[Edaily Reporter NA EUN-KYUNG ] HanmiPharm(128940)Hanmi Pharmaceutical is partnering with domestic artificial intelligence (AI) companies to build its own infrastructure for the “eppiday” digital healthcare platform’s chatbot system. The company plans to transition its AI response generation capabilities—which previously relied on overseas cloud services—to a system based on domestic large language models (LLMs) and AI semiconductors.

HanmiPharm announced on the 2nd that it signed a “Strategic Memorandum of Understanding (MOU) for the Transition of the eppiday Chatbot to a Korean-Style LLM and the Establishment of Domestic AI Infrastructure” on the 30th of last month at the HanmiPharm C&C Square Multi-Hall in Songpa-gu, Seoul, with AI semiconductor company FuriosaAI, AI solutions specialist Xperix, and Xperix’s healthcare subsidiary Bagel Labs.

This agreement was established to transition the “AI Health Coaching” response-generation feature of the Epiday app from its existing overseas cloud environment to a domestic infrastructure based on in-house servers, including large language models (LLMs) and AI semiconductor technology. Through this, HanmiPharm plans to strengthen its control over the processing of patient health information and stably manage costs and operational risks associated with future service expansion.

Epiday is a digital healthcare service that supports personalized health management based on medical questionnaires and health information entered by patients. It allows users to continuously track and manage daily data such as weight, meals, and exercise, and supports lifestyle improvements during obesity treatment through AI health coaching.

HanmiPharm launched the Epiday app late last month, coinciding with the start of domestic prescriptions for “Epe” (generic name: efeglenatide), a Korean-developed obesity treatment. In addition to Epe, patients using other obesity treatments—such as Wegovy and Mounjaro—can also verify their medications and manage their dosing schedules through the app. The app also offers a feature that allows users to share the health information they have recorded with their healthcare providers.

Obesity is a condition in which not only weight loss through medication but also diet and exercise management, as well as weight maintenance after weight loss, are crucial. HanmiPharm’s strategy is to use Epiday to manage patients’ lifestyle habits and medication history even outside the doctor’s office, thereby increasing long-term treatment adherence.

Building on this agreement, HanmiPharm plans to first transition the core feature of Epiday’s AI coaching—response generation—to a Korean-developed large language model (LLM). The company also intends to gradually replace existing features currently in service—such as medication and dietary information searches and meal photo recognition—with domestic AI technology.

This transition is part of the effort to build what is known as “Sovereign AI.” Sovereign AI is a concept that reduces dependence on external technologies by operating data, AI models, and computing infrastructure within a nation’s own technological and management systems. For digital healthcare services that handle sensitive data—such as health information—it is crucial to have direct control over where data is processed and who has access to it.

In this collaboration, the four companies have divided roles according to their respective areas of expertise. HanmiPharm, as the service operator, will set the overall direction and priorities for the collaboration, while Bagel Labs will be responsible for developing the Epiday app and chatbot, replacing the AI model, and verifying the quality of responses.

Xperix will build and operate a domestically developed AI server and provide a domestic LLM inference environment. FuriosaAI will handle the supply of domestically developed AI semiconductors and provide technical cooperation for model optimization. AI inference is the process by which a trained model analyzes a user’s question and generates a response; response speed and operating costs can vary depending on semiconductor performance and server operation methods.

The four companies have completed verification in a test environment to determine whether the AI coaching response generation function can be migrated to their own domestic servers. They are also comparing and verifying the accuracy and response speed of answers before and after the migration using standard test books and an automated grading system. The timing and specific scope of the actual service implementation will be finalized following quality and safety verification and consultations among the four companies.

HanmiPharm expects this collaboration to reduce its reliance on overseas cloud services and strengthen domestic control over patient health information. Another goal is to increase the predictability of infrastructure operating costs and ensure service stability in response to growing user numbers.

The four companies plan to share the status of the phased transition and quality verification results through regular working-level meetings. Starting with the Epiday chatbot, they intend to expand the scope of domestic AI technology applications in the digital healthcare sector.

Kim Na-young, Vice President of the Innovation and Growth Division at HanmiPharm, stated, “The competitiveness of digital healthcare lies not in the adoption of technology itself, but in providing patients with consistent quality and tangible benefits.” She added, “Based on the expertise of our partners, we will thoroughly verify the response quality and operational stability of the Epiday chatbot and expand it into reliable, personalized health management features.”

Hwang Sang-yeon, CEO of HanmiPharm, remarked, “This agreement is significant in that it lays the groundwork for processing patient health information more securely and reliably by transitioning Epiday’s core chatbot functionality to a domestically developed sovereign AI environment.” He added, “We will combine our pharmaceutical and biotech capabilities with digital and AI technologies to create digital healthcare services that patients and medical professionals can trust.”