Figma Config India 2026

Figma held Config India this week. A lot of what was shown at Config in San Francisco earlier this year is now moving into open beta, plus a few new things. Here's everything that was announced:

Code layers

Code layers are now in open beta. They let you put a real, working prototype directly on the canvas, so you can explore code-backed ideas together with your team instead of everyone working in their own IDE. You can start from a design frame, a GitHub repo, a local folder, a live website via the Figma Chrome extension, or just a prompt. Want to try a different direction? Duplicate the code layer and compare both side by side, or use Extract designs to turn screens back into editable Figma layers.

Good to know: code layers use AI credits, but creating and editing them is free on paid plans until November 4.

Specs

Specs is a new home for context in Figma, basically Markdown that lives on the canvas next to the work it describes. Drag in any .md file or press D to create a new one, and embed frames, FigJam boards, or code layers right inside it. The Figma agent can reply to comments, and connectors like GitHub or Hex can pull in outside context. Once you're ready to build, paste the link to your spec in your coding agent, and all of that context comes along through the Figma MCP server.

Payload in Figma Make

Payload, the open-source CMS that joined Figma last year, is now built into Figma Make. When you prompt a website that needs content, Make will set up:

  • A database and file storage
  • User login and access control
  • APIs (REST, GraphQL, ...)
  • An admin panel where anyone can update content without prompting (just add /admin to your published site)

Figma hosts everything, but since it's open source, you can also take it with you.

Design system updates

  • ‍Guidelines (open beta): add guidelines to your published libraries so the agent follows your patterns and avoids anti-patterns.
  • ‍Code sources (closed beta): import code from a GitHub repo or npm package so the agent builds with your actual components.
  • ‍Centralized design systems (preview): bundle your design libraries, code libraries, and guidelines in one place. A system map shows changes and dependencies, and a system check flags gaps like missing descriptions or interactive states.

Figma MCP and agent

The Figma agent is now generally available on the canvas, and the Figma MCP server got faster while using fewer tokens. From your favorite AI tool, you can now create code layers, edit specs, generate images, and run Weave workflows. There are also 20 new partner integrations, including OpenCode, Dia, Junie by JetBrains, and Meta Muse.

‍

‍

Claude Dashboards and Claude Motion

Anthropic launched two new features in Claude this week, both in beta.

Claude Dashboards connects to your data platform (BigQuery, Snowflake, Databricks, ...) or a CRM like Salesforce and builds dashboards that keep updating as your data changes. You can click any number to see the query behind it.

Claude Motion turns a report, chart, or idea into a short animated explainer. What's great is that it doesn't use a video model; Claude writes the animation as code, so you can still change every word, number, and timing. Export it as an MP4 or open it in your video editor.

On top of that, Claude Docs, Slides, and Design are now out of beta. The standalone version of Claude Design stays available at claude.ai/design until December 14, so make sure to migrate your design systems before then.

‍

ChatGPT Intelligent UI

Generative UI is clearly the theme this week. OpenAI is bringing GPT-6 to everyone in ChatGPT, and with it comes Intelligent UI. Instead of just text, ChatGPT can now answer with charts, maps, forms, interactive diagrams, or even a small tool like a bill splitter or savings calculator. The model decides which format fits your question best, and sometimes that's still plain text. It's built on a library of native components that stream in while the answer is being generated.

‍

Open Intelligent UI

It didn't take long for an open-source alternative to show up. CopilotKit released Open Intelligent UI, which brings the same kind of generated interfaces to your own app, with any agent and any model. The best part is that it can use your own design system and components, so the result actually looks like it belongs in your product. It can build algorithm visualizations, 3D scenes you can rotate, charts, calculators, games, and maps, all running in an isolated iframe. You can also download any generated interface as a standalone HTML file.

Check out Open Intelligent UI on GitHub

‍

Audio and layer effects in Lottie Creator

Lottie Creator now supports audio. You can import a track and edit it right on the timeline with waveforms, volume keyframes, and metering, and video exports will keep the audio. There are also new layer effects: Drop Shadow, Gaussian Blur, Fill, Tint, and Tritone, which you can stack, reorder, and keyframe. And if you want a head start, right-click a layer and choose "Animate with AI" to start animating it with Motion Copilot.

‍

Sketch Ghent (beta)

Sketch's next big update is out in beta, with a name close to home. The biggest addition is component variants, so you can create one button component with properties like State, Theme, and Size instead of a pile of separate components. There are also sections, which work like mini pages on your canvas to group flows or variants, and they automatically tile layers you paste inside them. And to match the rest of the industry, Symbols are now called components.

Good to know: documents you open with this version won't work in older versions of Sketch, so keep that in mind if you share files with others.

Try out the beta