Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

Figma rolls out new features that define how agents work on its platform

Figma used its first Config event in India to announce features that define how its own and third-party agents should work inside a company's design system, led by the Markdown-based Figma Specs editor. The company also moved its code layers feature into public beta and said its MCP now supports more than 20 new tools.

Published
Figma 发布面向 Agent 的新功能:用 Markdown 规范定义设计规则
Image source: techcrunch.com

Figma used its first Config event in India to announce a set of features aimed at giving agents a formal definition of how they should work inside a company's design system. According to TechCrunch, the design platform introduced Figma Specs, a Markdown-based editor for defining design language, brought in design and code libraries, and integrated the Payload headless content management system.

Figma Specs is meant to act as a central source of context. It can live as a standalone file inside Figma or sit directly on the canvas, and beyond plain text it lets users reference elements on the canvas and embed FigJam diagrams, code layers and designs. Teams can pull in design libraries, code libraries and guidelines so that both internal and third-party agents check their output against one shared system.

The other headline feature is code layers, which adds code to designs on the canvas for more interactive prototypes. Figma announced it in private beta back in June and is now opening it to public beta at the India event, letting more users test it out. Users can also ask the agent to turn a design into code, or change part of that code to iterate on a prototype.

On the ecosystem side, Figma said its MCP now supports more than 20 new tools, including Open Code, Dia, Grok Bot and the Indian AI startup Sarvam. Payload, the open-source headless CMS Figma acquired a year ago, is being integrated with Figma Make, the product for generating dynamic websites, so teams can manage content directly.

Together these moves address the unpredictability of AI-generated work. Figma opened its on-canvas agent to a wider user base only in the past week, and because generative models are non-deterministic, the company is supplying structured specs, design systems and code layers to steer them toward more predictable outcomes.

Figma's chief design officer Loredana Crisan told TechCrunch that Figma is the place where software teams come to decide what they build and ship. She noted that 17 times more lines of code are being produced today than in the past, while releases have grown only 1.3 times, meaning much of that code is not really built to ship but built to discover what the product is.

India was the venue and is also Figma's biggest market outside the United States. The company said its platform works with 40% of India's top 100 companies listed on the Bombay Stock Exchange, and it opened its first office in the country late last year with sales, partnership, go-to-market and product support functions.

The open questions now are whether documents like Specs become a de facto standard that third-party agents also follow, when code layers graduate from public beta to general availability, and how the Payload integration reshapes the path from design to a live site and its content operations.

Why it matters

The move signals that design platforms are turning into rule-setters for agents rather than just canvases for humans. Whoever owns the shared spec file shapes how AI-generated work gets produced and reviewed.

FigmaAgentDesign Tools
Back to realtime news

Nearby Updates

All

10/15, 10:46

Enterprise AI Provider Fooyo Launches FDE-as-a-Service to Push AI Into Production

Enterprise AI and automation provider Fooyo has launched FDE-as-a-Service, a packaged offering aimed at getting enterprise AI deployments past pilots and into production, as reported by StreetInsider. The service is built around the forward-deployed engineer model, in which engineers work alongside customer teams on site to ship working systems.

10/15, 10:45

OpenCode 2.0 Released, Letting Models Write Scripts to Orchestrate Tools

AI coding tool OpenCode 2.0 has officially launched, with its headline capability being that models can now write their own scripts to orchestrate tools, according to OSCHINA. That shifts coding agents from single tool calls toward scripted multi-step workflows, raising fresh questions about permissions and execution environments.

10/15, 10:13

Anthropic's 2021 internal research roadmap surfaces in copyright suit, sketching Claude Code years early

QbitAI reported on Oct. 15 that an internal Anthropic document marked "top secret" was exposed through a copyright lawsuit, and that it is a two-to-three page research roadmap written in July 2021. The file already named Claude Code, tool use, model self-judging, test-time compute, and trillion-parameter models with very large contexts, covering much of today's agent product language.

10/15, 10:04

DeepSeek hit by global outage across chat and APIs as a mysterious Harness account appears

DeepSeek's chat service and APIs were broadly interrupted on the morning of Oct. 15, with the official status page showing the cause identified and a fix underway for DeepSeek chat, the V4 Pro API, and the V4.1 Flash API. A new Bilibili account named DeepSeekHarness appeared at the same time, and the report states there is no official evidence linking the outage to a Harness update or a new model.