The AI Tooling Driving Our Web Dev Workflow

Our engineering team is CMS-agnostic, with expertise spanning major platforms such as WordPress, Webflow and Drupal. Our growing passion lies in leveraging headless technology stacks like Next.js to craft custom, microservices-driven solutions tailored to enterprise needs.

Although AI adoption in web development raised a lot of questions in terms of accuracy, security and dependency, our team was excited about its potential.

Since integrating AI tools like GitHub Copilot and ChatGPT into our workflow, we have seen some improvements in coding efficiency and team collaboration. For example, a moderately complex custom block that previously took about five hours to complete can now be done in around three to four hours. This represents a 20% to 40% reduction in development time for these types of tasks.

Here is some direct feedback from the team on how AI has enhanced their work:

“Copilot is very helpful when converting traditional CSS or SASS code into Tailwind CSS classes, or transforming jQuery into vanilla JavaScript.”

“I used ChatGPT to help rewrite a complex algorithm for a navigation menu layout, specifically its element hierarchy. Without ChatGPT, it would have taken much longer, as I’d have to analyse each line, check with console logs, and open the browser to search for solutions when having syntax issues.”

“I use AI for debugging and code conversion. Recently, I used it to get an overview of plugins available for specific client requests on a WordPress site. I got a more concise, useful list which you don’t tend to get with google.”

Sometimes, using Copilot feels like having an extra developer on your side helping with repetitive tasks. We only need to review and sometimes modify the AI-generated code to suit our needs. This is not necessarily a bad thing, but it requires to be on top of the solution to not end up with technical debt.

Just as we don’t expose environment secrets, when it comes to security we ensure sensitive data is not exposed when using these tools. We anonymise sensitive information and ensure AI is disabled in confidential projects.

Overall, our current approach is to use AI to increase productivity writing code and documentation, always ensuring there’s a developer in the loop checking if the output produced by the AI matches our standards.