AI in PLM: How MCP Can Turn AI Into an Engineering Agent
Artificial intelligence in Product Lifecycle Management is entering a new phase.
The first wave of enterprise AI focused heavily on generating answers, summarizing documents, finding information, and helping users interact with complex systems through natural language.
Those capabilities are valuable, but they represent only part of the opportunity.
The next step for AI in PLM is the ability to work with connected product information, interact with engineering systems and tools, understand relationships between product objects, and support real engineering activities within governed processes.
In other words, AI needs to move from:
“Tell me about this product.”
to:
“Help me work with this product.”
Making that transition requires more than a powerful language model.
For AI to become genuinely useful across the product lifecycle, it needs four foundational capabilities:
- Context
- Ontology
- Tools
- Governance
Model Context Protocol, or MCP, creates an interesting foundation for connecting these capabilities.
Combined with a connected Digital Thread and an AI engineering agent such as Nora Prima, MCP can help create a new way for engineering teams to interact with product information and enterprise systems.
Why AI in PLM Needs More Than Better Answers
A general-purpose AI model can reason over the information provided to it.
But engineering decisions rarely depend on a single document.
Consider a relatively straightforward question:
“Which released components could be affected by this requirement change?”
Answering it correctly may require understanding:
- the requirement itself
- the product structure
- associated designs
- affected BOM components
- component revisions
- released configurations
- engineering changes
- manufacturing information
- quality history
- supplier relationships
- workflow status
Some of this information may exist inside PLM.
Other information may reside in CAD, ERP, MES, QMS, requirements management systems, supplier platforms, document repositories, or other engineering applications.
The challenge therefore isn’t simply giving AI access to more documents.
The challenge is giving AI access to connected product context.
The Digital Thread Provides the Context AI Needs
The Digital Thread connects information across the lifecycle of a product.
Rather than treating requirements, CAD models, BOMs, changes, documents, manufacturing information, and quality records as isolated objects, the Digital Thread preserves relationships between them.
For example:
Requirement → Design → BOM → Change → Manufacturing Process
A requirement can be satisfied by a design.
A design can be represented by parts and assemblies.
Those components can form part of an engineering BOM.
A change can affect particular components or configurations.
Those components may ultimately be used within manufacturing processes.
When these relationships are connected, AI can reason over something much closer to the actual structure of engineering work.
This is significantly more useful than simply searching a collection of files.
Four Foundations for Practical AI in PLM
1. Context
Context gives AI the information surrounding an engineering question.
Within PLM, context may include:
- products
- parts
- BOMs
- requirements
- CAD models
- documents
- revisions
- configurations
- engineering changes
- manufacturing information
- suppliers
- quality records
- project information
- workflows
- approvals
- relationships between these objects
Without this context, an AI system may understand the words contained in an engineering document without understanding where that document sits within the lifecycle of the product.
Context allows the AI to understand the broader engineering environment.
2. Ontology
Context alone is not enough.
The system must also understand what the relationships between product objects mean.
This is where product ontology becomes important.
An ontology provides semantic meaning to the information within the Digital Thread.
For example, the system should be able to understand relationships such as:
- a requirement is satisfied by a design
- a design is represented by a component
- a component is part of a BOM
- a change affects a revision
- a revision belongs to a configuration
- a part is supplied by a supplier
- a manufacturing process uses a particular component
This semantic layer allows AI to reason over engineering relationships rather than simply matching keywords.
Imagine the difference between locating every document containing a component number and understanding that the component belongs to three assemblies, appears in two configurations, was modified through an engineering change, and is associated with a recurring quality issue.
That is the difference between information retrieval and engineering context.
3. Tools
Even context and reasoning do not make AI an active engineering participant.
For that, AI needs tools.
An AI engineering agent might need to:
- query a BOM
- inspect product structures
- retrieve CAD-related information
- compare revisions
- identify affected configurations
- search quality records
- retrieve supplier information
- review change history
- examine workflow status
- create structured findings
- initiate permitted actions within enterprise systems
This is where Model Context Protocol becomes particularly relevant.
