As an interaction designer, I’ve always been wary of code-as-prototype. The risk? Code rarely gets thrown away, even when better UX insights emerge later in the process.

But working on my own open source projects (esp. Docsify-This.net) has shifted my perspective. Generative AI significantly reduces the sunk-cost barrier of creating functional prototypes while delivering the highest fidelity possible - real, working functionality that users (and myself!) can actually interact with.

This is an interesting parallel perspective with Luke Wroblewski’s recent “AI Has Flipped Software Development” post on how AI is flipping traditional software development processes. The speed and disposability that AI enables may finally resolve the classic designer dilemma: getting true fidelity without being locked into suboptimal solutions.

For designers, this could mean:

• Testing with real functionality instead of simulated interactions
• Faster iteration cycles with working prototypes
• Less friction between design vision and technical reality

The question isn’t whether this changes design practice, it's how quickly we adapt our processes to leverage these new capabilities.

Link to Luke's article: lukew.com/ff/entry.asp?2112

This post is part of the 🌱 Sustainable Open Source Projects Pack - a collection of posts about building projects you can maintain in the long run.