Rute Cotrim
September 11, 2025
Lately, I’ve been experimenting with Flutter and Android Studio in my free time to build a small app prototype. Naturally, I saw this as the perfect opportunity to try Vibe Coding (letting AI help write or assist with code) and see how it fits into my Product Designer's workflow. I have tried many ways for different purposes, from Replit to Cursor, Claude or the classic ChatGPT. This is not an article about the agents themselves, and while this might not be news to everyone else, here’s what I discovered about the actual process of Vibe Coding:
PROS
1. Fast development
AI can generate code almost instantly, which is a huge time saver for anyone familiar with coding. Even for someone like me, who already worked with Front-end full-time but now only codes more occasionally, Vibe Coding made it possible to get functional results much faster than doing everything manually.
2. Pair-programming vibes
Discussing ideas with the AI agent felt a lot like pair-programming. It often suggested approaches I hadn’t considered and helped me work through tricky problems more efficiently than brainstorming alone.
3. Quick debugging & iteration
One of the biggest perks was being able to test multiple solutions in minutes. I could experiment with different ways of structuring the app, try small tweaks and also quickly see which approach worked best.
4. Perfect for personal projects
Since this is something I am doing in my free time, Vibe Coding is ideal, as I didn’t need to worry about production-level constraints. It’s low-pressure and fun for dev side projects or even quick UI prototyping.
5. It fills the knowledge gaps
Vibe coding is great for handling those small coding challenges that aren’t your core expertise. It helps bridge gaps in your knowledge, letting you get a product out faster without getting stuck on minor issues outside your usual coding scope.
CONS/CHALLENGES
1. Constant code verification is key
Just like everything else related to Generative AI, the generated code can be a bit unpredictable. I often had to double check that it didn’t break existing features built before or introduce subtle bugs. This is especially important if you plan to integrate the AI-generated code into a larger project or even publish it publicly.
2. Time spent reviewing
While the AI speeds up initial development and debugging, I still needed to go through the code carefully. Some parts weren’t well-documented, others could be cleaner, and optimization for performance, security or scalability wasn’t always considered.
3. Limited learning for beginners
For those trying to learn coding, Vibe Coding can be a double-edged sword. Since the AI builds logic automatically, it’s easy to copy and paste without fully understanding the underlying concepts.
4. Prompt-dependent and context-blind
The AI does exactly what you ask, which means bad prompts lead to bad solutions. It doesn’t always understand bigger business or product goals as well. This means you still need constant human oversight to make sure the code aligns with your vision and product context.
5. Ethics & code ownership
For me personally, there's an open question about who really owns AI-generated code. If you’re using it in a product that could be monetized or shared publicly, it’s worth considering potential ethical or legal implications.
MY MAIN TAKEAWAY
Vibe Coding is fast, fun and surprisingly powerful, but it’s not magic, and it shouldn’t replace a developer’s full workflow. Keeping a careful eye on code quality, context and ethics makes it a serious productivity booster, especially for side projects, prototypes, or personal experiments.