What Is Model Context Protocol?
Model Context Protocol provides a standardized approach for connecting AI applications with external data sources and tools.
For PLM, the important idea is not simply that AI can retrieve additional information.
MCP can provide a consistent mechanism through which an AI agent can interact with the systems and capabilities required to complete a task.
Instead of building an isolated AI experience around one application, the agent can potentially work with authorized tools and connected enterprise information across the engineering environment.
That could include systems such as:
- CAD
- PLM
- ERP
- MES
- QMS
- requirements management
- supplier systems
- additional enterprise applications
The result is an AI environment capable of working across the Digital Thread rather than remaining limited to a single database or application.
From AI Assistant to AI Engineering Agent
This distinction matters.
An AI assistant generally waits for a question and generates an answer.
An AI engineering agent can potentially interpret a goal, gather relevant context, use permitted tools, evaluate connected information, and prepare an engineering outcome.
Consider several examples.
Impact Analysis
An engineer asks:
“Which released components could be affected by this requirement change?”
The AI agent could potentially:
- identify the requirement
- inspect its relationships with designs
- trace associated components
- determine which product structures use those components
- identify released configurations
- review related engineering changes
- summarize the potential impact
Instead of manually navigating several objects and systems, the engineer receives a structured starting point for impact analysis.
Similar-Part Analysis
Another request could be: “Find similar parts and compare their manufacturing and quality history.”
This requires significantly more than semantic similarity.
Useful analysis may involve:
- geometry
- classification
- part attributes
- manufacturing history
- quality events
- supplier information
- lifecycle state
- previous changes
An AI agent connected to the relevant systems could gather this context and prepare a comparison for engineering review.
Configuration Analysis
An engineer might ask: “Identify the configurations affected by this component revision.”
The agent would need to understand where the component is used, which revisions are applicable, which structures represent released configurations, and potentially which downstream processes depend on those configurations.
Again, this depends on connected relationships rather than isolated documents.
AI Must Work Across More Than PLM
Product knowledge rarely exists in one system.
Engineering organizations typically operate across multiple applications.
CAD
CAD provides detailed design and geometry information.
PLM
PLM manages product structures, lifecycle information, revisions, changes, documents, workflows, and related engineering information.
ERP
ERP may contain sourcing, procurement, planning, costing, and enterprise information.
MES
Manufacturing Execution Systems provide production and manufacturing context.
QMS
Quality Management Systems contain quality events, non-conformances, inspection information, and related records.
Supplier Systems
Supplier platforms contain external product, sourcing, qualification, and collaboration information.
AI becomes substantially more useful when it can access relevant knowledge across these systems while retaining the meaning and relationships surrounding the data.
This is where the combination of a Digital Thread, product ontology, and standardized tool connectivity becomes powerful.
Governance Is What Makes Enterprise AI Usable
Giving an AI agent access to engineering systems without governance would create significant risk.
Engineering information is controlled for a reason.
Different users have different permissions.
Product states determine what can be changed.
Certain activities require formal review.
Released information cannot simply be modified because an AI model recommends a change.
Practical AI in PLM therefore requires governance at every level.
An AI engineering agent should operate within:
- user permissions
- role-based access
- lifecycle rules
- workflow controls
- approval processes
- traceability requirements
- audit histories
- organizational policies
It should also distinguish between tasks it can perform automatically and tasks requiring human approval.
For example, AI might prepare the findings for an engineering change review.
That does not mean the AI should independently approve the change.
The role of AI should be to improve engineering productivity while preserving the controls required for product development.
Human-in-the-Loop AI for Engineering
This leads to another important principle: human-in-the-loop operation.
Many engineering processes involve decisions with significant downstream consequences.
Changing a component could affect:
- compliance
- manufacturing
- suppliers
- cost
- inventory
- serviceability
- product configurations
- quality
- released documentation
AI can help engineers understand those relationships faster.
