Skip to main content
R
RV_109304424-Participant
4-Participant
May 8, 2026
Question

LLM Accelerator that leverages ThingWorx 10.1

  • May 8, 2026
  • 2 replies
  • 69 views

How can the LLM Accelerator be integrated with the native MCP capabilities introduced in ThingWorx 10.1 to enable interoperability with Mistral AI models? Additionally, are there any sample implementations or reference code available for building a chatbot within ThingWorx 10.1 and generating reports natively using these AI capabilities?

2 replies

A
anursinghCommunity Manager
Community Moderator
June 1, 2026

Hi ​@RV_10930442,

Thank you for your question. 

Your post appears well documented but has not yet received any response. I am replying to raise awareness. Hopefully, another community member will be able to help.

Also, feel free to add any additional information you think might be relevant. It sometimes helps to have screenshots to better understand what you are trying to do. 

 

Regards,

Anurag 

R
RV_109304424-ParticipantAuthor
4-Participant
June 8, 2026

Thank you, Anurag, for helping reach out to additional participants. PTC shared some sample code for implementing a chatbot‑like feature, but it relies on Microsoft Azure OpenAI and does not make use of MCP capabilities. I have included below the link to the PTC GitHub repository that contains the accelerator code.
GitHub - thingworx-field-work/Thingworx-LLM · GitHub

S
slangleyCommunity Manager
Support
June 2, 2026

Hi ​@RV_10930442 

We don’t have a lot of information on this at this point, but you should be able to leverage the built-in functionalities of ThingWorx to facilitate communication between the LLM Accelerator and Mistral AI. This would involve utilizing the APIs and connectors available in ThingWorx to establish a seamless data flow and interaction between the two systems.

I’m sorry I can’t be more specific at this time.  Hopefully someone from the Community with actual experience can provide some insight.

Regards.

--Sharon

 

R
RV_109304424-ParticipantAuthor
4-Participant
June 8, 2026

Thank you, Sharon, for your response. I was able to integrate Mistral with ThingWorx 10.1 and successfully invoke the Tools created in ThingWorx to retrieve some results. However, our requirement aligns more closely with the PTC demo, which showcases how an AI chatbot can pull data from an IoT application and generate reports on the fly. I have shared an image from the PTC demo below for reference.

 

N
Nathaniel_SCommunity Manager
Community Manager
June 11, 2026

Hey ​@RV_10930442,

The LLM Accelerator is quite old (in AI tech terms) and is just using API calls to communicate with Azure OpenAI. It was written well before MCP was on the table for ThingWorx and was not built with that type of connection in mind.

The best resource I can find regarding what you’re describing is in the ThingWorx Help Center MCP Sample Use Cases page. Specifically you’ll want to look at use case #3, which goes over making data resources available for the MCP client (and associated LLM) to pull from. You’ll have to change things around to match your specific use case, as the sample is utilizing VS Code and Claude desktop on the client end. However, creating the resource on the server for the client to utilize should be universal.

Unfortunately, there are no sample implementations available for MCP enabled AI chatbots.  ThingWorx does have built in functionality in the MCPService Resource entity to allow for Tool, Resource, and Prompt creation and management, but the individual components are left up to the implementer to add for their specific AI applications. 

Best regards, Nathaniel
    Terms & ConditionsAccessibility statement