# Figma turns its design canvas into a coding environment at Config 2026 > Source: > Published: 2026-06-24 17:14:17+00:00 *Figma's Config 2026 announcements collapse the design-to-engineering handoff into a single canvas, putting the company in direct competition with AI coding tools like Cursor and Replit.* For years, the handoff between designer and developer has been one of the most reliably painful parts of shipping software. A designer finishes a mockup in Figma, exports assets, writes a spec, and hands it to an engineer who then rebuilds it from scratch in code, guessing at spacing, reinterpreting intent, and inevitably producing something slightly wrong. That friction costs product teams weeks per cycle, and every startup that has ever tried to move fast has felt it. At Config 2026 in San Francisco this week, Figma moved to end that pattern entirely. The headline feature is Code Layers, and it's the sharpest signal yet of where Figma wants to go. Designers can now clone a GitHub repo and surface design-to-code flows directly on the canvas, without needing production-quality assets first. From a code layer, you can drag out editable design elements, adjust them visually, and convert them back to code. Figma Make, the companion system, goes further: connect it to your local codebase, describe a change in natural language, and an AI coding agent makes the corresponding code edits, then opens a pull request, all without touching a terminal. That's not a design handoff tool. That's a development environment with a design interface on top. The competitive implication is hard to ignore. Cursor built its following among professional developers by making AI assistance feel native to a coding workflow. Replit offered a cloud environment where anyone could scaffold and ship an app quickly. Figma is now coming at the same territory from the design side, which means it arrives with something neither of those tools has: the actual design system already loaded. Your components, your tokens, your brand, all present before a single line of code is written. For a solo founder or a two-person product team, that's a meaningful difference. Code Layers got the attention, but the motion and visual effects announcements are what signal how aggressively Figma is trying to eliminate the entire third-party tool ecosystem that grew up around it. Figma Motion now lets teams coordinate animations on a shared timeline, with comments and collaborative editing built in. Previously, that kind of work lived in After Effects or Lottie, then got exported, integrated, tested, and iterated on in a separate loop. Now it stays on the canvas. The 3D transforms feature adds genuine spatial depth to the design surface. Until now, depth in interfaces was largely faked with stacked shadows and layered elements. Designers can now shape perspective directly inside Figma. And the new AI shader tools, built on WebGPU, let you describe an effect you want and have an agent generate it with adjustable controls: dithering, frosted glass, polished chrome, pixelation. That used to require a specialist. As TechCrunch reported on June 24, these effects were previously only possible through third-party plugins. They're now native. The AI Agents update is worth separating from the rest because it points at something different from design-to-code. Agent Skills lets anyone write a text prompt to create a repeatable custom command, something like /contrast-improvements or /resize-for-mobile, and share it with the whole team. Agents can connect to GitHub, Notion, and Excel, which means a designer can, in theory, trigger a workflow that updates a component, checks it against brand guidelines, and commits the change, without leaving the canvas. Whether teams actually build those workflows depends on how reliable the integrations prove to be, but the architecture is there. Don't mistake this for a slow category creep, either. Figma is moving on a clear thesis: that the AI coding tools eating into the software development workflow are vulnerable at exactly the point where design and code meet. Cursor and Replit are strong in their domains, but they start from code. Figma starts from the design system, which is where most product decisions already live. Frankly, any tool that makes it easier to go from a design decision to a merged pull request without a handoff meeting is going to win time-starved teams over, and Figma now has a credible claim to that workflow end-to-end. For small teams in particular, this changes the calculus on what you need to hire. If one person can own the canvas, connect it to the repo, and ship working code with animations and production shaders without spinning up three separate tools, the cost of moving from idea to product drops. That's the actual promise of Config 2026, and it's more concrete than most annual conference keynotes manage to be. **Also read:** [The AI model that broke into NSA systems is now the one the NSA cannot use](https://startupfortune.com/the-ai-model-that-broke-into-nsa-systems-is-now-the-one-the-nsa-cannot-use/) • [Goldman Sachs leads $110 million bet on Taktile as banks move from AI pilots to autonomous decisions](https://startupfortune.com/goldman-sachs-leads-110-million-bet-on-taktile-as-banks-move-from-ai-pilots-to-autonomous-decisions/) • [OpenAI and Broadcom unveil Jalapeño, a custom inference chip that puts Nvidia's pricing power on notice](https://startupfortune.com/openai-and-broadcom-unveil-jalapeo-a-custom-inference-chip-that-puts-nvidias-pricing-power-on-notice/)