In recent years, generative AI has evolved from simple chat functions to sophisticated agents, driven by advancements in large language models (LLMs). A key technology in this area is the Model Context Protocol (MCP). Our company has applied this to the PI System to enhance maintenance activities.

By building a PI System MCP Server, we've enabled LLMs and AI agents to access AF tags, time-series data, and event frames, facilitating natural language data retrieval. This integration improves plant monitoring, reducing manual effort.

Our heavy-duty gas turbines, monitored 24/7 globally, now benefit from the MCP framework, shifting from deviation monitoring to fixed-point monitoring of underperforming areas, expanding predictive maintenance possibilities.