For the past few weeks, I’ve been thinking about AI nonstop. Yes, me who was skeptical and reluctant to use it, it’s been growing on me. In my almost decade-long journey of building software, I’ve never seen anything like it. And still, it feels like this is just the beginning.

The hype is real and is easy to get caught by it. But despite all the noise that might come with it, there’s one undeniably fact: the software game has changed.

It changed the moment LLMs started getting closer to our tooling. When they became integrated into our editors and IDEs – able to read, edit, and delete our source code. When they were given a web browser to check the docs of your library of choice. When they started running commands in your terminal to test things out and self-correct. The moment they were called “Coding Agents.”

And if you’ve been in the industry long enough, we might agree that what truly hooked us wasn’t any of that. It was speed.

What took countless hours to debug now takes minutes. We are able to iterate faster than ever by simply describing in plain language what we need. It’s truly amazing.

And yet, it’s still not easy. There’s way too much to learn and new skills to develop in order to become productive. It requires running experiments to steer them toward the expected results. Engineering principles and best practices must still be enforced in code and followed by agents. And regardless of their imperfections, I’ve seen them take me really far.

This first post is mostly about setting the ground of what I’ll like to share moving forward. Things are moving at a ridiculously fast pace. So fast that sometimes I feel like I’m being left behind by the overwhelming amount of software being created right this second. But that is what is making me excited and I’m looking forward to sharing what I’m building as well.