With the rapid development of artificial intelligence, more and more engineers working in mechanical design, automation control, and equipment development are beginning to consider a very practical question: can AI already complete the development of automation projects independently?

This question has attracted even more attention since the release of OpenAI Codex. Many people hope to use AI to automate the entire workflow, from mechanical concept design, 3D modeling, electrical design, PLC programming, and HMI development to equipment commissioning. But can Codex really do all of this?

Based on the current level of technology, the answer is no. Codex cannot yet fully replace engineers or independently complete an entire automation project. However, it can already play an important role in many key stages, helping engineers improve efficiency and reduce repetitive work.

Codex is essentially an AI model that can understand natural language and generate code. It was initially used mainly in software development, where it could automatically generate program code, analyze logic, identify potential errors, and suggest optimizations based on user requirements. As model capabilities continue to improve, Codex is gradually entering the industrial field and showing strong potential in mechanical design, electrical control, and automation development.

During the mechanical design stage, Codex can help engineers quickly develop design ideas. For example, if a user asks for a packaging machine with automatic film feeding, bag making, filling, and sealing functions, Codex can generate a functional block diagram, suggest mechanical structure options, analyze transmission methods, and provide servo system configuration recommendations. Although these suggestions still need to be verified and optimized by engineers, they can significantly shorten the concept design process.

In 3D modeling, Codex still cannot independently complete complex assembly designs like an experienced SolidWorks engineer. However, it can support modeling work by writing CAD interface scripts, generating parametric modeling logic, calculating key dimensions, and organizing standard parts lists. For example, when designing a conveyor, Codex can calculate roller length, motor power, sprocket specifications, and frame dimensions based on conveyor width, providing accurate parameters for later modeling.

Codex becomes even more useful during the electrical design stage. For automation equipment, IO lists and electrical component lists often involve a large amount of repetitive work. Codex can quickly generate input and output address tables based on the equipment configuration and organize them into a standardized format. It can also recommend PLC models, HMI models, servo drives, power supplies, relays, and other commonly used electrical components according to project requirements, greatly improving design efficiency.

PLC programming is currently one of the most valuable application scenarios for Codex. Engineers only need to describe the equipment logic, such as “after pressing the start button, the conveyor runs, and after detecting a product, it stops after a delay.” Codex can then generate flowcharts, state machine logic, and structured text code. It has a certain level of support for mainstream control platforms such as Siemens, Mitsubishi, Omron, and Beckhoff. In modern packaging machines, where state machine programming is widely used, Codex can quickly build program frameworks and help engineers reduce a large amount of basic programming work.

In addition to PLC programming, Codex can also assist with HMI interface design. By simply describing the requirements, it can generate page structure plans, alarm management interfaces, parameter setting pages, production statistics functions, and even multilingual switching logic. This can significantly reduce the workload of HMI development.

Codex can also provide assistance in motion control. For applications such as electronic camming, electronic gearing, multi-axis synchronization, and flying shear control, it can explain control principles, generate PLCopen motion control code, and help calculate motion curves. However, for high-performance motion control tasks in high-speed packaging machines, labeling machines, and complex robotic systems, experienced engineers are still needed for in-depth commissioning and optimization.

The stage that is most difficult for AI to replace is on-site equipment commissioning. During machine operation, many complex factors are involved, including vibration, noise, mechanical clearance, air pressure fluctuations, and product characteristics. These problems often cannot be accurately judged through data and code alone. For example, when packaging film drifts off position, Codex can suggest checking tension, guide rollers, and photoelectric sensor calibration. However, it cannot observe the machine’s real operating condition and quickly locate the problem like an experienced on-site engineer. Therefore, commissioning still relies heavily on human experience.

Overall, the greatest value of Codex is not replacing engineers, but helping engineers work more efficiently. It can assist in concept design, parameter calculation, electrical design, PLC programming, HMI development, and technical documentation, thereby shortening project cycles and reducing development costs.

In the future, as CAD, CAE, and PLC development platforms become more deeply integrated with AI, the automation project development process will become increasingly intelligent. The role of engineers will gradually shift from executors to reviewers and decision-makers.

The truly competitive companies will not simply be those that own the most advanced AI tools, but those that enable engineers and AI to work together efficiently.