Artificial intelligence (AI) is now one of the most significant forces shaping industrial automation. Across manufacturing, design and maintenance, AI is shifting from a future concept to an everyday tool. A recent survey found that 40% of manufacturers have prioritised AI and automation above other digital transformation projects, with 80% allocating resources to implementation. However, around 30% still cite integration challenges as a key concern.
Automation specialist Beckhoff Automation is addressing those challenges by expanding its portfolio of AI-driven tools for the engineering and control environment. Project manager Ben Harrison says the company’s modular architecture allows rapid adoption of new technology. “Because our systems are modular and IPC-based, we can integrate technologies such as AI and machine learning easily,” he says. “Adding these into our TwinCAT runtime follows the same design pattern as our other components. It means engineers can experiment, prototype and deploy AI tools faster using our rolling trial licence.”
TwinCAT CoAgent is Beckhoff’s built-in AI assistant designed to support engineers directly within their programming workspace. Through a simple chat interface, users can prompt the system to write PLC or HMI code, review and explain logic, generate reports or access documentation. CoAgent is connected to Beckhoff’s Infosys knowledge base, so it understands TwinCAT’s structure and language.
Harrison says tools like CoAgent represent a major shift in how engineers will work. “AI assistants will make development faster and more intuitive,” he says. “You’ll spend less time searching for answers and more time refining ideas. Instead of looking through manuals, you’ll ask questions and get workable examples you can apply straight away.”
While he believes coding agents are becoming far more reliable, Harrison maintains that quality assurance remains essential. “Automated testing is still the best way to make sure code does what it’s supposed to do,” he says. “Use AI to accelerate development, but keep validation systems in place to maintain accuracy and safety.”
Machine learning intelligence
Beckhoff’s TwinCAT Machine Learning tools bring AI capability into live control environments. The TE3850 Machine Learning Creator allows engineers to create trained models directly from datasets using a clear, graphical interface. The software automatically generates initial model versions, helping to reduce errors and accelerate projects ranging from classification and forecasting to anomaly detection.
Harrison explains that machine learning allows complex behaviour that traditional automation can’t easily achieve. “In the past, engineers had to hand-code every scenario. If a motor vibrated beyond a limit, or a part was the wrong shape, that logic had to be written line by line. Now, trained models can recognise patterns and make those decisions automatically.”
Machine vision is one of the areas where this technology is already making an impact, enabling real-time image classification and automated inspection. “By giving a system the ability to see and evaluate what it’s doing, you make it more self-aware and responsive,” he says.
Still, Harrison emphasises that AI is not a universal answer. “Rule-based algorithms will remain the fastest and most reliable solution for many problems,” he says. “Machine learning adds value when the process is too complex or variable for traditional control logic.”
Future outlook
As AI continues to automate repetitive coding and optimisation, the engineer’s role is evolving. Harrison believes creativity and domain expertise will become key differentiators. “Customers will expect machines to self-optimise and self-diagnose as a baseline,” he says. “What will make one system stand out from another are the creative, problem-solving features engineers build on top of that foundation.”