by abduzeedo
I share my personal experience with vibecoding, using Gemini and Claude to build a custom photo editor without needing deep software engineering skills.
Design is shifting right under our feet. I feel like we are finally moving away from the era of static mockups and entering what many are calling the age of "vibe coding." To me, this term describes a brand-new way to create. It is a process where I get to act as a conductor. I use natural language to guide AI models, focusing entirely on the "vibe" and the final output rather than getting stuck in the weeds of specific lines of code. I recently set out to build my own personal photo editor to see just how far AI could take a non-engineer like myself.
I am a designer by trade. I understand logic, but I will be the first to admit that I lack the patience for traditional software engineering. This has always been a major hurdle for me, and I know it is a common one for many creatives. Over the years, I have managed to build a few of my ideas, but honestly, a huge number of them never saw the light of day. They stayed trapped in folders or as half-finished mockups because I couldn't bridge the technical gap. That ends now with vibe coding. In the past, if I lacked technical depth, my best ideas usually stayed on the canvas. Now, tools like Gemini 3.1, Claude Code, and Antigravity are completely changing the math for me.
I tested these models to build a functional web application from scratch. I didn't start with a complex architecture or a thick manual. I started with an intent. I used these models to bridge the gap between my visual taste and the technical execution. This project really highlighted the core of vibe coding for me: it is all about speed and experimentation. During the process, I moved between different platforms to find the best fit for my workflow. I used Claude Code CLI because I loved the simple interface. I explored Antigravity to test out different agent tools. This way of working isn't about mastering one specific language. It is about using the right AI agent for the right task at the right time.
The result is a photo editor that could I actually use, supporting features like custom blurs, halftone effects, and even video support. What I love most about this work is the lack of "corporate" polish. It feels raw, personal, and functional. I know my logic can be a weakness sometimes, and in a traditional setting, that would have stopped me in my tracks. In this new world of vibe coding, the AI handles the heavy lifting of the logic. This allows me to stay focused on the soul of the product. I can iterate on the UI/UX in real-time and see the changes instantly. It creates a tight feedback loop that used to be impossible for me to achieve as a solo designer.
I’m keeping this project open for anyone to improve. This spirit of collaboration is exactly what defines this new era for me. By sharing my progress and the YouTube demo, I want to invite you to fork the code and play with it. This project is proof that the barrier to entry has finally vanished. You don’t need a degree in computer science to build your own tools anymore. You just need a clear vision and the right prompts. As I look at the future of digital design, the "vibe" is easily the most important variable we have.
You can play with it at https://abduzeedo.github.io/reeded-glass-effect-generator/
