A designer’s look at Figma Make
As a designer, I basically live in Figma. It’s the industry standard for interface design, prototyping, and collaborating with developers. In recent years, it has expanded beyond that, and I’ve been curious to see where it’s heading.
One of its latest additions is Figma Make, an AI-powered tool for creating interactive prototypes. Now that it has been out for a good few months, I feel I have been able to put it through its paces properly and learn where it fits in the design process.
Here are the key things I found, including where it excels, limitations to be aware of, and what it means for my process:
How I work with it
My process starts before Figma Make. I build screens with a component library like shadcn/ui, arrange layouts, map flows, and decide on visual hierarchy. By defining the core concept first, I can then ask the AI to help me communicate what I’ve already figured out.
With the groundwork in place, I bring Figma Make in for specific, focused work. Rather than handing it a complex layout and hoping for the best, I use it to prototype individual features or interactions, the kind of thing that helps me understand how screens and components like buttons actually interact, not just how they look side by side.
I’ve found this specific approach delivers far better results. Each prototype works best as a spotlight on an experience rather than an attempt to build everything at once. When it’s ready, I can publish as a live site for stakeholders to explore on their own devices whenever they like.
What works well
Prototyping can be fast. Figma Make generates Tailwind CSS and uses shadcn/ui components (matching our stack at Zudu), so the gap between designs and prototypes is quite short.
A good example of this was building out a prototype flow for an onboarding experience. It’s the kind of isolated, end-to-end journey that suits Figma Make well. It has a clear start and finish, and a focused set of interactions in between. From there, I was able to quickly convert to responsive layouts for tablet and mobile, and copy frames back into a regular Figma file to refine further.
What to watch out for
The output can look impressive, but in my experience getting it to feel consistent with existing styles and components takes time. Colours, spacing, and typography often need manual attention before anything sits comfortably within a design system.
The results often shine at first, but I’ve found that when too much complexity is introduced, the tool can lose its way. Features can blur together and earlier wins get overwritten.
Sometimes it simply doesn’t behave as expected. When that happens, it’s easy to fall into a prompting cycle by tweaking the wording again and again, but this just eats into time better spent elsewhere. I’ve found the best move is to recognise that point early: restart, revert to a previous version, or step back into Figma and work through the problem manually. Knowing when to stop prompting is just as important as knowing how to prompt well. It’s also worth noting the credit cap, which became a constraint faster than expected, especially when iterating.
Finally, I’ve noticed Figma Make can produce something that looks good, but lacks the thoughtful decisions that make it truly work for users. A slick prototype isn’t the same as a well-crafted product, and it’s easy to confuse speed with real progress when things look polished from the start.
Where I’ve landed (for now)
Figma Make earns a spot in my process, but in a specific one: quick visualisation and focused interactive prototyping, definitely not end-to-end product design. If you’re going to try it, get your design fundamentals right first, the tool will make a lot more sense when you do.
As for what’s next in our Figma experiments, we’ll be exploring the MCP layer between Figma and Claude, treating Figma as the source of truth rather than just the starting point.
