine," which technologizes MIDAS Group's management philosophy of understanding people and organizations through relationships. While the knowledge graph provides the company's reality and business context, the Zain Engine offers a perspective for interpreting human and organizational relationships, enabling agents to make more human-like and reasonable judgments. Ha Seong-woo, TF Leader of the AX Foundation Development Team at MIDAS In, stated, "AI agents will become extremely numerous and large-scale in the future. The differentiating point of MIDAS agents is that they have a compass guiding the direction, and the end of that compass is people. As the system grows, it will become more human-like and smarter, and the system's growth will, in turn, drive human growth."

MIDAS IT, which provides construction engineering software technology, plans to apply inAX technology to its global technical support operations. The technical support agent currently under development will assist domestic and international members and customers in three areas: sales support, user technical support, and training support. The sales support agent can generate product features, competitor comparison materials, technical documents, and presentation content needed for customer meetings, and provide real-time answer suggestions during meetings. The technical support agent analyzes the design models and input data used by customers, identifies error causes, and guides correction methods. The training agent can demonstrate software installation, activation, and project-level modeling and design processes step-by-step, and provide a summary report after the training session. Lee Hye-yeon, Team Leader of the Technology Planning Team at MIDAS IT, stated, "The most important thing for the technical support agent is to accurately understand the product. The team built a knowledge graph of the product's numerous functions and the purpose and relationships of each function, enabling the agent to understand customer errors, problems, and business context. Based on daily accumulated issues and feedback, the agent is designed for continuous improvement."

MIDAS Group will release the initial version of inAX to its group members in October this year. Currently, CEO Shin and some other members are piloting inAX to verify its functions, such as proactively suggesting business priorities and necessary actions. The MIDAS IT technical support agent will be launched at the end of December this year, followed by plans to sequentially develop specialized AI agents that work alongside members in various domains such as business, products, HR, and finance. Shin Mi-young, CEO of MIDAS In, emphasized, "By next year, we will build an environment where numerous agents work together with members, realizing the future of 'The MAX' where agents handle the work and humans provide judgment and support."