Analysis
Figma used its Config event in India on October 14 to expand how AI agents work inside its design platform, adding a new Specs feature, wider code-layer tools and a deeper MCP integration, TechCrunch reported. Pulse has covered Figma's rockier months this year, including a stock drop tied to revenue guidance and executive departures -- this launch is Figma's attempt to show its AI roadmap is still shipping regardless.
What shipped
Figma Specs is a Markdown-based editor for defining a company's design language, usable as a standalone file or live on the canvas, that both human teams and AI agents can reference while designing. Centralized design-system context lets teams feed design libraries, code libraries and brand guidelines to agents so outputs follow house rules rather than generic defaults. Code layers -- in private beta since June -- moved into public beta, letting users ask an agent to convert a design directly into working code or edit code already on the canvas. Figma also expanded its agent-on-canvas rollout to more users and added a Payload CMS integration for generating and managing dynamic websites, plus more than 20 new MCP tool integrations including Open Code, Dia, Grok Bot and Indian AI startup Sarvam.
“Figma also said downloads of Payload, the open-source CMS it acquired last year, are up sixfold over 12 months, and that India is now its biggest market outside the U.S.”
The numbers Figma is leaning on
Figma's chief design officer, Loredana Crisan, said the platform now produces 17 times more lines of code than in the past, against only 1.3 times more actual releases -- a gap Figma frames as evidence that AI has made building cheap but exploration, not shipping, is where most of that new code goes. Figma also said downloads of Payload, the open-source CMS it acquired last year, are up sixfold over 12 months, and that India is now its biggest market outside the U.S. by users, reaching 40% of India's top 100 BSE-listed companies.
Figma going deep on agent tooling at an India-specific event, rather than its U.S. developer conference, signals the company is treating international enterprise adoption as a growth lever separate from its core U.S. AI narrative. The 17x-code-to-1.3x-releases gap is the more important number for competitors and customers alike: if AI-generated code isn't translating into shipped product, the bottleneck has moved from writing code to deciding what's worth shipping -- a design and product-management problem, not an engineering one.
Today's release is features, not usage data -- the real test comes when Figma discloses adoption numbers for code layers and the expanded agent once they're out of beta.