It can identify information that deserves attention.
It can prepare comparisons.
It can summarize evidence.
It can potentially perform predefined actions.
But engineers, reviewers, and other responsible stakeholders should remain involved wherever formal decisions or approvals are required.
The goal should not be uncontrolled autonomy.
The goal should be governed engineering assistance.
What This Could Look Like With Nora Prima and Nora IPLM
This is an important direction for Nora IPLM, Nora Prima, and Nora IPLM MCP capabilities.
Nora Prima is positioned as an AI engineering agent working across connected product data, enterprise systems, and governed engineering processes.
The underlying model can be understood through four layers:
Context
Connected product and enterprise information provides the data needed for engineering reasoning.
Product Ontology
Relationships between requirements, designs, BOMs, changes, processes, and other objects preserve meaning.
Tools via MCP
AI gains controlled access to capabilities and data sources required to perform useful engineering work.
Governance
Permissions, processes, workflows, approvals, and traceability define what the agent can do and when human involvement is required.
Together, these elements can allow AI to participate much more meaningfully in the Digital Thread.
Example: AI-Assisted Engineering Change Analysis
Consider an engineering team preparing a component revision.
Traditionally, the engineer may need to inspect several sources before understanding the complete impact.
They may review:
- the component
- CAD assemblies
- engineering BOMs
- affected products
- active configurations
- prior revisions
- open engineering changes
- supplier records
- manufacturing usage
- quality information
Now imagine asking Nora Prima: “Check the relevant systems and prepare the findings for engineering review.”
An AI engineering agent operating across connected systems could potentially gather this information and organize it around the engineering decision.
The final approval would still follow the organization’s governed workflow.
But a significant portion of the information-gathering process could be accelerated.
That is much closer to meaningful AI-enabled engineering productivity than simply generating a summary.
Why Product Ontology Could Become Increasingly Important
Generative AI has made natural-language interaction much easier.
But natural language alone does not describe the full complexity of a product.
Engineering organizations rely on structured relationships.
A component has a revision.
That revision belongs to particular structures.
Those structures represent product configurations.
A requirement may affect one design but not another.
A manufacturing process may use a component within a particular plant or configuration.
A quality event may be associated with a specific revision or supplier.
Product ontology captures these relationships explicitly.
This creates a semantic foundation that AI can use when reasoning about product information.
As engineering AI evolves from answering questions toward performing tasks, this semantic foundation is likely to become increasingly valuable.
AI in PLM Is Becoming an Architecture Question
The AI discussion in PLM is often framed around individual features.
- Can AI summarize a document?
- Can it create a requirement?
- Can it generate a report?
- Can it provide a chatbot?
These features can be useful, but the larger opportunity is architectural.
Organizations need to consider how AI will:
- access engineering context
- understand product relationships
- connect with enterprise systems
- invoke tools
- respect authorization
- maintain traceability
- participate in workflows
- interact with engineers
Once AI begins performing work, these architectural questions become much more important.
The future of AI in PLM is therefore not simply about adding another AI feature to an existing product interface.
It is about creating an environment where AI can participate safely and intelligently within the engineering Digital Thread.
From Product Data to Connected Product Knowledge
Traditional enterprise systems often organize information around individual records.
AI makes another layer increasingly valuable: connected product knowledge.
Connected product knowledge combines the information itself with the relationships surrounding it.
For example, instead of knowing only that a part exists, the system may understand:
- where the part is used
- which design defines it
- which requirement influenced it
- which configuration contains it
- which supplier provides it
- which change modified it
- which manufacturing process consumes it
- which quality events relate to it
This relationship-rich context allows AI to answer substantially more useful engineering questions.
More importantly, it creates the foundation for AI to support engineering actions.
The Future of AI in PLM Is Not Another Chatbot
Natural-language interfaces will remain useful.
Engineers should be able to ask complex questions without navigating countless menus or running multiple manual searches.
