I am not exaggerating with the title. Thirty minutes. A complete onboarding flow of four screens with transitions, real microcopy and state variations. What used to require half a day in Figma plus another half day polishing details now happens in a fraction of that time.
And what changed is not that I am faster but that the prototyping process transformed completely.
The prototype no longer starts in Figma
My process before was to open Figma, create the frames, design each screen, connect the flows and then show something that felt real. Now I start in Claude. I describe the flow I need, the user context, the project constraints and I ask it to generate the code for a functional prototype in HTML and CSS.
In minutes I have something I can open in the browser, that looks like a real app, that has basic interactions working. It is not pixel perfect and it does not have to be. It is a prototype to validate concepts not to deliver to development.
Prototyping to think not to deliver
That mindset shift was the hardest part. As designers we were trained so that everything that comes from our hands looks impeccable. But a prototype is not a deliverable, it is a visual question. And the faster you can ask that question the faster you get the answer you need to move forward.
Now I generate three or four variations of the same flow in an hour. Before I could do one variation in a day if things went well. That means I can explore paths I would not even have considered before because there was no time to build them.
What I do with AI prototypes
I use them for two things. The first is internal validation. Before investing time designing in high fidelity I show the AI-generated prototype to the product team or the client and see if the direction makes sense. If it does not work I lost thirty minutes instead of three days.
The second is quick testing. For informal usability tests where I just need to see if a user understands a flow, an AI-generated prototype is more than enough. It does not need to be pretty, it needs to be functional and clear.
Figma does not disappear but changes its moment
Figma is still where the final design lives, where components are documented, where the handoff to development happens. But it is no longer where I start prototyping. Figma became the refinement step not the exploration step.
And that is liberating. Because it allows me to arrive at Figma with a clear direction, with a concept already validated, with the confidence that what I am going to design in high fidelity has already been tested in low fidelity with real users.
Speed is not the point. The point is thinking faster. And prototyping was always thinking with your hands.