Introduction
ThingWorx 10.1 takes a major step toward intelligent interoperability by introducing the Model Context Protocol (MCP) in Public Preview. As industries accelerate their use of AI-assisted operations, MCP provides a standardized way for AI agents to securely connect with real-world industrial data and perform actions across systems.
Our key goal is enabling agent-driven automation along with intelligent interoperability. MCP allows AI models and agents to interact with external systems such as ERP, MES, CRM, and analytics platforms in a structured and secure way. It removes the need for custom connectors and integrations, enabling developers to use a single, open standard to bridge AI and operational data. In essence, MCP allows AI to work natively with ThingWorx services and data models.
ThingWorx and MCP
Low-Code Enablement for Agentic Automation
ThingWorx now embeds an MCP Server directly into the platform. This means that your existing ThingWorx services can be instantly made “AI-ready” without the need for external deployments. Through a simple low-code interface called MCPServices, users can define, manage, and expose three core MCP entities — Tools, Resources, and Prompts — that form the building blocks of contextual AI workflows.
MCP Services
Tools represent actions or functions AI can call (such as retrieving machine KPIs), Resources represent contextual data the AI can reference, and Prompts are reusable templates that guide AI on how to interact with them. This model creates a foundation for secure, context-aware automation that can be reused across multiple agents and applications.
Secure by Design, Open by Standard
MCP in ThingWorx follows industry-standard security protocols by introducing OAuth-protected metadata endpoints, compliant with RFC 9728. These endpoints let clients authenticate and discover the resources they’re authorized to use — ensuring data access remains secure while supporting open interoperability.
This aligns with ThingWorx’s broader goal: creating an ecosystem where AI agents can safely access contextualized industrial data across systems. Whether connecting to SAP, Salesforce, or another MCP-compatible server, your ThingWorx instance can now participate in a larger agentic ecosystem.
Seamless Interoperability and Scalability
All MCP configurations — tools, resources, and prompts — are stored within ThingWorx’s persistence layer, ensuring that your MCP setup scales with your enterprise environment. Agents can connect to multiple ThingWorx servers deployed globally, retrieve contextual data from each, and feed it into AI-driven workflows that span factories, regions, or entire business units.
This design lays the groundwork for domain-specific large language models (DSLMS) and enterprise AI assistants that understand and act on operational data directly from ThingWorx.
Feature Summary
For instance, you may have ThingWorx Server 1 supporting factory sites in the USA, another in Germany, and a third in Mexico. These local ThingWorx deployments, powered by MCP capabilities, can expose rich contextualized data for agentic automation.
MCP clients could then be used to perform enterprise-wide KPI calculations to benchmark performance across factory sites over standard metrics such as OEE.
As local systems (ERP, MES, or factory systems) adopt MCP, ThingWorx MCP clients can directly access their data, enabling seamless integration without building custom REST connectors — saving significant development effort in API integration and data mapping.
MCP Usage with TWX - Concept
Capability |
Description |
|---|---|
|
Embedded MCP Server |
Native support inside ThingWorx – no extra setup required. |
|
MCPServices Resource |
Low-code interface to manage Tools, Resources, and Prompts. |
|
OAuth Security |
RFC 9728-compliant protected metadata for secure AI access. |
|
Persistence Layer |
Stores MCP configurations across databases for scalability. |
|
AI Integration |
Enables context-aware, agentic automation via MCP Tools. |
Feature Benefits
Overall, MCP support with ThingWorx delivers several key benefits:
-
Standardized, secure, and discoverable data access via the ThingWorx MCP Server, making operations AI-ready.
-
Dynamic population of MCP tools without the need for custom code.
-
Ability for AI agent developers to create tailored agents that use selective ThingWorx services enriched with context for higher accuracy.
-
Seamless alignment with agentic AI architectures for automated workflows.
-
Direct interoperability with enterprise systems already supporting MCP servers, allowing AI agents to connect and retrieve data easily.
Upgrade to ThingWorx 10.1 to try MCP
The introduction of MCP marks the start of AI-native industrial automation in ThingWorx. By adopting 10.1, you gain early access to a framework that will power the next generation of connected, intelligent systems — helping your business stay ahead as AI integration accelerates across industries.
We encourage feedback from users who have tried the MCP preview — what capabilities would you like to see next, and how can we improve interoperability and automation through ThingWorx
If you’re new to ThingWorx release phases, read about Public Preview and other stages here
Vineet Khokhar
Principal Product Manager, IoT Security
Stay tuned for more updates as we approach the release of ThingWorx 10.1, and as always, in case of issues, feel free to reach out to <support.ptc.com>
