Why Claude Wins Over Figma for Prototyping at Scale
Figma may have made a costly miscalculation by introducing Figma Make’s new credit pricing on March 18th. After months of testing their product, it’s clear that while it worked well during testing, iterative AI-driven design is not commercially viable for heavy prototyping at scale under the new credit plans. That’s why I’m now using Claude Code.
My Experience with Figma Make
Since launch, I’ve created multiple working prototypes using Figma Make, from a simple running calculator app to an MVP I demoed to a prospective client with 7 million customers. Across these projects, I’ve used tens of thousands of credits. Some of this was trial and error, and a portion came from correcting issues introduced by Figma itself. While this was manageable during free trials, enforced credit limits quickly made it unsustainable.
The point of AI prototyping is to test, iterate, and see what works. At our current usage, costs could add several thousand pounds a year on top of subscription fees. For larger teams, expenses would quickly spiral into tens of thousands.
Beyond cost, this model introduces a creative bottleneck: designers become conscious of credits rather than focusing purely on iteration. The trial-and-error process is what produces good design, not a perfect prompt constrained and critiqued by cost.
This isn’t about using Figma Make to “vibe code” an app. The vision and design are defined in Figma, then Make constructs and iterates on that idea. Most layouts and elements were created in Figma or via other methods. I defined the vision and used Make purely as a bridge to development. With the new credit pricing, that bridge is no longer cost-effective.
How I Migrated to Claude
Migrating a prototype wasn’t straightforward. It required a few hours of hands-on work, plus Claude running in the background through multiple iterations to reach a version comparable to the Figma Make prototype. Once the setup was complete, the long-term benefits became clear.
Here’s the workflow I used to migrate an existing prototype:
Connected the Figma MCP to Claude Code via desktop and terminal
Converted each key page into layered Figma files by sharing each frame with Claude - around 20 for this prototype
Asked Claude to build each Figma frame into HTML
Linked everything to create a functioning prototype
Iterated using Figma frames if needed or directly within Claude without returning to Figma
I’m also one step closer to checking out my app using branches, then I’m straight into the dev workflow. Figma Make showed what was possible; Claude is leading the way for production teams already using it and wanting design and development to operate on the same platform.
If I had started prototyping in Claude, the entire process would have been ten times easier, as designs could have been fed directly into Claude from the outset.
Cost and Usage Reality
The Claude Max model uses a set number of tokens within a 5-hour window. To put this in context: prototyping an app that uses 10,000 credits in Figma would exceed a standard seat allowance and cost $240/month extra. In Claude, once I had reached a prototype comparable in size, I had used less than 10% of my usage limit, which then resets. The Max plan starts at $100/month, and combined with Chat and CoWork, the value is compelling for teams already using Claude for coding.
Why Claude Makes Sense for Us
As a business, we are now fully locked into Claude. Figma Make was intended to bring the teams closer, and while it worked well to begin with, the new pricing and the fact that our developers weren’t using it turned moving away into an opportunity to align the teams even more effectively. Figma currently remains our environment for exploration and static UI design, but for prototyping, Claude is more efficient and cost-effective.
It’s disappointing because Figma Make had promise. As a designer, I found it the most natural platform to use after trying alternatives like Google’s Stitch and Pencil. I understand Figma must face cost pressures for processing AI prompts, but there must be a better way to deliver prototyping without prohibitive credit limits. We would likely have snapped up a few “premium AI subscriptions” with higher AI credit limit ceilings if those were $50/month more per seat.
Looking Forward
I expect more technically oriented design teams to move to Claude for prototyping or even abandon Figma entirely. Alternatively, Figma may already be ingrained so deeply in some organisations that migration to Claude for prototyping is impractical. The coming year will show whether Figma Make’s credit-based prototyping is sustainable and if Figma can maintain its relevance under this new model.