Figma

Figma is a design and product development company whose shared browser canvas connects visual design, prototyping, and implementation. Its AI expansion includes Figma Make, which turns prompts or existing designs into working prototypes and web apps; an MCP server that supplies design context to coding agents; and tools that let agents edit Figma files. Founded by Dylan Field and Evan Wallace in 2012, Figma began NYSE trading in July 2025 and reported $1.056 billion in revenue for that year.[3][2][6]

Figma Make, launched in 2025, lets users move directly from a prompt to a working prototype, then refine it through further prompting, direct code edits, or visual manipulation. Prototypes and web apps can be published to a URL. Make Kits bring existing design systems and component libraries from Figma Design into Make; Copy Design moves a Make preview onto the design canvas for detailed editing and side-by-side exploration.[3]

Figma announced its MCP server beta on June 4, 2025, initially with tools for code, images, and variable definitions from a selection or node ID. Its contribution is structured design context: Code Connect can point an agent to the code file for an existing component, while variable metadata can identify the intended token even when several tokens share the same red color value. Screenshots supply broader relationships between screens and sections; interactive code representations describe behavior that a metadata tree alone may miss.[4]

The September 23, 2025 update added remote access without the desktop app, retrieval of underlying Make-file code, and in-app Code Connect mapping between design components and code files. Figma credited Anthropic, Cursor, Windsurf, and VS Code as partners for Make-file integration. These capabilities were announced in beta; importing external context back into Figma was still a future intention.[5]

On March 24, 2026, Figma opened canvas editing to Claude Code, Codex, and other MCP clients through use_figma. The complementary generate_figma_design tool translates HTML from live apps and websites into editable Figma layers; use_figma then creates or modifies assets using components and variables. Markdown skills specify workflow steps, sequencing, and conventions, with /figma-use as the foundation. Community examples included Ian Guisard of Uber’s screen-reader specifications and Firebender’s token synchronization with drift detection. Figma offered the API free during beta while describing eventual usage-based pricing.[6]

Figma announced its native design agent on May 20, 2026, with a gradual, selected beta rollout. The agent works on the canvas and in the left rail, allowing parallel prompts and manual edits while it iterates. Figma described it as fine-tuned for editing Figma files, drawing on frequently and recently used components and accepting explicit references to libraries, tokens, and variables. Its tasks include bulk edits, design-system documentation, and organizing feedback. The announcement distinguished its design-layer work from Make’s code-layer generation and said beta usage would not consume credits, with credits intended at general availability.[7]

Field’s July 2025 founder letter framed AI as a long-term investment, warning that spending could reduce efficiency for several years. His argument was that easier software creation increases the importance of design, craft, and point of view. Figma’s 2025 annual report said its AI products primarily relied on off-the-shelf foundational models and described further development of its own models as an investment plan.[2][3]

Before these agent products, Figma built a professional editing environment on the web and experimented with creation tools. That earlier work provides the context for the canvas and reusable design structures its AI products now use.[9][2][6]

Field and Wallace began experimenting with web design tools at Brown University and started Figma in 2012 with funding from a Thiel fellowship. Figma incorporated in Delaware that October. Field described their founding vision as “eliminate the gap between imagination and reality.” In his 2025 Latent Space interview, he recalled exploring computational photography, image editing, and converting 2D images into 3D scenes before incorporation, crediting most of the demo work to Wallace, Figma’s original CTO.[2][3][1]

Figma dates the launch of Figma Design to 2015, although Wallace’s December 7 engineering explanation said the product was still behind a waiting list. The 2015 editor used C++ cross-compiled through Emscripten to asm.js. Controlled memory layout allowed compact numeric representations; allocation in a preallocated typed array kept its C++ objects outside JavaScript garbage collection. Figma also built a custom tile-based WebGL renderer with its own DOM, compositor, and text layout engine to address rendering and text inconsistencies across browsers. Wallace called the result “a browser inside a browser” and recalled personally fixing high-DPI cursor support in Chrome, Firefox, and WebKit.[3][9]

