AI drug development is only half the battle without smart factories

As the government pushes policies to strengthen the global competitiveness of the pharmaceutical and biotech industry, AI-driven drug development has emerged as a defining challenge for the sector. With the efficiency of AI in identifying new drug candidates from vast datasets now well established, calls are growing for data-based innovation to extend beyond the research stage and into actual manufacturing and quality control.

The government's recently announced second-half 2026 economic growth strategy identified the establishment of an AI drug development roadmap and a data foundation as key policy priorities to propel the biotech and healthcare industry forward. Experts say that for AI drug development to bear fruit, research data and manufacturing floor data must be connected without interruption — making the digital transformation of production processes essential. In the commercialization of new drugs particularly, regulatory compliance hinges on manufacturing data integrity, and resolving data fragmentation has become an urgent industry-wide challenge.

Longstanding problems on the manufacturing floor, however — siloed data between systems and fragmented management structures — continue to hold the industry back. Despite the requirement to comply around the clock with Good Manufacturing Practice (GMP) standards, manufacturing execution systems (MES), laboratory information management systems (LIMS), and warehouse management systems (WMS) have typically operated in isolation, making smooth data linkage difficult. Manual data entry and fragmented systems leave room for arbitrary data modification or manipulation, and become an obstacle to demonstrating process compliance when AI-developed drugs move into commercial production.

Against this backdrop, Daewoong Pharma's Osong smart factory has drawn attention as a model for the manufacturing innovation the AI drug development roadmap envisions. The Osong facility has built what it calls a "structural quality assurance model" by fully integrating its WMS, MES, LIMS, and electronic document management system (EDMS).

At the Osong factory, every process from raw material intake to final packaging runs on data integrity controls. When chemical raw materials arrive, they undergo full non-destructive Raman spectroscopy inspection — which uses a laser to verify ingredients without opening the packaging — and test results are automatically registered in LIMS as PDF files that operators cannot modify. In the weighing process, scales link in real time to the plant's computer systems, and if a weight discrepancy occurs, a "lock" mechanism automatically halts the process.

Vision cameras detect micro-defects as small as 50 micrometers, while laser-guided vehicles (LGVs) and sealed intermediate bulk containers (IBCs) eliminate human contact with materials, cutting off the possibility of human error and contamination at the source.

This systematic completeness extends to the production of Daewoong Pharma's own drug, Enblotab. Despite each tablet containing just 0.3 milligrams of active ingredient, every measurement from weighing through in-process control (IPC) is recorded as real-time data, and the entire production line automatically shuts down if any value falls outside the permitted tolerance.

Daewoong Pharma developed the system in-house through IDS&Trust, an IT affiliate within its group, maximizing its capacity for maintenance and process adaptation. Quality personnel account for 42 percent of total factory staff — well above the 24 to 40 percent range typical of major domestic pharmaceutical companies.

Significant challenges remain, however, before this kind of manufacturing data integration can deliver results across the industry. The cost of building large-scale IT infrastructure, compatibility issues with existing equipment, and varying skill levels among floor workers are practical barriers that individual companies must overcome on their own. Experts advise that beyond technical completeness in data linkage, customized design suited to each pharmaceutical company's product characteristics and rigorous security verification must go hand in hand.

"For the pace of AI drug development to translate into higher success rates in actual product manufacturing, efforts to systematize and integrate manufacturing data — as seen at Daewoong Pharma's Osong smart factory — need to spread across the industry," an industry official said. "For this kind of floor-level innovation to take root as genuine competitiveness, institutional support for the industry's digital transformation, alongside technical improvements, is essential."