Hanmi Pharm Signs Epday AI Infrastructure Deal

Hanmi Pharm signed a strategic memorandum of understanding with FuriosaAI, Xperix and Bagel Labs to move the Epday chatbot to a Korean large language model (LLM) and build domestic AI infrastructure. Epday is a digital healthcare service that supports personalized health management based on patients’ health information and medical questionnaires. It lets users track daily data such as weight, meals and exercise, and provides AI health coaching to support care, including obesity treatment. Hanmi Pharm said the service uses accumulated health information to provide tailored guidance and encourage healthy habits.

Under the agreement, Hanmi Pharm will lead the partnership and set its direction and priorities. Bagel Labs will develop the Epday app and chatbot, replace the AI model and check response quality. Xperix will provide and operate domestic AI servers and an inference environment for a Korean LLM. FuriosaAI will supply domestic AI chips and work with partners on model optimization.

The four companies have completed testing to determine whether AI coaching responses can be generated on their own domestic servers. They are comparing response accuracy and speed before and after the transition using a standard question set and automated scoring. The timing and scope of commercial deployment will be decided after quality and safety checks and further discussions among the four companies.

The agreement aims to shift AI health-coaching responses from overseas cloud services to a Korean LLM and AI chips on domestic infrastructure. Hanmi Pharm expects this to strengthen systems that handle patient health information and help manage costs and operational risks as the service expands. The company also expects lower reliance on overseas cloud services, greater domestic control over health information, more predictable infrastructure costs as users increase and stable operations.

Hanmi Pharm plans to move response generation to a Korean LLM first, then gradually replace the service’s existing medicine and dietary information search and meal-photo recognition functions. CEO Sang-yeon Hwang said a domestic technology-based “sovereign AI” environment would provide a basis for handling patient health information more safely and reliably. Nayoung Kim, vice president of the Innovation and Growth Division, said the partners would use their expertise to verify chatbot response quality and operational stability, then expand the service into personalized health management.

Hanmi Pharm signed a strategic memorandum of understanding with FuriosaAI, Xperix and Bagel Labs to move the Epday chatbot to a Korean large language model (LLM) and build domestic AI infrastructure. Epday is a digital healthcare service that supports personalized health management based on patients’ health information and medical questionnaires. It lets users track daily data such as weight, meals and exercise, and provides AI health coaching to support care, including obesity treatment. Hanmi Pharm said the service uses accumulated health information to provide tailored guidance and encourage healthy habits.

Under the agreement, Hanmi Pharm will lead the partnership and set its direction and priorities. Bagel Labs will develop the Epday app and chatbot, replace the AI model and check response quality. Xperix will provide and operate domestic AI servers and an inference environment for a Korean LLM. FuriosaAI will supply domestic AI chips and work with partners on model optimization.

The four companies have completed testing to determine whether AI coaching responses can be generated on their own domestic servers. They are comparing response accuracy and speed before and after the transition using a standard question set and automated scoring. The timing and scope of commercial deployment will be decided after quality and safety checks and further discussions among the four companies.

The agreement aims to shift AI health-coaching responses from overseas cloud services to a Korean LLM and AI chips on domestic infrastructure. Hanmi Pharm expects this to strengthen systems that handle patient health information and help manage costs and operational risks as the service expands. The company also expects lower reliance on overseas cloud services, greater domestic control over health information, more predictable infrastructure costs as users increase and stable operations.

Hanmi Pharm plans to move response generation to a Korean LLM first, then gradually replace the service’s existing medicine and dietary information search and meal-photo recognition functions. CEO Sang-yeon Hwang said a domestic technology-based “sovereign AI” environment would provide a basis for handling patient health information more safely and reliably. Nayoung Kim, vice president of the Innovation and Growth Division, said the partners would use their expertise to verify chatbot response quality and operational stability, then expand the service into personalized health management.