From Data to Decision How AI Agents and PLM Work Together

Related Vendor

Wuxi InfiMotion Technology Co., Ltd.

Artificial intelligence is fundamentally reshaping industrial software applications. Instead of isolated automation, we are now seeing the emergence of increasingly networked, context-sensitive agent systems.

According to Gartner, by 2028 around one-third of all enterprise applications will feature agent-based AI capabilities. Today, that figure is less than one percent. Analysts also predict that, in the medium term, approximately 15 percent of routine business decisions will be prepared, or even made autonomously, by AI agents, particularly in data-intensive, structured work environments.

In technical development areas where information must be managed, connected, and assessed over long product lifecycles, the key question becomes: What role can AI agents play in the context of existing PLM systems? And what structural prerequisites are required to integrate their capabilities effectively?

Bridging Data Management and Decision Support

Product Lifecycle Management (PLM) is well-established across many industries. It forms the technological backbone for managing engineering product data: from CAD models and bills of materials to change documentation, requirements, and approvals. The role of PLM systems is clearly defined: they ensure traceability, consistency, and process integrity across the entire product lifecycle.

However, as system complexity increases, development cycles shorten, and customization pressures mount, established PLM structures can be enhanced intelligently. This is particularly relevant in industries with highly modular or software-defined products, such as machinery, medical devices, or automotive manufacturing. Here, demands for speed, variant flexibility, and data consistency are growing. This is precisely where the Agentic AI approach comes into play.

How Agentic AI Differs from Traditional Automation

Agentic AI refers to the ability of AI systems not only to execute rule-based tasks but also to interpret them independently, put them into context, and proactively derive actions. The concept combines three defining capabilities:

• Context sensitivity: AI agents analyze data while considering dependencies, system boundaries, and priorities.
• Learning capability: With each task performed, the system becomes more accurate through feedback loops and structured knowledge management.
• Interoperability: AI agents operate across system boundaries, linking data and processes between different applications.

In PLM-supported development processes, this could mean that an AI agent analyzes a change request, identifies the affected objects, assesses variant dependencies, and considers technical as well as regulatory implications, before a human makes the final decision. The result: measurable workload reduction in operational development.

Data Access, Semantics, and Integration: The Key Requirements

To make AI agents work effectively within a PLM system, certain technological prerequisites must be in place. Three factors are essential:

• Vector-based data models: These enable the inclusion of unstructured content, such as free-text fields, requirements documents, or test reports, into systematic analyses. AI agents access machine-readable representations to link and interpret information.
• Semantic interfaces: Translating natural-language user queries into structured database queries is a crucial step. For example, “Which open change requests affect component X?” would automatically be translated into a combination of object reference, filter logic, and process status.
• Open APIs: These interfaces allow AI agents not only to read information but also to trigger processes, such as automatically generating approval proposals, updating linked objects, or forwarding data to downstream systems.

Together, these elements create an environment where technical decisions can be prepared based on consistent data, aligned across systems, and anchored in a coherent Digital Thread, also known as Intelligent Product Lifecycle.

Practical Use Cases

The benefits of AI agents are most evident in areas with high variant diversity, frequent changes, and extensive documentation requirements. Three examples:

• Digital traceability: The AI agent automatically detects changes to requirements, engineering data, or system models. It analyzes the impact, identifies relevant objects, and suggests possible follow-up actions.
• Product Line Engineering (PLE): Based on natural language, AI agents create initial system models, standardize components, and identify redundancies. In the event of changes, they dynamically adapt configurations - always within the defined variant space.
• Variant analysis: In technical system architectures with high design flexibility, AI agents simulate the technical feasibility of certain configurations, calculate resulting unit costs, and assess potential compliance risks. The results are prepared for decision-making.

Organizational Implementation

Integrating Agentic AI into existing development environments is not purely a technical project, it also changes roles, workflows, and decision-making patterns. Step-by-step introduction is therefore essential, ideally starting with clearly defined pilot projects supported by training measures and interdisciplinary guidance.

Transparency is key: stakeholders are more likely to embrace AI agents as a support tool if they understand how recommendations are generated, rather than perceiving them as a black box. Coordination with existing roles in engineering, variant management, and system architecture should take place early to establish clear interfaces and ensure smooth collaboration.

Agentic AI is not a replacement for existing systems, it is a functional enhancement. It augments PLM environments where decision processes are structured but resource-intensive. The technology for gradual implementation is already available. Combined with a consistent Digital Thread and a willingness to embrace change, a new form of development support can emerge

*Michele Del Mondo is currently Global Advisor Automotive at PTC.