AI tools for frontend work have gotten way better lately and I’m starting to worry about my career. I used to think AI coding was pretty bad - tools like cursor would generate messy code that you couldn’t actually use in real projects.
But now things are different. Even cursor has improved a lot compared to two years ago. The bigger concern is all these new figma-to-code tools that are popping up everywhere. Me and my colleague tested kombai recently with a figma export and the generated code actually looked pretty decent. We also tried figma’s MCP with claude and got surprisingly good results.
You definitely still need a developer to review everything and fix issues, but I’m worried that companies might start hiring people who are great at design but weaker at coding, then just use AI to bridge the gap.
I know everyone talks about “AI replacing developers” but this feels different somehow. Should I spend more time learning design skills instead of just focusing on coding? What do you think about where frontend development is headed?
i get what ur saying, but don’t freak out! ai tools def help but they still struggle with the more complex stuff. like, code logic and optimizations still need a human touch. good design is great, but dont forget to keep ur coding skills sharp!
Interesting! Have you seen specific project types where these AI tools just don’t work at all? I’m wondering if there’s still a huge difference between basic landing pages and complex interactive features. What surprised you most about the Kombai results?
The shift’s real, but I see it creating opportunities, not killing them. I’ve worked with several AI-generated codebases this past year, and yeah, the initial output looks great. But here’s the thing - maintenance and scaling become a nightmare. AI nails isolated components but can’t maintain consistent architecture patterns across big apps. The devs crushing it right now? They quickly spot performance issues, accessibility problems, and maintainability red flags in AI code. Don’t pivot completely to design. Get good at code review, system architecture, and optimization instead. Companies will always need someone who gets the technical constraints and can talk to design teams about what’s actually feasible.