This Beyond PLM article argues that the flurry of announcements by major PLM vendors about AI, copilots, and virtual assistants may not lead to the future of agent-native PLM.
Vendors such as Dassault Systèmes, Siemens, PTC, and Aras are embedding AI features inside their existing platforms. For example: Dassault frames AI as part of a “Generative Economy” with virtual twins and assistants; Siemens adds copilots to help navigate BOMs, requirements, and documentation; PTC uses an agentic-AI narrative across PLM, ALM, CAD, and service; Aras plugs conversational AI into its low-code Innovator SaaS.
The critique: these moves mostly improve usability inside the confines of legacy architectures, but they do not rebuild PLM from the ground up to support truly autonomous, reasoning-capable agents. The author contends that transformative “agentic PLM” requires a fundamentally different data and system architecture, one that supports persistent product memory, semantic context, cross-tool interoperability, and event-driven reasoning.
Without such a foundation, current AI additions deliver only incremental gains, such as better search, smarter document retrieval, and easier navigation, not a radical rethinking of product-development workflows. The promise of agents that can reason over design history, simulate “what-if” scenarios, or coordinate across disparate tools remains unfulfilled.
The article warns that labeling these enhancements as “agentic AI” risks overpromising. Instead of chasing hype, the industry needs to redesign PLM data infrastructure around flexible, semantically rich knowledge graphs and event-aware architectures before expecting meaningful AI-driven automation.