Hanmi Pharmaceutical to Fully Deploy AI Across Quality Assurance, Training on 4,000 SOPs
核心洞察
Hanmi Pharmaceutical has been selected as the final beneficiary of South Korea's "AX One-Stop Voucher Support Program" to integrate AI into quality assurance operations.
The two-year "Hanmi Q-Agent" project will train a proprietary language model on approximately 4,000 SOPs at the Pyeongtaek Bio Plant, with first-year costs of roughly 1.7 billion won.
The AI system will support quality assurance for "Epe (搜索)," a Korea-customized obesity (搜索) treatment slated for launch in the second half of the year.
Hanmi Pharmaceutical is moving to fully integrate artificial intelligence (AI) into its quality assurance (QA) operations across its pharmaceutical and bio business, aiming to automate the search, analysis, and drafting of vast volumes of quality documentation and eliminate inefficiencies inherent in manual processes.
The company announced on the 7th that it has been selected as a final beneficiary in the pharmaceutical and bio category of the "AX (AI Transformation) One-Stop Voucher Support Program," administered by South Korea's Ministry of Science and ICT, with the National IT Industry Promotion Agency (NIPA) serving as the dedicated implementing body.
A Domestic Consortium for the "Hanmi Q-Agent" Project
The project, named "Hanmi Q-Agent," will be executed through a consortium composed entirely of domestic infrastructure providers. Hanmi Pharmaceutical will oversee overall project management and handle SOP (Standard Operating Procedure) integration, quality documentation, IT governance, information security, and linkage with core systems including MES, LIMS, and QMS. Samsung SDS (搜索) will be responsible for building the AX infrastructure, while MegazoneCloud (搜索) will handle AI solution implementation and tuning. SelectStar (搜索) will perform third-party verification through dataset construction and AI reliability assessment.
The project kickoff ceremony was held on the 28th of last month at Hanmi C&C Square in Songpa-gu, Seoul, with key representatives from all four companies in attendance.
Two-Year Initiative Centered on the Pyeongtaek Bio Plant
The project will use the Pyeongtaek Bio Plant as its forward operating base, running over a two-year period through December of next year. Work commenced in July, with first-year project costs totaling approximately 1.7 billion won (approximately $1.3 million), of which roughly 1 billion won (approximately $740,000) will be covered by government subsidies.
In the first year, the company will train AI on approximately 4,000 SOPs at the Pyeongtaek facility, applying the system to document retrieval and regulatory gap analysis. In the second year, Hanmi plans to build a system that automatically drafts section-by-section APQR (Annual Product Quality Review) report outlines based on the trained data.
In the pharmaceutical and bio industry, documenting proof that products were manufactured according to regulations is as critical as the manufacturing itself. However, the current structure—where vast volumes of semi-structured quality documents are manually searched, compared, and drafted—has been constrained by inefficiency and reactive, after-the-fact responses. Through this initiative, Hanmi Pharmaceutical aims to transition manual quality documentation workflows to an AI-collaborative process, enabling employees to focus on higher-value activities such as new drug development and global business operations.
A Proprietary Language Model Supporting Epe Quality Assurance
A key aspect of the project is avoiding off-the-shelf general-purpose AI. Hanmi will build a proprietary language model fine-tuned for its specific operations, based on Upstage (搜索)'s "Solar" model. This fine-tuned model will be deployed across Hanmi Pharmaceutical's internal quality enhancement efforts and quality assurance operations for innovative new drugs.
In particular, the company plans to apply the model to QA operations for "Epe (搜索)," a Korea-customized obesity (搜索) treatment slated for launch in the second half of the year. With Epe preparing for full-scale production at the Pyeongtaek facility, the AI-based quality assurance system is expected to provide support from the earliest stages of manufacturing.
Once validated at Pyeongtaek, the model will be expanded to the Paltan facility and R&D center. Beyond QA, Hanmi plans to extend application across QC (quality control), R&D (research and clinical), RA·PV, and other enterprise-wide domains. The rollout will also extend across the group's affiliates, including Hanmi Science, Hanmi Fine Chemical, and Beijing Hanmi Pharmaceutical.
Hwang Sang-yeon, CEO of Hanmi Pharmaceutical, said, "The QA field, which handles vast volumes of quality documentation, has been constrained by manual processes and reactive responses. Through Hanmi Q-Agent, we will support Epe (搜索)'s quality assurance operations and expand the AI innovation model across the entire group." He added, "Starting with Pyeongtaek, we will successfully scale this AI innovation model across all group companies to further elevate the global export competitiveness of K-pharma."