Velotic Ships ThingWorx 10.2 With Agentic AI for Factory Floors

Velotic has released version 10.2 of its ThingWorx Industrial Intelligence Platform, adding three agentic AI components aimed squarely at plants running software and equipment that predate the AI era. The Boston-based vendor positions the release as a way to connect AI to the industrial data, systems and applications manufacturers already operate — rather than layering another disconnected analytics layer on top.

The 10.2 release bundles ThingWorx AI Assistant, ThingWorx MCP Server, and AI and Agent Services. Together they form what the company calls an enterprise foundation for building secure, extensible AI assistants and agents that work with industrial data, context and existing systems.

What does the MCP Server change for legacy plants?

The most consequential piece for practitioners is support for Model Context Protocol (MCP — an open standard that lets AI agents retrieve context from external systems). ThingWorx can now create MCP servers for legacy systems, meaning AI agents gain access to operational data from platforms that were never designed to talk to AI at all. For plants with decades-old equipment and applications, that is the practical bridge: AI reaches the historian, the MES (manufacturing execution system) and the workcell without a rip-and-replace program.

"Manufacturers don't need another disconnected AI environment. They need intelligence that can understand and work with the operations they already have," said Chris Balwin, chief product officer at Velotic. "ThingWorx provides the platform backbone for our AI strategy, connecting operational context with new forms of intelligence and giving customers a practical path to bring AI into the environments they rely on today."

Why industrial context is the hard part

Velotic's argument rests on a familiar plant-floor reality: industrial data and context sit scattered across equipment, applications and systems built up over decades. ThingWorx already lets customers model those environments and use connected data in industrial applications. The new capabilities build on that foundation, giving developers a more direct route to AI assistants and agents that actually understand the environment they operate in.

The payoff, according to one implementation partner, is measured in engineering hours. "The real factory upgrade is fewer hours lost finding answers. ThingWorx 10.2's agentic AI capabilities give factory teams a more direct way to ask questions in plain language, investigate problems faster, and spend less engineering time searching through data," said Shashi Preetham, CEO of PSPTechHub, an IT services and consulting firm specializing in industrial digital transformation, Internet of Things (IoT) and smart manufacturing.

What else is in 10.2?

Beyond the AI additions, ThingWorx 10.2 advances the platform's core industrial applications, including:

• Connected workcells
• Real-time production performance monitoring

Those functions matter because they feed the operational context the new agents depend on — an AI assistant is only as good as the production data it can reach.

What to watch next

The release is available now with ThingWorx 10.2. Watch for early deployments that test the MCP-server claim against real legacy installations: whether AI agents can reliably query pre-AI systems, and whether plain-language troubleshooting actually cuts engineering search time on live lines. Adoption among Velotic's installed base over the next two quarters will show whether agentic AI in manufacturing moves from platform promise to measured downtime reduction.