"With current technology, only 30% of multi-omics (Multi-Omics) data can be interpreted by people, even with an optimistic view. The data structure is complex, and Korea lacks specialized talent. We can break through these limits with artificial intelligence (AI) and cloud automation."
Multi-omics, which integrates analysis of the genome, transcriptome, and proteome, is considered a key pillar of new drug development and personalized precision medicine. But the data are complex, making integrated analysis difficult, and as AI technology advances, the industry's approach is changing.
Lee Nam-yong, head of Cellkey AI, gave a lecture on "Redesigning the bio R&D workflow with AI" at the "2026 SME AX (AI transition) Leaders Forum" hosted by ChosunBiz at the Westin Josun Hotel in Jung-gu, Seoul, on the 9th.
Founded in 2021, Cellkey AI is a bio technology company building an AI-based precision medicine platform by combining AI and cloud automation technologies. It holds algorithm technology that can efficiently analyze more than 1 million proteins and glycoproteins in the human body. It is currently promoting joint research and business through open innovation with global research institutes and large corporations in the United States, Japan, the United Kingdom, the Middle East, and Germany.
Lee pointed to data fragmentation as the biggest problem in today's bio research field.
Research institutions typically commission analyses to different vendors, receive results by email, and save them on individual researchers' PCs. When a researcher leaves, data utilization is cut off, and standard data that could be used for AI training are not accumulated.
The company developed a multi-omics platform that combines AI and cloud automation to solve these limits. It is the agent AI-based bio R&D platform "Omics Farm." When a researcher requests analysis on the platform, it connects with partner research institutions to produce data, standardizes and assets them, and uses them for AI training.
Lee said, "Omics Farm is an automation platform that covers the entire process from generation, analysis, learning, and interpretation of multi-omics data to automatic report generation," adding, "Our goal is to build a proprietary 'Sovereign Multi-Omics Foundation Model' based on the accumulated data."
The company is also developing a Deep Learning model that integrates genome, transcriptome, and protein data to predict biomarkers and new drug candidates. It also built a bio-specialized vertical AI agent, "BioEOS," that automatically analyzes papers and patents.
AI is driving process innovation in biopharmaceutical CDMO (contract development and manufacturing). Bio manufacturing is cited as a representative "green field (Green Field)" where digital transformation is slow because it remains with traditional manufacturing methods.
Cellkey AI moved away from the conventional cell development method of randomly finding gene insertion sites, enabling an AI agent to collect and analyze a vast number of papers and patents to identify optimal "hotspot (Hotspot)" regions.
In the purification process that removes impurities, instead of repeating physical experiments dozens of times, it derives optimal experimental conditions in a simulation environment using an AI-based "surrogate model (Surrogate Model)."
Lee said, "With this approach, we shortened purification processes that used to take months to a few weeks and achieved a sharp reduction in expense by 70% to 80%."
An innovation case in the cell analysis process for animal alternative experiments was also shared. Previously, researchers faced a bottleneck of about three weeks as they checked each cell's growth and ratio by looking through an electron microscope.
AI is also applied in reading cell images. He said, "It used to take about three weeks for a researcher to check 96 cell images one by one with an electron microscope, but we enabled an AI agent to analyze the images and automatically generate a quantified report," noting, "We achieved more than a 3,000% improvement in work efficiency."
Lee cited "data," "automation," and "physical AI (Physical AI)" as the three pillars of future bio innovation.
He said, "Automation experiments with robots produce high-quality data, and that data trains AI to efficiently redesign the next experiment. We must build a 'virtuous cycle in which data grows data' to secure a lead in global competitiveness."
An automated research facility of the Chinese AI new drug development corporation Insilico Medicine was introduced as a case. Lee said, "In the near future, the fusion of physical AI and bio will become a reality in Korea as well."