Back in 2020, Figma became the undisputed winner in the race to become the global standard for product design tools. Adobe XD and Sketch threw in the towel, and haven’t been heard from since. Figma’s success came down to its game-changing collaboration features and a steady stream of updates like variants, auto layout, and components. These tools completely transformed the way we design.
But there was always one thing it was really, really bad at: prototyping.
If you wanted to build a prototype that actually replicated a real product experience – not just a basic click-through – you need conditional logic and variables. Things like “if the user clicks this, then do this, else do that” or “if they enter this value, carry it over here.” When Figma announced a big prototyping upgrade at Config last year, the hype was real. We all thought, this must be the one we’ve been waiting for. But inside our design team, the launch was an anticlimax of Game of Thrones finale proportions. It still didn’t even support input fields. Seriously?
The struggle with other tools
Tools that offered the level of prototyping functionality we needed just didn’t fit into our workflow. We still wanted to design in Figma, which had now gotten so far ahead of the competition that designing in another application was almost inconceivable. But these other tools required either an awful import process for Figma designs, or rebuilding designs from scratch. Both options were wildly inefficient. So we did what we could within Figma and built one-off bits in other applications when absolutely necessary.
Then came the wave of AI-powered design tools. Every one of them promised the same thing: import directly from Figma, upload your design system, instant results. But when they were actually put to the test, it was painful. Recreating designs through prompts alone felt like shooting in the dark. Even as we got better at writing prompts, editing the output was still frustrating. One small change could completely mess up everything else.
What made it worse were the moments of brilliance. These tools would show real potential with something complex, then totally fall apart when you asked it to do something trivial.
Figma Make arrives
Then in May, Figma announced Figma Make, their own AI-powered tool. Naturally, we got excited again. The dream was back. The idea that you could design inside Figma, then click a button to turn it into a fully functional prototype, complete with input fields no less, felt within reach.
At least, that’s what we thought.
The first time we pasted in a Figma frame, we were instantly disappointed. A warning sign should have been that you were required to describe the design in words, as well as pasting it in. Granted, it did work better than some of the other tools we had tried, but it still felt like robot one was describing the design to robot two, and robot two was the one making it based purely on the description. Why couldn’t robot two just look at the design?
And maybe the weirdest part? It felt completely disconnected from the rest of Figma. None of the seamless integration that made Figma such a force years ago. I have to admit, I lost faith at this point. But others on the team persevered. And I’m so thankful that they did.
A month or so later, when one teammate said, “You know it’s not actually that sh*t anymore,” (which, for us, was a glowing review), I decided it was time to give it another go.
Surprisingly… it works
And she was right.
It does still feel rubbish whilst you’re using it. You enter a prompt, then sit through this weird wait whilst it processes it. Not long enough to do anything else, not short enough to ignore. Just… awkward. And it’s this waiting that makes the process feel slower than it actually is.
Because after a few hours of this, I suddenly realised I had built a pretty complex prototype that worked exactly how I wanted. It could do things way beyond what Figma’s native prototyping ever could, and it would have taken weeks to build in something like Axure or Framer.
This was what we’d been hoping for. It just didn’t look like it at first glance.
I kept going, and spent a few days fleshing out the prototype so that it fully replicated a complex new feature for our B2B SaaS platform. It’s still early days, but here’s my hot take:
1. It still starts with a design
Designing something from scratch using only text prompts is still slow and frustrating. The best starting point is still a classic Figma file. I think that’s going to remain true for quite a while. The more specific your prompt is, the better the result. And it’s very hard to be specific without an initial design.
2. It still feels like a craft
One of the biggest fears with AI tools is that they’ll take the craft out of what we do. The joy of tinkering with the details. The process of refining something, and ultimately taking pride in it. But writing text prompts instead of using a UI is just a different way of working. In a world flooded with bland, AI-generated content, that obsession with quality and detail matters more than ever. It’s what makes your work stand out.
3. It’s only going to get better
I built this prototype faster than I ever could have done manually. And not only that, the end result is much better than anything I would have produced without Figma Make. When you can simulate the real user experience of a feature, you notice things that would otherwise get missed. You come up with better ideas. You experiment more. It’s actually really satisfying, and it feels like the tool we’ve been missing. And it’s still (at the time of writing), only in Beta.
Although other AI design tools do still have a look in, the AI product design tool race does feel like Figma’s race to lose. It’s already ingrained into our design process (and budget), and it currently hosts all of our design systems and files. The dream for me is that the design team can work on different prototypes that all dynamically pull from a single design system, and that updates can seamlessly ripple through as they can already with global styles and components. If this is where they’re going with it, then it’s hard to imagine that a competitor would be able to achieve this.
I believe that it was John Hegarty who once said we tend to overestimate new tech in the short term and underestimate it in the long term. AI is no exception.
Based on my understanding of these new tools, I still don’t believe the product designer role is disappearing anytime soon, despite what the LinkedIn echo chamber says. But it is going to change, and fast. But then again, it always has. Not long ago, we were all “web designers.” UX wasn’t even part of the curriculum a few years back.
Honestly, the week I’ve just had – going 300 versions deep into this prototype – has been both wildly productive and genuinely fun.
Which, really, is all you can ask for from a working week.