We’ve all been there. You just need to resize an image. That’s it. But to do that, you have to launch Photoshop, wait for the splash screen to load, navigate three different menus, and export. It takes way too much mental energy for a task that should take three seconds.

Sure, there are small web apps for this. But they’re usually missing that one specific feature I need, or they force me into a workflow that feels just as clunky as the big software.

I used to stare at my screen thinking, "If I could just put a button right here that does X, my life would be perfect."

If an actual developer heard me, they’d probably roll their eyes and say, "Everyone wants a different button, buddy. We can't build everything." And they’d be right.

But until recently, that "dream app" stayed a dream. I’m not a coder. I honestly don’t even know the difference between programming and coding.

But that era is over. We’ve entered the age of "Vibe Coding."

My Ambitions Were... Humble

While my childhood friends were dreaming of being a doctor to cure diseases or building rockets to land on Mars, my ambitions were much smaller: I just wanted to convert a 26MB PNG into an 800KB WEBP without a headache.

So, I started using LLMs (ChatGPT, Gemini, etc.). I realized I didn't need to learn programming; I just needed to explain what I wanted. I started building simple scripts to download videos, change formats, and even auto-change my company laptop wallpaper (don't tell IT).

I did all this without writing a line of code. I was just the middleman, copy-pasting code blocks from the chatbot to a text file. If it broke, I’d send a screenshot of the error back to the bot and say, "Fix this."

It was messy. The code structure was probably a nightmare. But it worked.

Entering the "Agentic" Era

Recently, things shifted again. I decided to try Google’s Antigravity tool. This is where I learned the difference between an LLM and Agentic AI.

An LLM is like a text message buddy—you ask, it answers. Agentic AI is like a coworker sitting at your computer.

I fed it a Python script for a simple image tool I’d made and asked for improvements. It didn't just dump a block of code at me to copy-paste. It analyzed the project, made suggestions, and actually edited and created the files.

That’s when I realized: I’m not just prompting anymore. I’m collaborating.

The "Image Converter Bro.app"

I didn't have to deal with the "dark world of the terminal" or figure out why a library wasn't installing. I just told the AI what I wanted:

• "Change the background color to #202020"

• "Make the buttons bigger."

• "Use this icon for Browse button."

In a single evening, I built a fully functional desktop app with a GUI. I even learned how to use GitHub during the process (something I thought I’d never touch).

Just Use It

There is a lot of noise right now about whether AI will take our jobs or ruin the creative industry. Those are big, valid debates.

But for me, right now? It’s about removing friction.

It’s about realized that you don't have to wait for a software company to build the feature you want. You don't have to tolerate a workflow that slows you down.

The barrier to entry for building your own tools has completely vanished. You don't need to be a "developer." You just need to be curious enough to ask, "Hey Chat, can we build this?"

Go build that button you’ve always wanted.