In this Beyond PLM article, author Oleg Shilovitsky argues that the phrase AI-native PLM, much like earlier terms such as “true cloud,” is increasingly used as marketing shorthand, often with ambiguous meaning. He warns that buyers should be wary: the label promises transformed workflows, but what lies beneath may still be classic PLM systems with a veneer of AI tools.
Shilovitsky notes the shift from deterministic to probabilistic computation as the real turning point. Traditional PLM relies on exact behavior: querying “Part A, Revision 3” returns the same record every time. AI introduces uncertainty, responses can vary, meaning the system must balance trust and flexibility.
He outlines what credible AI value looks like in PLM:
- Reasoning over both structured and unstructured data to extract insights;
- Automating classification and enrichment of metadata;
- Conversational interfaces to query product history;
- Agentic workflows that can trigger checks, supplier queries, or compliance audits without manual orchestration.
The article describes how true AI-native systems will likely layer probabilistic reasoning atop robust governance frameworks. Relational databases still serve critical roles; newer architectures may use graphs or vector databases to represent product knowledge intelligently.
Shilovitsky’s conclusion is pragmatic: the label “AI-native PLM” is less important than the capabilities behind it. He encourages engineering leaders to focus on outcomes, improved decision support, traceability, and data reuse, rather than buzzwords.