Figma uses AI to transform digital design

我们的全新“执行功能”系列聚焦于那些通过人工智能推动变革的领导者视角。

Figma 是一个让团队汇聚一堂，将创意转化为世界顶级数字产品和体验的平台。我们采访了 Figma 的 AI 产品负责人 David Kossnick，探讨了 AI 在设计中的影响、如何赋能创造力，以及如何为 Figma 员工打造 AI 流畅度。

你曾将 AI 描述为一种平台变革和核心能力。AI 正在如何改变设计？Figma 又如何在这场变革中定位自身？

随着 AI 让数字产品的创作变得前所未有的简单，优秀设计将成为越来越重要的差异化因素。但设计不仅仅是像素，更是一门工艺：包含同理心、工作流程理解和问题解决能力。

因此，AI 被深度嵌入 Figma 的各个环节——从产品内的文本编辑、图像生成，到自动重命名图层和网站视觉，帮助创作更快、更直观，也让更多人能够轻松参与。

同时，AI 也是一种平台变革。Figma 专为数字产品打造，这让我们能够从根本上重新思考工作流程。

举个例子，Figma Make 是一个从语言、图像或结构化框架生成生产级代码的“提示到应用”工具。它为程序员和非程序员——设计师、产品经理、工程师、市场人员——提供了更进一步的方式，能够原型设计和表达想法，不再受技术障碍限制。

很多人说 AI 是“副驾驶”，而非替代者。你怎么看这种关系如何赋能创造力？

Figma 以对工艺的深度执着著称——让用户完全掌控每个细节。借助 AI，我们的能力已超越视觉层，涵盖语言、视觉和代码，新增了代码作曲器和“代码图层”等工具，允许用户原生编写和发布 AI 辅助代码。

“AI 代理最令人兴奋的一点是，它们能帮你走得更远，完成大量繁琐工作，帮你起步。”

AI 代理可以处理繁琐事务，但许多工具在此之后限制了定制。在 Figma，你可以完全编辑每一层——无论是语言、视觉还是代码——以匹配你的愿景并保持工艺水准。我们还支持跨模态工作流程，无论你擅长代码、设计还是语言，都能以自己的方式工作，就像全栈开发而不失专长。

最终，Figma 制作的产品是为人类服务的。人类带来判断力、同理心和品味——这些品质使他们成为真正的驾驶员，而非仅仅是副驾驶。

“AI 会帮助人类更快地探索，走得更远，但所有的人类判断、同理心、工艺和品味，才是成为驾驶员的真正含义，而非副驾驶。”

Figma Make 和 Dev Mode MCP 服务器是将 AI 融入端到端工作流程的重要步骤。你们从设计师和开发者通过代码与 AI 交互中学到了什么？

设计师与开发者的协作关键在于交付真正可用的产品——不仅仅是避免沟通误差，更是确保用户价值。Figma Make 帮助团队验证和测试多种想法，当达成共识时，能坚定地构建正确的产品。

Dev Mode 通过结构化数据（如 CSS 和设计令牌）简化了交接流程，MCP 服务器更进一步，允许开发者调用编码代理，将设计稿转换为具备完整上下文的生产级代码，无需手动复制粘贴。

虽然 Make 主要用于原型设计，设计师往往能精准地提示交互，工程师直接复制代码，使其逐渐成为工程交接的产物。

更广泛地说，Figma 天生支持多人协作，这与早期多为单人使用的 AI 工具不同。现在，我们正朝着更具协作性的 AI 体验迈进，邀请更多人参与创作过程。

你如何看待支持协作和多人参与的 AI 工具？

多人协作是 Figma 的核心，Figma Make 和代码图层等工具支持实时协作——即使是与 AI 一起。两个人可以在同一文件中工作，看到彼此头像，并与 AI 助手共同创作，将会议变成共享的互动构建时刻。

图像生成也成为 FigJam 和 Slides 的亮点，帮助团队共同创作符合品牌的视觉内容或并肩迭代。还有文化层面，比如我们制作 FigJam 周年贺卡的传统，团队成员用 OpenAI 的图像编辑功能重新混合头像，创造有趣且个性化的致敬作品。这些创意仪式促进了团队的联系和精神，是大多数工具难以实现的。

