Software-Defined Manufacturing (SDM) is a production paradigm where the control logic, process orchestration, and operational intelligence are decoupled from specific, proprietary hardware and executed in a virtualized, software-centric environment. In an SDM architecture, the capabilities of the factory are defined by code rather than by fixed mechanical constraints or hard-wired circuits. This allows for rapid reconfiguration, over-the-air updates, and the continuous optimization of manufacturing processes via AI agents.
SDM is the necessary environment for Agentic AI and Physical Intelligence to flourish. It draws parallels to "Software-Defined Networking" (SDN) or "Software-Defined Vehicles." In a traditional factory, changing a production line often requires rewiring cabinets, moving conveyors, and reprogramming PLCs one by one. In SDM, the hardware is generic and flexible; the "personality" of the line is loaded via software.
Masson's research positions Agentic AI as the "brain" and SDM as the "body" infrastructure that responds to the brain's commands. This decoupling is essential for agility. It allows the "brain" to be upgraded (e.g., deploying a smarter AI model) without replacing the "body" (the robots), and vice versa. It transforms manufacturing from a CAPEX-heavy, rigid discipline into an OPEX-driven, agile discipline capable of "Mass Customization" and rapid response to market demand.