Meg Folcarelli is the MedTech Industry Marketing lead. Known for her thoughtful storytelling, Meg helps translate ideas into messages that resonate, making communication more practical, engaging, and impactful.
Where should your organization adopt existing MCP capabilities, and where should it build?
In short: Adopt PTC-supported capabilities where they exist, extend them for your specific workflows and integrations, and build where you can create differentiated business value.
Product development leaders are under pressure to show results from AI, and MCP can help teams move quickly. Model Context Protocol (MCP) is an open protocol that gives AI applications a consistent way to discover and use capabilities from enterprise software. An AI application can use an MCP client to discover and invoke capabilities exposed by an MCP server. PTC provides MCP server capabilities for its products, while customers and partners can build AI agents, applications, and experiences that consume those capabilities.
MCP can make it easier and faster to connect AI applications to product capabilities, but access is only the beginning. In production, an MCP capability needs to be operated, tested, secured, supported, and maintained as both the underlying product and AI ecosystem evolve. It must also respect the authorization, access controls, business rules, and audit requirements of the system of record.
The right question is not just “Can we build it?” but “Which parts should we own, and where will that ownership create value?”
Product data needs product context
Giving an AI agent access to product data is only part of the challenge. The agent also needs the product and lifecycle context required to interpret that data correctly. In Windchill, for example, that context can include revisions, configurations, structures, relationships, lifecycle states, requirements, changes, and the business rules that connect them.
This product-aware semantic layer is much harder to recreate than a basic connection.
An engineering change may span parts, documents, requirements, configurations, workflows, and multiple related objects. Without the right product context, an AI agent may find the data but still be left to guess which revision, configuration, relationship, or lifecycle state actually matters.
In product development, those errors can have consequences:
- Quality: Decisions based on incorrect revisions or configurations can contribute to rework, scrap, or field issues.
- Compliance: In regulated environments, answers need to be traceable to the appropriate system of record.
- Security: AI connections need to respect the same access controls that govern the underlying product data.
- Cost: Poorly targeted requests can lead to unnecessary data retrieval, tool calls, and token consumption. Product-aware capabilities help AI find the right information with less processing.
MCP itself is not the governance layer. The underlying PTC product remains responsible for authorization, access control, business rules, and governance. PTC-supported MCP capabilities provide a supported path for AI clients to interact with those governed product capabilities while operating within the product’s existing controls.
Your AI, your choice, with PTC product expertise
PTC’s role is to provide and maintain the common, product-aware foundation behind the capabilities it exposes. The advantage is not merely operating an MCP server. It is PTC’s understanding of the product model, semantics, relationships, business rules, and lifecycle context that AI applications need in order to work with product data responsibly.
That approach is taking shape across the PTC portfolio:
- Windchill: PTC-supported MCP capabilities are intended to give AI clients a consistent way to interact with governed Windchill information and capabilities while respecting Windchill permissions and controls.
- Onshape: The Onshape FeatureScript MCP Server connects coding-focused AI applications with Onshape’s FeatureScript language, helping engineers translate design knowledge into reusable custom tools.
These examples point to the broader MCP-compatible AI ecosystem PTC wants to enable. Specific availability and compatibility will depend on the product capability and release.
These examples point to the broader MCP-compatible AI ecosystem PTC wants to enable. Specific availability and compatibility will depend on the product capability and release.
See how AI helps teams get more value from trusted product data
Windchill AI brings AI-driven assistance, insights, and automation into PLM workflows.
Explore Windchill AIThe key difference is the product knowledge behind the connection.
A product-aware MCP server can account for how the underlying system works, including how parts relate to structures, how changes move through workflows, and which rules apply. Because PTC can maintain that knowledge alongside its software, customers have less product-specific logic to recreate and support themselves while retaining the flexibility to build around their own needs.
Adopt the foundation. Extend and build the differentiation.
The most effective boundary is straightforward: PTC owns and maintains the common product-aware foundation. Customers and partners differentiate through workflows, cross-system orchestration, agents, user experiences, integrations, and proprietary business logic.
That makes “adopt, extend, or build” a more useful framework than a simple build-versus-buy decision:
- Adopt PTC-supported MCP capabilities where they already provide the governed, product-aware foundation you need.
- Extend that foundation for organization-specific workflows, integrations, and processes.
- Build custom agents, experiences, and business logic where they create differentiated value.
Partners remain critical to this model. Supported product capabilities give partners a stronger foundation on which to create differentiated solutions, rather than requiring them to reproduce product semantics, permissions, and lifecycle logic before they can begin innovating.
Five questions to guide the decision
Before investing in a new MCP capability, consider five questions.
- Are we reproducing a capability or product-aware foundation that PTC already provides?
If the requirement is common across organizations using the same PTC product, recreating and maintaining a separate version may add ownership without meaningful differentiation.
- What are we agreeing to own?
Include maintenance, upgrades, security reviews, testing, observability, support, and compatibility with future product changes in the business case, not only the initial development effort.
- Who will maintain the product context and business rules?
Someone must keep revisions, configurations, relationships, lifecycle states, permissions, and other product semantics accurate as the product evolves.
- Will the approach hold up technically and economically at enterprise scale?
A pilot for one team may work well without meeting the governance, audit, access, reliability, and support requirements of broader adoption. At scale, inefficient data retrieval, repeated tool calls, and unnecessary token consumption can increase production costs, making the efficiency of the MCP capability part of the economic equation.
- Is custom development focused on differentiated value?
Engineering and partner capacity have the most impact when focused on proprietary workflows, agents, experiences, integrations, and business logic, rather than rebuilding a common product foundation.
Build where it creates an advantage
MCP is making it easier for AI applications to access enterprise software capabilities, but easier access does not eliminate the need for product context, governance, or the long-term responsibility of operating production capabilities. A supported, product-aware foundation can reduce the amount of product-specific logic customers and partners need to recreate while preserving the governance of the underlying system. The goal is not to discourage building. It is to build at the right layer: adopt the governed product foundation where it exists, extend it around your workflows and systems, and invest custom engineering where it creates a real advantage.