Between the moment I started writing this article and today, the landscape has changed again. New tools have shipped, others have evolved enormously, and concepts that felt experimental only a few months ago have already settled into product workflows.
We’re no longer talking only about an AI that generates a mockup or some interface copy. We’re now talking about agents that can:
- read a codebase;
- edit files;
- run commands;
- build prototypes;
- work directly inside a design canvas;
- move between a coding environment and Figma.
Claude Code is one visible example, but it’s only one among many: depending on the team, the technical constraints and the working environment, Codex, Mistral Vibe or other agents can serve the same purpose.
This blog is still a personal account, the perspective of a still-junior designer, curious but far from an expert, who is gradually discovering that the role is no longer limited to producing screens.
It’s expanding toward understanding the product, the code, the team’s workflows, and the tools that help turn an idea into something tangible more quickly.
The Landscape Has Changed
Two years ago, a UX/UI designer’s daily routine still followed a fairly readable path:
- Research;
- Wireframes;
- Figma mockups;
- Prototypes;
- Handoff to development.
In 2026, that path still exists, but it’s no longer as linear.
Google Stitch is pushing the idea of vibe design: you describe an intention, a mood, a use case, and the tool generates interfaces from natural language. Since its March 2026 overhaul, Stitch has become a genuine AI-driven canvas, you can now talk to it by voice, ask for real-time design critiques, and chain several screens of the same flow together.
Figma, for its part, is tightening the link between design and code with Figma Make, the Figma Agent and MCP. The canvas is no longer just a space where you draw screens: it has become a place where you clarify an intention, generate variations, work through states and where an agent can now write directly onto the canvas, respecting your design system.
For a broader look at today’s AI tools, I recently watched this video by theyo:
Framer has also crossed a milestone with its agents built into the canvas. With Framer 3.0, the tool is no longer limited to helping you design or publish a site: its agents can generate pages, adjust components, manage breakpoints, create effects, write code, connect the CMS… and every change stays real Framer work you can review in a branch before publishing.
So what’s changing isn’t just the list of tools. It’s the very nature of the workflow. Designers no longer work only with design software. They work with agents, canvases, coding environments, prompts, design systems, and much shorter iteration loops.
We have to stay clear-eyed: these tools are impressive for exploring quickly, producing first drafts, generating visual directions or building simple prototypes. But as soon as you move into a more complex product with specific user journeys, business constraints, product trade-offs and a real design system, AI isn’t enough on its own. It accelerates execution. It doesn’t replace thinking.
My Turning Point
At BearStudio, AI is no longer some distant thing to keep an eye on. It’s concrete and everyday, already shaping how we design, how we build, and how we collaborate.
At first, I discovered these tools mainly through code. The team had set up a Design Playground, a GitHub environment that let me experiment, understand the structure of a project, work in a terminal, spin up a local server, and take my first steps outside Figma. That’s where Claude Code entered my journey. But let’s be precise: Claude Code isn’t magic, and it isn’t “the” point in itself. It’s an AI coding agent. It can read a project, suggest changes, edit files, run commands, explain a codebase or help move a technical task forward. Tomorrow, that same role could be played by another agent depending on the context: Codex, Mistral Vibe, or an internal agent connected to the team’s tools.
So what struck me wasn’t only the tool. It was the shift in posture. Before, the moment I stepped out of the mockup, I felt like I was entering territory reserved for developers. Today, I understand a little better what happens behind the interface: I can test an idea, follow the steps, read the errors, ask the agent for help, then come back to a developer with something more concrete.
I’m not a developer. I don’t understand everything yet. But I’m progressing. I know how to start a project, read a file tree, understand what npm run dev means, plug in a template, open a PR, and above all ask better questions. To a developer, these might be simple gestures. To me, each one is a step closer to a more complete designer’s role.
The best example is probably my own design template. With resources found online and what the team taught me, I managed to put together a ready-to-use React base clear structure, reusable components that I can clone to spin up a site, an app or a prototype in minutes. A few months ago, I wouldn’t have thought I was capable of it.
Why Staying on the Sidelines Is No Longer an Option
The hesitation is understandable: AI moves fast, the tools change every month, and it quickly feels like you have to learn everything at once. But the landscape won’t wait. Teams that adopt AI ship faster and explore more directions, and designers who can bridge design and development become more autonomous and more sought-after.
The profession you’re stepping into today is no longer the one from three years ago. A portfolio of Figma mockups will soon no longer be enough: what will matter is understanding a product as a whole, using AI as a lever, and speaking the developers’ language even imperfectly.
The real risk isn’t picking the wrong tool or fumbling a prompt. It’s trying nothing, and watching the profession move forward without you.
Curiosity Makes the Difference
I don’t master all the tools mentioned here, and I still have days when everything moves too fast. But I’m no longer on the sidelines and it’s curiosity that made the difference, not talent or a computer science degree. That said, I didn’t do this alone: at BearStudio, I’m pushed to explore and given the space to learn. Being surrounded by people who believe in this shift changes everything.
The UX/UI design profession isn’t dying, it’s expanding. If this article sparks even the urge to try a tool or to ask a dev colleague how a repo works, that’s already a first step off the bench.
Keep reading
Curious about how we use AI on the technical side? Check out our dedicated article:
- Discovering MCP: a new approach for your AI agents (French only)
Interested in how we work with our clients using AI? Take a look at these articles: