Over the past year, I have navigated the evolving landscape of AI tools, integrating them deeply into my work as a Laravel developer. This journey started with my initial exploration of ChatGPT and has grown into a fully integrated part of my coding workflow. Here’s how it unfolded:

I began by using ChatGPT as a companion in my daily work. While I’ve been familiar with the tool since its launch, it wasn’t until last year that I decided to make it a core part of my development process. Using it all day long for questioning and ideation proved transformative. ChatGPT became an assistant that provided real-time solutions and suggestions that complemented my thought process and work style.

My journey didn’t stop there. I discovered Cursor, an editor with ChatGPT built into the coding environment. It was a fantastic experience, allowing me to code with unprecedented ease and fluidity. However, the tool came with its own set of challenges—it was expensive, and as its popularity grew, so did the number of timeouts. These limitations led me to search for alternatives.

Enter GitHub Copilot integrated with VS Code. This combination was a game-changer. The accuracy of suggestions improved, the stability increased, and my workflow became faster and more efficient. The way Copilot could read, understand, and even edit code without missing a beat revolutionized how I approached development. With Copilot, coding became more intuitive, almost like having a reliable technical assistant that could understand complex instructions and execute them in real-time.

Using AI bridged the gap between my thought process and the actual code I wrote. I realized that my role shifted; I was no longer just a coder but a technical client to my own projects. I needed to train myself not just in new frameworks or updates but in crafting better prompts and instructions. This mindset shift was profound—it wasn’t just about learning to code efficiently; it was about learning to communicate with AI more effectively.

Then came an unexpected surprise: an invitation to try GitHub Copilot Workspace in its technical preview. I had been following the news about this tool, expecting its launch in 2025, but suddenly, it was at my fingertips. Eager to explore, I connected client repositories and decided to put it to the test.

I wrote an instruction set as a client would, with a bit of technical detail but without specifying class names or complex logic. My prompt was simple: add new columns, link them to the correct model, and update my Filament Admin panel resources accordingly.

When I hit the brainstorm button, I was shocked by the result. Copilot Workspace created the migration flawlessly, added the columns to the right model, updated the fillable array, and inserted the correct column types in my Filament resources. It even labeled them in Swedish, although I had written my prompts in English. The only hiccup? A slightly incorrect label on one column. This level of precision and automation was beyond what I imagined AI could handle, and it opened my eyes to a new way of working.

Now, I use GitHub Copilot Workspace daily. I create pull requests that I review and potentially integrate, streamlining my development process. But what should I tell my clients about this powerful new tool? I decided transparency was the best approach, and integrating them into this process was essential. But how?

Here’s my idea: We’ve been using Basecamp for over a decade. Imagine a workflow where clients add their to-dos in Basecamp, which we pull via API and translate into English if necessary. We could have an intermediate editor for reviewing and adding technical context before sending the task to GitHub Copilot Workspace.

Copilot would then brainstorm a solution, create a pull request, and deploy it to a staging environment. We’d send the client a link with a translated description of the pull request for their review. They could test, approve, or request changes, creating a new iterative cycle of development.

This vision raises an important question: Can the traditional hourly billing model survive this kind of efficiency? Probably not. A new pricing model is on the horizon, one that reflects the value of output rather than time spent. Feedback and insights into this new approach are invaluable as we navigate this shift.

Could we take this even further by integrating FlareApp? Errors detected by FlareApp could automatically trigger a task in Copilot Workspace, which would suggest and create a pull request for fixes. This would add an additional layer of automation and error-proofing to our workflow.

The future of AI in development is here, and it’s already changing how we work. I’m excited to see how this new chapter unfolds and eager for feedback from others in the community. Can this model work for us all? Will it bring more flexibility and a better work-life balance while maintaining high client satisfaction?

I look forward to your thoughts and ideas.