Schneider Electric, Nvidia and Hewlett Packard Enterprise (HPE) announced a collaboration to run autonomous AI agents inside process plants. The companies call the offering Industrial AI Agents at the Edge and describe it as a practical route to “physical AI” for continuous process industries.

Most physical AI discussion centers on robots, humanoids and autonomous machines. The three companies say process operators face a different problem. They run complex, continuous operations where reliability, safety, cybersecurity and energy efficiency drive performance, and until now there has been no clear blueprint for applying Physical AI there.

“While much of today’s focus is on autonomous machines, we see a significant opportunity to apply Physical AI to the industrial systems that power the world’s critical industries,” said Gwenaelle Huet, executive vice president of industrial automation at Schneider Electric. “Together with Nvidia and HPE, we are bringing together expertise in automation, AI and edge computing to help customers transform operational data into action, improving resilience, efficiency, cybersecurity and sustainability while laying the foundation for more autonomous industrial operations.”

Built for the plant floor, not the cloud

The agents run HPE Private Cloud AI, which HPE co-engineered with Nvidia as part of its HPE AI Factory with Nvidia portfolio. They are paired with Schneider Electric’s operational technology expertise. The companies say keeping AI at the edge addresses three barriers to scaling industrial AI:

• Data sovereignty: Operational data is often too sensitive or regulated to leave plant boundaries. Local deployment keeps control, security and compliance in the operator’s hands.
• Cybersecurity: Added connectivity can add risk in critical infrastructure. Processing data close to its source reduces the attack surface.
• AI consumption costs: Cloud inference, data transfer and token-related costs can climb as AI usage spreads across sites and assets. Edge deployment can help reduce some of those costs and gives operators more control over where AI workloads run.

Customers deploy, manage and govern multiple agents through Schneider Electric’s EcoStruxure Industrial Insights apps, which provide a single environment for lifecycle management and security oversight.

Three initial agents

The first agents focus on OT monitoring, cybersecurity and energy optimization:

• OT monitoring agents combine real-time operational data, equipment configurations, process topology, alarms, maintenance history and engineering context to understand the physical state of operations. They identify emerging issues, reason for likely operational impacts, recommend or initiate corrective actions, and verify that the intervention delivered the expected physical outcome.
• Cybersecurity agents correlate network activity, control system behavior, asset configurations, operational events and process context to detect abnormal conditions. They assess likely impacts on operations, recommend or execute response measures, and validate that systems have returned to a secure and stable state.
• Energy optimization agents integrate process data, energy consumption patterns, equipment performance, operating conditions and production requirements. They identify optimization opportunities, recommend or implement adjustments, and verify the effect on energy efficiency, operational performance and sustainability.

Proven foundation at CMPC Chile

The agents build on Schneider Electric’s Operational Technology Monitoring System (OTMS), which CMPC Chile has run in production for three years. CMPC, part of Empresas CMPC S.A. and one of Latin America’s largest pulp, paper and packaging producers, uses OTMS to monitor the health and availability of control systems across its continuous pulp and paper operations.

OTMS gives CMPC teams a unified view of control systems, servers, workstations and switches. Specialists can spot emerging issues earlier, diagnose problems faster and set clearer maintenance priorities.

“In continuous industrial operations, identifying issues early is critical to protecting production and avoiding costly disruption,” said Matías Jory, digital and process manager in CMPC’s industrial operations vice-presidency. “Looking ahead, we see significant potential for AI Agents to build on this proven foundation, helping our specialists investigate issues faster and move more effectively from insight to action, while keeping people in control of critical decisions.”

Partner perspectives

Nvidia’s Nemotron open AI models and HPE Private Cloud AI are part of the stack. “Industrial operations generate vast amounts of data, but turning that data into action remains one of industry’s biggest challenges,” said Sri Kodali, global head of developer relations for energy and utilities at Nvidia. He said combining those technologies with Schneider Electric’s industrial expertise and OTMS platform lets the agents “monitor, reason and optimize operations in real time.”

HPE’s Alexia Clements, SVP of hybrid cloud solution sales, said industrial organizations want assurance that sensitive data and critical systems stay protected. “This gives customers a way to expand AI use across their facilities while maintaining the security, privacy and control their businesses require,” she said.

The companies say the longer-term goal is to “automate the automation,” moving toward safer, more efficient and increasingly autonomous process operations. For now, they emphasize that people remain in control of critical actions.