Why You Still Need a Developer in the Age of ChatGPT

Posted by Will Wakeford on August 13, 2026

AI can easily create “feel-good” code that looks shiny and seems to work well, giving you an immediate sense that a problem has been solved. There are situations where this is great - for example, visualising data or automating repetitive processes. However, there is a significant difference between creating something that works and creating something that is secure, maintainable, and appropriate for the software it is being built into. An experienced developer understands the software they are working with and, crucially, how its different components fit together. That knowledge is essential for making sure systems are built properly, remain reliable, and can be maintained as they evolve. A lot of the time, the hard part isn't the code.

Let's use OpenCart, our field of expertise, as an example. A non-developer store owner may want to change they way their OpenCart shop looks, so they ask ChatGPT to create some updated templates. Initially, everything works well and the website looks exactly how the store owner wanted it to. However, after a few orders some serious issues start popping up - analytics are no longer being recorded, the captcha on the contact form has stopped working, and the product filter extension has disappeared. It turns out the issues were all related to the new AI-generated theme files uploaded to the shop. OpenCart and many of its extensions rely on the existing application structure and use systems such as OCMOD and vQmod to modify or extend existing files. These modifications often depend on specific code being present in the original files. If an AI-generated template replaces or significantly changes that expected structure, those modifications may no longer apply correctly. To the store owner the new templates may have appeared to be purely visual changes, but to a developer they have altered the underlying architecture that the functionality of the site depends on.

This is just one example for one system, but issues like this are becoming increasingly relevant as AI is used more widely to create and modify software.

This isn't to say that AI is bad for development. For an experienced developer, AI can be an extremely powerful tool. It can help locate bugs, provide potential solutions to problems and help speed up tasks that previously might have taken hours.

In this sense, AI doesn't remove the need for a developer; but it does change how a developer works. Instead of spending time writing ever line of code manually, developers can utilise AI to handle routine implementation while they handle the architecture, problem solving and testing to make sure the solution is robust. The big difference is that a developer will remain in control, responsible and able to review and understand every line of AI-generated code.

In conclusion, replacing a good developer with some AI agents is almost certain to land you in trouble. For businesses that rely on software or an online store to operate, it's crucial to have someone with the knowledge and experience to make sure the code being written (whether that is by AI or humans) is actually right for the system.