Designer vs AI
AI in my toolkit
AI is here to stay, and it’s a powerful tool. But what can it actually do for my product design workflow?
Here is a comparison between my first full fledged project, a language learning app MVP that took me many months of work, and the same app created with prompts in Figma Make, the latest and greatest prompt-to-app tool, also known as vibe coding. The latter took me only three short sessions to produce.
I'll share the insights I got from this experiment but you can see the results for yourself.
Let me introduce you to SpeakTu, a modern language learning app.
SpeakTu’s main role is to help you find the perfect tutor for your language learning needs. An intelligent search page with filters makes it easy to discover exactly the tutor you’ve been looking for. SpeakTu also includes all the tools you need for your lessons: messaging, file sharing, video calls, and more. We even fun built communities!
The whole process of creating this MVP is thoroughly documented, but the point of this case study is to compare the final outcome: my MVP prototype of the app versus the fully functional app coded with Figma Make.
Having said that, if you're interested in the details click below to see the full case study.
Comparison
SpeakTu MVP from Junior UX Designer (2021)
Click here for improved interaction ▼
SpeakTu App with Figma Make (from 0 → 1 in 2025)
Key Takeaways
Main Takeaways
≫ The main purpose of creating a Minimum Viable Prototype is to test the concept and determine whether the design has the potential to solve the problem. In this context, AI tools can be especially useful for presenting more complex features or major overhauls fast!
≫ For smaller fixes and design changes, the classic design method often proves faster and more reliable than prompting. The reason is that you need the full environment and design styles in place to iterate effectively. When a design system is already established, a skilled designer can implement minor changes quickly and efficiently.
Good to know
≫ Vibe coding generates the underlying code infrastructure that developers would normally painstakingly build—almost like magic. But in reality, it’s not always that practical. Just as designers rely on their own design systems, developers stick to their familiar toolkits, which usually don’t line up with the AI-generated code.
≫ Classic workflow shines in precision, scalability, and maintainability, ideal for final products, production-ready apps, and complex systems.
Conclusion
In my day to day I use AI as an assistance for brainstorming, ideation and creativity but here I was trying to test if it can also do the design and UX part of things. It can but with plenty limitations. A Hybrid approach would be best to use, AI for rapid experimentation and classic UX workflow for consistency and quality.
Thank you for reading until the end.