What AI Agents Should Manufacturing Companies Pursue?
Session Overview
Manufacturing PMs lose sleep over the same questions: Which AI agents will actually help us win in the near future? What separates a good agent from a bad one? How do we prepare data, evaluate shipped products, and build the right “alliance” across teams and vendors?
Drawing on real deployments from semiconductor infrastructure, this talk outlines the strategic choices manufacturers must make to survive—and lead—in the age of agentic AI. You’ll see what works (and what doesn’t) when moving from pilots to production on the factory floor.
Key Topics:
- The legacy of DX—and what it means for AX
- The “internal data” nightmare and the reality of scaling laws
- Good vs. bad AI agents: design principles and failure patterns
- “Benchmark is dead”: toward task-grounded, operations-centered evaluation
- Do agents only live online? Bringing agents to physical operations and OT systems
Speaker

Moon Chan-hee plans and develops agentic AI for Samsung’s semiconductor infrastructure. He has helped shape and execute division-wide strategies from DX to AX, and leads vertical AI adoption with a holistic view that connects the shop floor with AI systems.