Figma’s native – Figma has announced an AI agent built natively inside its collaborative design environment, meant to generate and edit interface elements directly on the canvas—while drawing on a team’s design system context. The company argues it boosts productivity without
By the time you can click an app screen in your Figma canvas and see a star appear—signaling that a natural-language instruction can reshape what’s already there—the work of designing starts to feel less like “prompting” and more like collaborating.
Today, Figma announced an AI agent built natively inside its collaborative environment. The company positions it as a shift away from disconnected. floating prompt boxes toward assistants that live in the same workspace as the interface itself. Figma says the system can generate interface elements and remove “mindless drudgery of pixel-pushing,” keeping creators focused instead of scattered.
The update, Figma argues, is more than an add-on. It is presented as a reengineering of the digital drafting board into an autonomous engine—inviting the marketing department. code-wranglers. and project supervisors to participate in the design process. In Figma’s framing, that changes who gets to be a creator, not just what tools they use.
The agent doesn’t just respond to vague wishes in a separate chat window. Figma contrasts its approach with other AI-powered UX exploration tools that create interfaces in isolated. “sterile” chat windows—often producing different screens for an app. In Figma’s version, the control is more granular. The company says the tool can work at the level of individual elements, down to radial buttons and icons.
The interaction is built into the workspace and elements. When someone clicks an app screen on the canvas, a star appears next to it—an in-product signal that the agent is ready to adjust the visuals with natural language. The user tells it what to do on the interface element being worked on.
Figma also describes broader capabilities beyond incremental edits. The agent can generate initial design layers. explore multiple visual directions. and change color palettes either for one element or an entire screen—and do so globally across the design. It can handle tedious formatting tasks in bulk. including changing spacing in a progress bar and applying those changes across all progress bars in an app.
Teams, Figma says, can deploy multiple agents simultaneously alongside human colleagues, with each agent controlled by different users. The system continuously reads the “room. ” referencing the team’s existing design system logic and the ongoing conversations happening on the canvas. Users can toggle between typing natural-language commands and manually manipulating the design. keeping the pace of creation in the same place where the work already happens.
That promise comes with a tradeoff—and the tension is hard to ignore.
Figma’s tool is clearly aimed at time savings for experienced designers. But the company also sees it as a way for non-designers to start designing. In theory. Figma describes that as a democratization of creation: someone with a clear idea. but without the skills or budget to hire execution. could still build something compelling.
The darker possibility sits beside the excitement. The same automation that can lift outcomes for novices could also dilute the craft of product design into a flood of safe. standardized output. The result. in Figma’s critic’s dilemma. is a future where creativity becomes easier but less distinctive—less guided by taste and more by what the system can produce on demand.
Figma’s chief design officer, Loredana Crisan, pushed back on that fear. She told MISRYOUM that automation doesn’t erase the artisan; it isolates where the artisan matters.
“When an agent can take you 80% of the way, that last 20% is how you stand out,” Crisan said. In her view, human taste is what creates true distinction. She added that automation raises the baseline quality for novices while “it also brings up the ceiling of what designers can envision.” Her argument ends up at a familiar destination: the more people who care about design—and the more support they receive to push it—the better.
Crisan’s explanation points to what she calls the “local context” inside Figma. Figma says it doesn’t rely on a single, generalized oracle. “Under the hood, we use a variety of models for different tasks, some off the shelf and some we’ve fine-tuned ourselves,” Crisan said.
More important than the models. she stressed. is the way the agent connects to the materials already sitting inside the workspace. “The ability to connect your own design libraries and reference other context on the canvas is what will make the agent outputs feel unique and relevant to your team. ” Crisan said. The agent, she argues, isn’t meant to operate in a vacuum.
At the same time, she made the human role explicit. “AI can help spark new ideas . . . but the designer is the one who picks the direction,” Crisan said.
Even with that guardrail. the jump from design to engineering is where tools tend to break—turning polished concepts into constraints. gaps. and rework. Figma’s recent advancements for the Model Context Protocol (MCP) server are meant to smooth the path from design to code. But Crisan faced direct skepticism about how an AI handles the friction when a layout. imagined by the system. conflicts with the realities of Cascading Style Sheets (CSS) or backend constraints.
Crisan didn’t dismiss the problem. “It’s true that sometimes visual concepts don’t translate well to engineering constraints. AI doesn’t change that dynamic entirely, but it does offer more tools to close the gap,” she said.
To close that gap, she points to the broader ecosystem around Figma’s tools. Figma Make, described by the company as a “vibecoding” tool, is presented as the bridge that lets teams convert raw layouts into interactive, programmable applications.
“With our MCP server, you can bring context in and out of Figma to generate design-informed code without losing intent and fidelity,” Crisan said.
What Figma is ultimately selling is not a magic wand. Crisan framed it plainly: “AI is a tool. and the output depends on how you drive it.” Treating it as a one-shot solution. she warned. leads to mediocrity. Used for grunt work. she argued. it can buy back time—time intended for the hardest part of the craft. the “final. most crucial 20%.”.
Figma AI agent agentic design collaborative design natural language editing design system context MCP server Figma Make vibecoding design to code interface elements