Field recalled that early Figma was slow enough for one user test to take hours, so he brought a bottle of wine. The participant, Payam, then a designer at Coursera, sent eight to ten pages of feedback the next day. Looking back, Field recognized substantial overlap with the eventual roadmap, while stressing that the team had not literally followed the document. He used the story to explain his preference for surfacing users’ voices and finding people who could see beyond local feature requests.[1]

Figma expanded beyond interface design with FigJam in 2021 for whiteboarding and ideation, Dev Mode in 2023 for developer inspection and implementation, and Figma Slides in 2024 for collaborative presentations. Its four new products in 2025 extended the workflow further: Make for functional prototypes, Sites for publishing websites, Buzz for template-based marketing assets, and Draw for finer vector editing within Figma Design. Sites and Buzz were still described as beta in the annual report dated February 18, 2026. Figma also acquired Payload in 2025 to expand its content-management and application capabilities.[3]

An earlier AI feature, Make Designs, was distinct from Figma Make. Noah Levin’s July 17, 2024 retrospective described off-the-shelf GPT-4o and Amazon Titan models without additional training or fine-tuning. Metadata from commissioned mobile and desktop design systems, example screens, and the user’s prompt guided component assembly into parameterized designs; Titan generated images. After designer Andy Allen reported weather-app outputs resembling Apple’s app, Figma removed problematic assets and disabled the feature. Levin attributed the issue to inadequately vetted components and examples added before Config. The feature later returned as First Draft.[8]

On October 3, 2025, Figma acquired Weavy, which joined its platform as Figma Weave. Weave combines AI media models and professional editing tools on a browser canvas. Its node-based workflow lets users branch, remix, and refine generated outputs with operations such as masking, lighting adjustments, and color grading, feeding each output into the next stage of a media pipeline.[3]

Figma entered an agreement to be acquired by Adobe on September 15, 2022. The companies mutually agreed to terminate it on December 17, 2023, citing their joint assessment that there was no clear path to the required regulatory approvals. Figma received a $1 billion termination fee on December 20, recorded as other income.[3]

In 2024, Figma released eligible employee and former-employee restricted stock units and raised approximately $419 million through a primary share sale to cover the associated employee tax liability. A separate secondary tender offer closed July 3: new and existing investors purchased $566.7 million of shares from investors, employees, and former employees. In December, Figma issued approximately 2.5 million Class A shares to existing investors for $60 million in proceeds.[3]

Figma’s Class A shares began trading on the NYSE as FIG on July 31, 2025. The IPO completed August 1, with approximately 12.5 million company-issued shares at $33 each and $393.1 million in proceeds after underwriting discounts and commissions, before offering costs. Selling stockholders sold a separate 30 million shares, including the over-allotment, from which Figma received no proceeds. Field cited community participation in ownership as a reason for going public. Public ownership retained concentrated voting control: Figma reported that Field held approximately 72.3% of voting power at year-end, including shares covered by Wallace’s proxy.[3][2]

Figma reported revenue of $1.055788 billion in 2025, up 41% from $749.011 million in 2024, and 1,886 employees as of December 31, 2025. It also reported a 2025 GAAP net loss of $1.250463 billion; the IPO triggered a one-time cumulative stock-based compensation expense of $975.7 million for vested restricted stock units.[3]

At year-end 2025, Figma reported 13,861 paid customer accounts with at least $10,000 in annual recurring revenue and 1,405 with at least $100,000. These are separately billed accounts, so one organization can count more than once. ARR annualizes active agreement values and is not a forecast of recognized revenue. Its reported 136% net dollar retention compared the ARR of the same customer cohort that had met the $10,000 threshold twelve months earlier, incorporating expansion, contraction, and churn.[3]

As of December 31, 2025, Figma sold access through monthly or annual per-seat subscriptions, alongside free viewing and a limited free Starter plan. In March 2025, it changed seat upgrades to require administrator approval before provisioning and introduced multi-product seats. It introduced AI credits across seats during 2025; its February 2026 annual report described credit-limit enforcement and additional credit purchasing planned to begin in March 2026.[3]