随着更多设计流程（如图层命名、文案撰写、视觉搜索和生成）被 AI 驱动，你如何看专业设计师的角色演变？

工艺依然是最核心的技能——同理心、品味，以及探索和打磨的能力。随着尝试成本降低，人们能更深入地挖掘有效方案，让每个细节——从动画到交互——都成为卓越的机会。虽然噪音会增加，但优秀的工艺会脱颖而出。

我们也看到角色从执行者向问题解决者转变，角色融合，更多人成为创造者。设计师开始编写代码，未来属于那些能够独立将想法从概念执行到落地的愿景承载者。

我有一位媒体娱乐行业的同事，他描述了创意生成和提案的高昂前期成本——许多好点子在还没机会被采纳前就被筛掉了。现在，借助 AI，这一瓶颈正在缓解。创意者能更自由地探索和分享，点子激增。

这让我想起漫威宇宙中的奇异博士——他能看到所有可能的未来。AI 正成为设计的奇异博士：探索无数路径，选择最适合特定问题的那条。

你认为 Figma 的 AI 会解锁哪些此前不可能实现的用户或用例？

我们已经看到许多惊人的例子——甚至在正式发布前。在内部测试中，一位没有编码或设计背景的人力资源团队成员发现了 Workday API，利用 Figma Make 仅用两小时就构建了一个游戏：显示四张脸和名字，玩家需要匹配它们——这是一种帮助新员工认识同事的有趣方式。现在它已成为我们的入职流程一部分。

这是内部工具团队从未优先考虑的想法，但因为 AI 降低了门槛而得以实现。它证明了非技术人员也能凭借好点子构建真实可用的工具，有时甚至能直接部署，无需工程团队。

我们看到许多意想不到的用例，极具启发性。Figma Make 和 Figma Design 等工具让人们能够表达和激活那些原本可能沉睡的创意。

你们如何打造 AI 流畅度——那些让人“恍然大悟”的时刻，让人意识到自己能做以前做不到的事？有什么经验分享？

内部自用是我们文化的核心，我们全力投入 Figma Make。举办了“伟大的 Figma 烘焙大赛”——一场全公司范围的竞赛，鼓励大家构建酷炫项目，并在各时区举办现场即兴创作会。这种亲身体验帮助对 AI 好奇的员工建立信心，尤其是对新手。社交激励和现场指导极大促进了参与度。

此外，我们在公司内部推广了 ChatGPT Enterprise。它带来了变革——市场团队用它来完善提案、撰写邮件等，且在安全、注重隐私的环境中使用。

我们还举办 Maker 周——为期一周的黑客马拉松，向所有人开放，不仅限于产品团队。大家构建从视频、帮助文档到集成 Slack 的 GPT 等各种项目。它赋予每个人尝试、失败和学习的许可，尤其降低了非技术岗位的实践门槛。

这是更偏向哲学层面——营造 AI 流畅度文化，还是有具体的进展衡量方式？

Figma 的 AI 流畅度始于文化。我们招聘那些渴望尝试和探索新工具的人，并通过专门的时间和预算支持学习——无需强制。

“我们打造了一支愿意生活在未来的团队，也打造了一支不断追求改进、对新工具和新技术充满热情的设计师团队。”

我们强调成功案例，比如人力资源团队构建的 Workday 游戏，展示即使是 10 分钟的尝试也能激发真正的影响。迈出第一步往往是最难的。

为支持安全探索，我们创建了实验工具的合规快速通道——对数据使用设有护栏——让团队能无障碍测试新 AI。大多数工具不会完美，但降低尝试成本有助于发现真正价值，推动整个组织的创新。

你分享了很多关于内部 AI 流畅度建设的见解。那么在消费者端，公司应如何将 AI 融入产品和体验？

作为 AI 用户和构建者，我们发现自下而上的实验推动了采用。员工开始非正式使用 ChatGPT，进而产生了对安全、支持路径的需求，最终促成了 ChatGPT Enterprise 的推广。

关键是：一旦人们尝试了 AI 工作流程，发现它们多么简单，就会感到有能力去构建。这种心态转变是推动有意义 AI 采用的关键，无论是在公司内部还是面向客户。

最后，能谈谈你个人在 Figma 的工作流程中如何使用 AI 吗？

我每天使用 ChatGPT，处理从整理评审笔记、撰写沟通稿到深入调研的各种事务，经常用“这个问题通常如何解决？”来快速探索解决方案。

我还依赖 Figma Make 进行原型设计和想法探索，使用 Slack AI 来总结复杂讨论，保持团队协同。最后，我经常用 Grammarly，虽然它不那么显眼，但它默默地通过一键操作提升我的写作质量。

Figma 利用 OpenAI API 支持 FigJam AI 及其平台上的图像生成功能，并在整个组织内部部署了 ChatGPT Enterprise，以提升员工的 AI 流畅度。

Our new Executive Function series features perspectives from leaders driving transformation through AI.

