Rockwell Automation just proved that artificial intelligence and the factory floor go together really well.
Picture this: a technician standing in front of a broken machine, cracking open a manual that runs hundreds of pages, and hunting for the right error code to fix the problem. If the answer wasn’t in there, the next step was tracking down a coworker who might have dealt with the same breakdown before. Meanwhile, the production line sat idle, racking up losses by the minute. That was everyday life for countless professionals in factories around the world, including at Rockwell’s plant in Singapore.
But since October 2025, that scenario has changed completely. Enter the GenAI-Powered Maintenance Copilot, a generative artificial intelligence tool developed in-house by the company that transformed how technicians diagnose and fix machine failures. And the numbers that followed are pretty hard to ignore. 👇
What the Maintenance Copilot Is and How It Works
The Maintenance Copilot is a generative AI solution built by Rockwell Automation itself to work directly within the industrial maintenance process. The tool was trained on three main sources of knowledge: the expertise of the workers themselves, data from the company’s manufacturing software, and machine instruction manuals. All of that was brought together in a conversational interface where a technician simply describes the problem and gets clear, context-aware guidance on how to solve it.
Behind that simple experience sits a pretty robust architecture. The solution runs on Microsoft Azure cloud, was built using Azure OpenAI within Microsoft Foundry, and operates on OpenAI’s GPT-5.4-mini language model. In other words, a cutting-edge large language model is working behind the scenes to help the maintenance team diagnose and resolve issues way faster than before.
Someone who feels this firsthand is Mangleswaran Mahalingam, an engineering assistant at the Singapore plant. Before, he had to go looking for coworkers or senior technicians to ask if they had ever seen a particular error. Now, he just types his question into the copilot on a tablet and gets the answer right away. In his own words, it became faster and easier. And it doesn’t stop there: Mahalingam has already learned to access the AI using augmented reality smart glasses, something that would have felt like science fiction just a few years ago. 🤓
In practice, the technician no longer needs to be an expert on every piece of equipment on the production line. They interact with the Copilot as if they were talking to a really well-informed colleague who knows every detail of the manuals, every error code, and every repair procedure.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Quick Access to Decades of Veteran Knowledge
One of the most interesting parts of this story is how the system captures decades of experience. When an error is detected at the Singapore factory, the AI assistant pulls specialized information from a database built by veteran Rockwell engineers — professionals with 20 to 30 years of hands-on factory floor experience.
The tool organizes that information by symptom, cause, and response, so the technician can immediately see the most likely causes of the error, whether it’s a worn gear or a faulty sensor. On top of that, they can read what other colleagues have noted about the same problem and check whether it has happened on other production lines. And if the technician still wants to consult those thick instruction manuals, they can do it in plain language, with the copilot responding in an easy-to-follow step-by-step format.
For Li Wang, director of the Singapore plant, the big problem they set out to solve is knowledge loss — what they call tribal knowledge. It’s all that tacit expertise that walks out the door when an experienced worker retires. Another major win, according to her, is consistency: there is now a standard approach for every problem, regardless of who is handling it.
The Results That Turned Heads
When Rockwell Automation released the first round of data after rolling out the Maintenance Copilot, it was clear that the bet on artificial intelligence for the industrial environment was paying off. This consistent, targeted approach cut machine downtime by 33 percent. Costs dropped too: with fewer maintenance service calls and replacement parts, Rockwell’s internal tracking shows a cost reduction of roughly 25 percent.
But maybe the most impressive number is the impact on new employee training. Mahalingam remembers feeling worried when he joined the factory in 2021 because he knew nothing about the machines. It took him over a year before he felt confident enough to take on night shifts, when fewer coworkers are around. Today he oversees about 50 machines.
According to Bob Buttermore, Rockwell’s supply chain director, the onboarding time for new workers has dropped dramatically. Instead of taking nine months to learn how to diagnose hundreds of machines, it now takes just three months. He sums it up well: a massive improvement, and AI was the catalyst. 🚀
It’s also worth pointing out that these results were achieved in a real production environment, not in a controlled test scenario. The Singapore plant operates with the typical challenges of any factory — load variations, natural equipment wear, and the constant pressure to hit targets. The fact that the Maintenance Copilot performed well in that context is a strong sign of how mature the solution really is.
The Factory of the Future
The Singapore facility manufactures industrial automation equipment like controllers and networking components for high-growth sectors such as pharmaceuticals, automotive, and semiconductors. And recognition came from outside: in June, the plant was designated a global lighthouse site by the World Economic Forum.
In its citation, the WEF noted that the facility increased units produced per person-hour by 43 percent, reduced defects by 35 percent, and shortened time-to-competency by 67 percent. Numbers that put the factory in a league of world-class references.
The maintenance copilot was just one of several AI-powered capabilities deployed at the site. According to Li Wang, other additions include a multi-agent quality assurance system that detects and corrects defects on the assembly line in real time, and a predictive maintenance system that helps prevent unplanned shutdowns of critical equipment.
For Patrick Dey, vice president of data, analytics, and AI innovation at Rockwell, these solutions show how it’s possible to combine the company’s operational and automation expertise with AI technology to drive innovation. And he makes a point of emphasizing: this innovation doesn’t just improve internal operations — it’s also used to help customers accelerate their own digital transformation journeys.
Architecture Built to Scale
Dey’s team built these tools by leveraging the breadth of Microsoft’s technology ecosystem. That included Azure OpenAI Service and Azure AI Search for conversational experiences and intelligent reasoning, along with Azure’s security, identity, and governance services to ensure enterprise-grade protection and compliance.
That careful attention to architecture has a clear reason behind it: global scalability. Microsoft’s cloud-native architecture supports expansion from a single factory to worldwide deployments, according to Dey. And the integration within the Microsoft ecosystem provides seamless connectivity across data, AI, applications, and collaboration tools.
This isn’t just talk. Rockwell has more than 25 plants around the world and over 26,000 employees. The maintenance copilot, for example, already has a rollout date set for the Twinsburg factory in Ohio, as well as facilities in Poland and Mexico. The idea is that the lessons learned in Singapore will serve as a blueprint to drive digital transformation across the entire company and its customers as well.
When AI Becomes a Coworker
Maybe the most compelling part of this story is how these systems are being perceived within the company. For Li Wang, these solutions are increasingly becoming like a teammate. According to her, these applications are more than just a tool — they are guiding the entire team into a new world.
That forward-looking vision is shared by Chris Nardecchia, Rockwell’s chief information officer. For him, the path forward is one where AI is present at every stage of the work. That kind of productivity will allow people to do more and create new possibilities, he says. And he leaves a valuable thought: for AI to succeed, you need good data, a trained workforce, and a culture focused on learning and results.
Rockwell Automation seems to have made the right call by developing the tool in-house and testing it at its own plant first. That approach ensured the product was shaped by the real needs of the people using it, rather than by a vision disconnected from day-to-day operations. That is exactly why the Singapore factory works as a living showcase for customers.
As Buttermore puts it, the ultimate goal is to be a reference for customers. By deploying, testing, and scaling these solutions at its own facilities first, the company can bring proven approaches to manufacturers looking to improve productivity, resilience, and workforce effectiveness. And that is industrial efficiency in action: not just doing more with less, but doing it better, with the right knowledge available at the right time and for the right people. 🤖⚙️
