Agentic AI and the Engineering Workflow

• Rockwell’s FactoryTalk Conductor orchestrates AI agents across engineering tools from Q1 2027
• Conductor supports Nemotron, keeping open models and data inside manufacturers’ own environments
• FactoryTalk Optix 1.8 adds AI design assistance and high availability clusters
• ctrlX Agent iXi lets customers choose their model and generates PLC code
• Industrial AI works when embedded in workflows, verifiable and bounded by role limits
• Claude configured TIA Portal hardware, wrote the PLC program and tested it on a digital twin
• GenAI builds pharma operator screens in minutes, shifting effort to data and validation
• Plant Simulation now uses GenAI copilots and multi agent digital twin content
• Agentic AI is positioned to coordinate factory exceptions within deliberately assigned authority

‍

MES and Enterprise Integration

• 93% of manufacturers have an MES, but only 23% have integrated it enterprise wide
• SAP will sunset legacy MII and ME by 2030 and brings MCP servers to DM Edge
• Siemens repositions Opcenter MES from system of record to system of intelligence
• Evenort gains visibility and traceability through Rockwell’s Plex MES
• QAD Redzone and NVIDIA link the factory floor, ERP and supply chain
• In mass customization, agents draft recipes, the MES enforces limits and humans handle exceptions
• Litmus Data Catalog targets inconsistent plant metric definitions, launching 17 November

‍

Physical AI and Robotics

• Siemens and NVIDIA bring physical AI to the Erlangen electronics factory
• Erlangen combines factory wide data, a digital twin and robots for unpredictable tasks
• Erlangen robot arms use Omniverse training data to correct packaging placements
• ABB launches IRB 1700 robots for precision electronics server manufacturing
• ctrlX Roxi humanoid skips legs and already works in customer projects
• AI Robot Trainer powers an upgraded vision based Fashion Inductor
• A SEMICON West panel on 14 October maps the path to autonomous fabs
• AI and agents target materials discovery and fab twins against $380B+ US investment

‍

Machine Vision, Digital Twins and AI Infrastructure

• NVIDIA VSS Blueprint 3.3 deploys a vision AI agent in under 30 minutes
• A vision language model answered eight checklist questions in 84.6 ms, 8 ms more than one
• The ctrlX Vision ecosystem adds partners and compute platforms from edge to data center
• MEGA MACHINE combines AI laser cutting analytics, weld inspection and assembly verification
• Foxconn’s NVIDIA AI factory reports 2.48x faster training and 25% fewer scheduling hours
• OpenUSD and simulation help the German Mittelstand test production changes virtually

‍

Scaling Industrial AI and Measuring Value

• 84% of German manufacturers generate AI value, but only 20% operate at scale
• Scaling physical AI fails without control, and winners deploy it consistently
• Sequence matters: eliminate, standardize, connect, then automate
• An AI model found equipment problems but never entered routine operations
• ADNOC reports engineers support two to three times more rigs across 120+ rigs
• Europe’s agenda: brownfield assets, edge AI, data to action and speed to value
• Munich summit guidance: pursue big opportunities, budget people and AI together

‍

Platforms, Partnerships and Market Structure

• Schneider Electric’s proposed PTC acquisition raises interoperability and partnership questions
• ABB Genix is a Gartner Leader for Industrial AIoT Platforms for a second year
• Bosch Rexroth pairs ctrlX AUTOMATION with Microsoft AI and cloud
• A podcast asks whether AI will change which industrial software firms dominate
• The Open Industry 4.0 Alliance evaluates the Margo specification for edge workloads
• Machine tool builders move toward software subscriptions and outcome based models
• Tetra Pak, NVIDIA and Tesla extend value into ecosystems and services

‍

Workforce and Skills

• Almost half of Asia Pacific manufacturers are increasing automation, and 51% expect to redeploy staff
• PLC programmer value shifts from writing code to defining behavior and owning results
• ABB Robotics and NVIDIA publish a white paper on workforce readiness for physical AI
• A four week hackathon has students building physical AI with real Rockwell plant data
• Controls frameworks deliver value only when documentation and examples enable other engineers

‍

Thanks to Rob Kemp, Mateusz Hanisz, Alvin Clark, Michael Grollmus, Zakaria SANKARA, Sebastian Creischer, Tony LeRoy, Nirat Keswani, Rev Lebaredian, Brion Carroll, Subhajit Bodhak, Markus W. Hacker, Ricky Watts, Christopher Stephens, Bin Wang, Jerry Gootee, Amit Sharma, Jasper Wildeboer, Omid G., João Roque, Peter Fisk, Colin Masson, Rahul Patel, Milos Kucera, Karthik Amirtharajan, Oliver Scheb, Chetana Rodrigues, Jana Vašíčková, Ben Stewart, Tobias Gerhard, Joakim Rosenqvist, Steffen Winkler, Paul Haikal, Craig McDonnell, Ilya Kurganov, Metin Kaplan, Kudzai Manditereza, Scott Wooldridge, Vladimir Troy, Patrick Ho, Ewerton Gozzi, Jacobo Lloret Casal, Richard Jeffers, Derek Riley, Yiyi Wang, Dr. Sebastian Scheurle, Hans Michael Krause, Stefan W., Bart Nieuwborg, Dr. Elnaz Nayeri, John Younes, Roland Laucher, Cedrik Neike, Jeff Winter and everyone else who contributed insights to this edition. Find the full list of posts and voices on LinkedIn: