How Agentic AI Can Transform Industrial Operations
Agentic AI is emerging as a transformational paradigm across industries, with the potential to reshape how companies design, manufacture and manage operations. Agents represent a significant step towards autonomy in AI systems, as they can orchestrate complex workflows, coordinate across multiple platforms and execute actions.
Yet, transformations take time and, especially in the case of AI agents imbued with a level of autonomy, require careful planning and robust infrastructure to ensure human experts maintain final review and approval authority.
In a recent podcast, host Conor Peick is joined by guests Matthias Loskyll, Senior Director of AI and Robotics at Siemens Digital Industries and Samir Desai, Senior Director and Global Program Head for Data and AI at Siemens Digital Industries to discuss how AI agents are already factoring into the future of industry, and how companies can best approach their adoption.
Check out the full podcast here, listen to the previous episode here, or keep reading for a summary of the highlights from the final part of the discussion.
Agentic AI Capabilities Varies with Context
The discussion opens with a focus on the development of agentic AI, how agents are distinct from AI copilots or assistants, and how agents may evolve in upcoming years. Desai shared how industry observers predicted significant growth of agents in 2026, dubbing it the “year of the agents”.
He also explained how these forthcoming agents can be organized into two categories, representing different capabilities aligned to the two complementary functions. First, domain agents operate within specific tools and systems, performing targeted tasks in familiar environments. These agents will work alongside users and will be capable of executing specific functions at the direction and under the supervision of humans. An AI agent that can generate machine code for review, modification and implementation by human engineers is one example.
Second, orchestration agents operate at a higher level, coordinating activities across multiple systems and platforms. These can also be thought of as “digital thread agents” because they coordinate between mechanical, electrical systems, simulation, manufacturing and other domains. When an engineer makes a change to a component in a car, for example, that modification ripples through numerous systems. The 3D model must be updated, data must be changed, simulations must be run, and manufacturing process plans must be revised. Digital thread agents orchestrate this entire sequence, ensuring that changes propagate correctly across all affected systems.
Future Agentic AI Moves from Monitor to Actor
Loskyll noted similar patterns emerging in agentic AI for production settings. As an example, Loskyll described how domain and digital thread agents may work together to manage a production line experiencing errors. One agent might monitor operations and spot the issue. Another could analyze root causes, determining what’s creating the breakdown. A third agent might initiate corrective actions, optimizing parameters in upstream processes. And another could feed information back to design or planning agents.
This level of autonomy also introduces new challenges. Loskyll noted the importance of strong guardrails and appropriate models and training data, “We give away certain control in this very critical environment,” notes Matthias Loskyll. “So how do we design that? How do we make sure these agents speak the language of our domain?”
Physical AI: Embodied Intelligence on the Shop Floor
Loskyll went on to outline another breakthrough coming in the deployment of agentic AI in production environments. This concept, often called physical AI, centers on embedding AI agents in robotics or other physical systems, making physical systems more intelligent and adaptive in dynamic real-world environments. Rather than large language models, physical AI is built on vision language action models that process visual, textual and action data to perceive surroundings, receive instruction and execute actions.
This capability opens possibilities for automating tasks that have resisted traditional programming approaches. Handling variable items like textiles, cables, or irregular components requires adaptability that rigid programming cannot provide. “This is the next big lever coming, hopefully, in the next few years,” Loskyll explains. “Automating the unknown, how we sometimes call it.”
Even with increasingly intelligent robotics, Desai and Loskyll did not forecast a future of “dark factories” operating entirely autonomously. “I’m deeply convinced there will be humans needed for many tasks, both in the engineering phase, but also operating and running the factories,” Loskyll emphasizes. “This will be an interplay between robots, autonomous systems, agents, and humans in the end.”
Practical First Steps Set the Stage for Transformation
Despite the speed of AI advancements, industrial adoption of the technology must proceed more deliberately. Manufacturers operate critical production environments with massive existing investments.
Industrial companies should approach the adoption of generative, physical and agentic AI by taking measured steps towards ambitious goals. Companies can begin by focusing on low-hanging fruit that delivers immediate value before progressively tackling more ambitious applications. Loskyll summarized the approach well, “First, get your data foundation right as we discussed. Second, start with the high value use cases and then start rolling those out. And third, upskill your people, don’t forget your people. Take them on your journey, think about this cultural transition that we are on and look for a good partner, a strong partner who understands industries.”
For companies ready to embrace this future with practical ambition and the right partnerships, the opportunity to transform their business is significant.
Siemens Digital Industries Software helps organizations of all sizes digitally transform using software, hardware and services from the Siemens Xcelerator business platform. Siemens’ software and the comprehensive digital twin enable companies to optimize their design, engineering and manufacturing processes to turn today’s ideas into the sustainable products of the future. From chips to entire systems, from product to process, across all industries. Siemens Digital Industries Software – Accelerating transformation.
