Summary: Figma is launching its own AI agent that operates directly on the collaborative design canvas, letting users generate, edit, and iterate on designs through natural language prompts. The move follows partnerships with Anthropic and OpenAI and the $200 million Weavy acquisition.
For months, Figma has been opening its canvas to other people’s AI. Partnerships with Anthropic and OpenAI gave coding agents such as Claude Code and Codex a direct line into the design tool via MCP. Now the company is shipping an AI agent of its own, one that lives inside the collaborative canvas and can generate, edit, and iterate on designs from a simple text prompt.
The assistant, launching first in Figma Design, lets users describe what they want in plain language and watch the agent produce it on the canvas in real time. Figma says users can run multiple agents simultaneously, each handling a different task, effectively adding AI collaborators to the same multiplayer workspace where human teammates already operate.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
The company claims its underlying models have been fine-tuned specifically for design work, giving the agent an understanding of layout, components, and visual hierarchy that generic large language models lack.
“Teams can now collaborate with agents on the multiplayer canvas to test out ideas, visualise edge cases, and refine concepts together without over-indexing on the more tedious parts,” said Loredana Crisan, Figma’s chief design officer, who joined the company from Meta last year after nearly a decade leading product and design teams across Messenger, Instagram, and Meta’s generative AI efforts.
The launch is the latest move in a rapid AI buildout at Figma. In February, the company struck back-to-back partnerships that embedded Anthropic’s Claude Code and OpenAI’s Codex into its design-to-development pipeline through MCP.
Both integrations let developers take a running interface and convert it into an editable Figma frame, or hand a Figma design to a coding agent for production-ready implementation. The new built-in assistant adds a different dimension: rather than bridging code and design, it makes AI a native participant in the design process itself.
That push has been underpinned by acquisitions. Last October, Figma bought Weavy, a Tel Aviv-based startup that had built a node-based AI canvas combining multiple generative models with professional editing tools.
The deal, reportedly valued at roughly $200 million, became Figma Weave, and AI credit monetisation from the product contributed to the company’s strong first-quarter results. Figma reported Q1 2026 revenue of $333.4 million, a 46 per cent increase year on year, with its net dollar retention rate climbing to 139 per cent, the highest in over two years.
The competitive context makes Figma’s AI bet feel less optional and more existential. Canva, which now claims 220 million users globally, launched its AI 2.0 platform in March with a proprietary foundation model built for design. Adobe’s Firefly holds 41 per cent business adoption. And a crop of AI-native startups, including Flora, Krea, and Dessn, are chasing the same audience of designers who want to move faster without sacrificing craft. Google, meanwhile, unveiled Pics at I/O 2026, an AI design tool built directly into Workspace that generates graphics from text prompts.
Figma’s advantage, if it has one, is the canvas itself. More than 690,000 paying teams already use it as their collaborative workspace, and the multiplayer architecture that made Figma dominant in the first place now doubles as the natural environment for AI agents to operate in. Where competitors are building AI tools that work on design, Figma is trying to build AI tools that work within design, sitting alongside human teammates on the same infinite canvas.
Whether that distinction matters will depend on execution. The company plans to extend the AI assistant to its other products over time and has signalled that it wants to pull design and code even closer together inside its apps. For now, the message is clear: the canvas that changed how designers collaborate is betting it can change how they collaborate with machines, too.
