I didn't come to this decision lightly. It took me most of 2025 to reach this conclusion,...

Larablog's resource offers insights, tutorials, and resources for Laravel developers. Readers can learn about PHP programming, Laravel framework features, and web application development. With tutorials, code examples, and best practices, Larablog provides  guidance and expertise for building modern and scalable web applications with Laravel.

Larablog

A personal announcement from Laracasts about a significant decision made after deliberation throughout 2025. The content appears to be a video episode discussing this conclusion, though specific details about the decision are not provided in the available text.

I'm Done

For me it boils down to:
Is AI more fun than coding: no.
Is AI more fun than unemployment: yes.

It’s certainly heartbreaking to see this. Whether the future of development is 100% AI or not (you never know!), it’s still going to disrupt the way we work for years to come.

AI code is a so ridiculously bad still. The problem is not that it can’t code. The problem is that it won’t say no or evaluate if the proposed solution takes up more cpu/database power/memory/storage then needed. I use AI for almost everything when coding. But I find myself having to delete a lot of it to rebuild it. I am myself head/senior AI implementation developer for a large company. Which is why I adopt it. But putting it online without rebuilding it is worst case scenario. Yes, use it to find your way when you want to implement a new feature. Cause it will give you new ideas you haven’t thought of. Use it to debug or help out, cause it’s great at it. But by god, don’t use its code. It is unreadable and overly complicated because of the exceptions, if’s, but’s and then’s it uses to make it work. Ai can code, but architecturally it is the worst. Reusable code structured in a way that is future proof is what it lacks. And every company will find that out once the funnels clog up and/or the hardware bills starts strangling them.

Wow, we can tell that the ai summary really doesn’t handle speech to text transcripts well at all. I think we really need a “flag this summarisation for being incorrect/needs improvement button”.