Figma is where teams come together to turn ideas into the world’s best digital products and experiences. We spoke with Figma’s David Kossnick, Head of AI Products, about the impact of AI in design, empowering creativity, and building AI fluency for Figma employees. 

You’ve described AI as both a platform shift and a core capability. How is AI changing design, and how is Figma positioning itself in that transformation?

As AI makes it easier than ever to create digital products, great design will increasingly be a key differentiator. But design isn’t just pixels; it’s craft: empathy, workflow understanding, and problem-solving.

That’s why AI is embedded throughout Figma—from in-product text editing and image generation to auto-renaming layers and site visuals—helping make creation faster, more intuitive, and accessible to more people.

At the same time, AI is also a platform shift. Figma is purpose-built for building digital products. This lets us rethink workflows from first principles.

One example is Figma Make⁠—a prompt-to-app tool that generates production-grade code from language, images, or structured frames. It gives coders and non-coders alike—designers, PMs, engineers, marketers—a way to go further, to prototype and express ideas without being blocked by technical barriers.

There’s a lot of talk about AI as a co-pilot, not a replacement. How do you see that dynamic empowering creativity?

Figma stands out for its deep commitment to craft—giving users full control to refine every detail. With AI, we’ve expanded beyond the visual layer to include language, visual, and code—adding tools like a code composer and “code layers” that let users write and publish AI-assisted code natively.

“One of the really exciting things about AI agents is they can get you really far, do a lot of the busy work, get you started.”

AI agents can handle the busywork, but many tools limit customization after that. At Figma, you can fully edit every layer—language, visual, and code—to match your vision and uphold craft. We also support cross-modality workflows, so whether you’re strongest in code, design, or language, you can work your way—like going full-stack without losing your specialty.

At the end of the day, the products made in Figma are for humans. And humans bring judgment, empathy, and taste—qualities that make them the true pilot, not just the co-pilot.

“AI is going to help humans explore much faster, go much further in their ideation, but I think all the human judgement, empathy, craft, taste is what it means to be the pilot not the copilot.”

Figma Make and the Dev Mode MCP Server were major steps in integrating AI into end-to-end workflows. What have you learned about how designers and developers want to interact with AI through code?

Designer-developer collaboration hinges on shipping what actually works—not just avoiding miscommunication, but ensuring user value. Figma Make helps teams validate and test many possible ideas so when they do align on a solution, there’s strong conviction to build the right thing.

Dev Mode⁠ streamlined handoff with structured data like CSS and tokens, and MCP⁠ takes it further by letting developers invoke a coding agent that translates mocks into production-ready code with full context—no manual copy-pasting required.

Even though Make is built primarily for prototyping, designers can often prompt interactions so precisely that engineers copy the code directly—making it start to become a handoff artifact for engineering.

More broadly, Figma has always been multiplayer by design, unlike early AI tools which were largely single-player. Now, we’re moving toward more collaborative AI experiences that invite others into the creative process.

How are you thinking about AI tools that support collaboration and the idea of multiplayer?

Multiplayer is core to Figma, and tools like Figma Make and code layers are built to support real-time collaboration—even with AI. Two people can work in the same file, see each other’s avatars, and co-create with an AI assistant, turning meetings into shared, interactive building sessions.

Image generation has also become a highlight in FigJam and Slides, enabling teams to co-create brand-aligned visuals or iterate side-by-side. There’s a cultural dimension too—like our tradition of making FigJam anniversary cards, where teammates remix avatars using OpenAI’s image editing to create playful, personalized tributes. These creative rituals foster connection and team spirit in ways most tools can’t.

As more design processes—like layer naming, copywriting, visual search, and generation—become AI-powered, how do you see the role of the professional designer evolving?

