11 months without Figma
I joined a geotechnical SaaS company as Lead Product Designer. We haven't opened Figma once. Here's what replaced it.
I joined Fugro 11 months ago to lead product design on GroundIQ, a geospatial intelligence platform. We haven’t used Figma once.
That sentence would have sounded insane to me two years ago. But here we are.
Why Figma stopped making sense
The team is AI-first. Everyone writes code. Designers and developers work toward the product, not their job titles. In that environment, Figma becomes a bottleneck. Design mockups are a static artifact in a world where the implementation can change three times before the mockup gets updated.
When I first joined, I tried the code-first approach: work directly in the real codebase. That failed. Real branches had data constraints and permission walls. You couldn’t freely experiment. You couldn’t rapidly prototype.
So I built a React/HTML shell repo. It looks like the platform but has no data restrictions. It deploys on GitHub Pages. Anyone in the company can see it, use it, and PR against it. It serves as both design mockup and developer handover, without the handover step. There is no handover. The design is the implementation spec.
What replaced the design tools
A living component library in code. Every component, every pattern, every interaction state lives in the shell repo. When a developer implements something, they pull from the same components the design team uses. There’s no translation layer between design and development. They’re looking at the same thing.
AI for research synthesis. User research, competitor analysis, industry reports, it all gets processed through AI that structures it into our knowledge base. Patterns emerge across sessions without someone manually tagging everything.
Voice notes for capture. I can send a 90-second voice note while walking between meetings and it gets transcribed, summarized, and filed into the right project context. This solves the ad-hoc conversation problem: the meeting that happens in the hallway and never gets documented.
PR-based design reviews. A developer opens a PR. An AI runs a heuristic UX audit as first pass. I review the results and spot-check. The feedback loop is minutes, not days.
What’s harder
The AI can build fast. Really fast. The problem becomes knowing what not to build. When everything is possible, the discipline is restraint.
Documentation also gets weird. When the design system lives in code and evolves in PRs, you have to be deliberate about capturing decisions. It’s easy to move fast and forget to write down why you moved.
And there’s the generation gap: our users are geotechs who work in printed PDFs. The internal process doesn’t mean much if the output still needs to feel trustworthy to someone who files paper reports. But that’s a different article.
Is this the future?
I don’t know if this is the permanent way of work. It’s still new. But I do know that going back to static mockups after working this way feels like trying to design a building with a crayon.
The designer’s job isn’t the artifact anymore. It’s the opinion.