But the conversational interface is only the surface.
Behind it, useful engineering AI will increasingly depend on:
- Context
- Ontology
- Tools
- Governance
The value comes from combining these elements.
AI should understand the connected product.
It should understand what relationships mean.
It should have controlled access to relevant tools.
And it should operate within engineering governance.
That is how AI can move from providing interesting answers toward creating meaningful engineering productivity.
Moving Toward an Agentic Digital Thread
The next evolution of the Digital Thread may therefore involve more than connecting systems and information.
It may also include AI agents capable of actively working across those connections.
An engineering agent could help identify impact, investigate alternatives, gather evidence, compare configurations, analyze related histories, and prepare decisions.
MCP can provide part of the connectivity layer needed to make those interactions more standardized.
Product ontology can provide the semantic meaning.
PLM can provide governed product context and processes.
And the Digital Thread connects everything together.
For Nora IPLM, the direction is clear:
AI should not simply sit beside product information.
Nora Prima should become an active, governed participant within a connected Digital Thread.
Connected product knowledge.
Shared meaning.
AI reasoning.
Tools.
Governance.
Together, these capabilities can move AI in PLM beyond experimentation and toward practical engineering work.
Frequently Asked Questions
What is AI in PLM?
AI in PLM refers to the use of artificial intelligence within Product Lifecycle Management processes to help users search, understand, analyze, create, and work with product information throughout the product lifecycle. More advanced approaches can connect AI with product structures, engineering changes, requirements, configurations, workflows, and external enterprise systems.
What is Model Context Protocol in PLM?
Model Context Protocol can provide a standardized way for an AI application or agent to connect with external data sources and tools. In a PLM environment, this could help AI interact with product data and authorized capabilities across PLM and connected enterprise systems.
How is an AI engineering agent different from a PLM chatbot?
A chatbot primarily answers questions. An AI engineering agent can potentially gather context, reason across connected information, use authorized tools, and support multi-step engineering tasks while respecting defined permissions and workflows.
Why does AI need a product ontology?
A product ontology gives semantic meaning to relationships between engineering objects. It allows AI to understand, for example, that a requirement is related to a design, a component belongs to a BOM, or a change affects a configuration rather than treating these as unrelated records.
Can AI work across CAD, PLM, ERP, MES, and QMS?
AI agents can potentially work across multiple enterprise systems when appropriate integrations, permissions, tools, and data relationships are available. This broader context is important because engineering information is rarely contained within a single system.
What role does the Digital Thread play in AI?
The Digital Thread connects product information and relationships across the lifecycle. This provides AI with richer engineering context and makes it possible to trace relationships between requirements, designs, BOMs, changes, manufacturing, quality, and other product information.
Should AI automatically make engineering changes?
Not necessarily. Engineering AI should operate according to defined governance, permissions, lifecycle rules, and approval workflows. AI can support analysis and prepare actions while people remain involved where engineering authorization or formal approval is required.
What is Nora Prima?
Nora Prima is Nora IPLM’s AI engineering agent concept for working across connected product data, enterprise systems, and governed engineering processes. The objective is to move AI beyond information retrieval toward practical engineering assistance within the Digital Thread.
Build AI on Top of a Connected Product Foundation
AI becomes far more useful when product information, engineering relationships, processes, and systems are already connected.
Nora IPLM brings product information, BOMs, CAD data, engineering changes, documents, workflows, and related lifecycle information into a connected PLM environment.
With Nora Prima, product ontology, and MCP capabilities, the goal is to extend that connected foundation into governed AI-assisted engineering work.
Explore Nora IPLM and see how a connected Digital Thread can provide the foundation for the next generation of AI-enabled product development.
See the consequence before you commit to the change
Discover how Nora IPLM Knowledge Thread turns connected product data into a traceable AI Change Impact Simulation, from qualified evidence and visual propagation to alternate-item, carbon, and AI-assisted analysis.