Craft remains the most essential skill—empathy, taste, and the ability to explore and refine. As the cost of trying ideas drops, people can go deeper on what works, making every detail—from animations to interactions—an opportunity for excellence. While noise will increase, great craft will stand out.

We’re also seeing a shift from implementers to problem-solvers, with roles merging and more people becoming makers. Designers are writing code, and the future belongs to vision carriers—those who can take an idea from concept to execution on their own.

I have a colleague in media and entertainment who describes the high upfront cost of idea generation and pitching—it required so much effort that many great ideas were filtered out before they even had a chance. Now, with AI, that bottleneck is easing. We’re seeing a proliferation of ideas because creatives can explore and share much more freely.

It reminds me of Doctor Strange in the Marvel universe—how he sees all possible futures. That’s what AI is becoming for design: a way to explore countless paths and pick the best one for a given problem.

What kinds of users or use cases do you think Figma’s AI will unlock that weren’t possible before?

We’re already seeing amazing examples—even before launch. During internal testing, someone on the HR team with no coding or design background discovered a Workday API and used Figma Make for just two hours to build a game: it showed four faces and names pulled from Workday, and you had to match them—a fun way to help new employees get to know teammates. It’s now part of our onboarding process.

This was an idea no internal tools team would have ever prioritized, but it came to life because AI lowered the barrier. It showed that non-technical people with great ideas can now build real, usable tools—sometimes even deployable—without needing an engineering team.

We’re seeing a lot of unexpected use cases, and it’s incredibly inspiring. Tools like Figma Make and Figma Design let people express and activate ideas that would’ve otherwise stayed dormant.

How are you building AI fluency—those “aha” moments when people realize they can do something they couldn’t before? Any learnings so far?

Dogfooding is central to our culture, and we went all-in with Figma Make. We ran the “Great Figma Bake Off”—a company-wide competition to build cool projects, with live jam sessions in every time zone. That hands-on support helped AI-curious employees build confidence, especially those new to these tools. Social incentives and live guidance made a big difference in helping people engage.

Beyond that, we rolled out ChatGPT Enterprise across the company. It’s been transformative—go-to-market teams use it for refining pitches, drafting emails, and more, all in a secure, privacy-conscious environment.

We also host Maker Weeks—weeklong hackathons open to everyone, not just product teams. People build everything from videos and help docs to Slack-integrated GPTs. It gives everyone permission to try, fail, and learn—lowering the barrier to hands-on experimentation, especially for those outside core technical roles.

Is this more philosophical—creating a culture of AI fluency—or are there ways you’re measuring progress?

AI fluency at Figma starts with culture. We hire people who are eager to experiment and explore new tools, and we support that with dedicated time and budget for learning—no mandates required.

“We’ve built a team that wants to live in the future. And we’ve built a team of designers that are constantly relentless and trying to find ways to make things better and excited about new tools and new technology.”

We highlight success stories, like an HR team building a Workday-powered game, to show that even a 10-minute experiment can spark real impact. That first step is often the hardest.

To support safe exploration, we’ve created a compliance fast path for experimental tools—with guardrails on data use—so teams can test new AI without friction. Most tools won’t work perfectly, but lowering the cost of trying helps uncover real value and fuels innovation across the org.

You’ve shared great insights on building internal AI fluency. But what about the consumer side—how should companies approach integrating AI into their products and experiences?

As both AI users and builders, we’ve learned that grassroots experimentation drives adoption. Employees began using tools like ChatGPT informally, which led to demand for a secure, supported path—ultimately prompting our rollout of ChatGPT Enterprise.

The big takeaway: once people try AI workflows and realize how easy they are, they feel empowered to build. That shift in mindset is key to scaling meaningful AI adoption—both inside the company and for customers.

To wrap up—how are you personally using AI in your workflows at Figma?

I use ChatGPT daily for everything from cleaning up review notes and drafting comms to deep research—often prompting it with “how is this problem typically solved?” to quickly explore solution spaces. 

I also rely on Figma Make for prototyping and idea exploration, and Slack AI to summarize complex threads and stay aligned across the org. Lastly, I use Grammarly constantly—it may not feel like AI, but it quietly improves my writing throughout the day with just a click.

Figma uses OpenAI APIs to power FigJam AI, as well as its image generation capabilities on its platform. It has also deployed ChatGPT Enterprise across its organization to enable AI fluency for its workforce.

Generated by RSStT. The copyright belongs to the original author.

Source