Executive Brief
Figma, the collaborative design platform used by millions of designers and developers, announced on May 7, 2025, a significant expansion of its AI capabilities. The update introduces generative design features that enable users to create interface components, modify existing designs, and receive automated layout suggestions through natural language prompts.
The new AI tools include a text-to-design generator that produces UI components from written descriptions, an intelligent editing assistant that can modify selected elements based on verbal instructions, and an automated layout system that suggests responsive arrangements for design elements. Figma stated that these features are available immediately to all users on paid plans, with limited functionality accessible to free tier users.
The announcement comes approximately 18 months after Figma's initial AI feature preview at Config 2023, which faced criticism over training data concerns. The company addressed these concerns by stating that the production AI models were trained exclusively on licensed datasets and publicly available design resources, not on user files stored in Figma.
Design tool competitors Adobe and Canva have both introduced AI features in their respective products over the past year. Figma's update positions the company to compete directly in the emerging AI-assisted design market. The company reported that over 4 million designers use Figma as of early 2025, making the AI rollout one of the largest deployments of generative design technology to a professional user base.
Figma CEO Dylan Field characterized the update as the beginning of a transition toward AI-augmented design workflows rather than a replacement for human designers.
What Happened
Figma's AI feature announcement on May 7, 2025, followed an extended development period that began with the company's Config 2023 conference in June 2023.
At Config 2023, Figma previewed several AI features including a text-to-design prototype. The preview generated significant discussion in the design community, with some users expressing concern that the AI models might have been trained on designs stored in Figma without explicit user consent. Figma paused the AI feature rollout in response to these concerns.
In September 2024, Figma published a detailed explanation of its AI training methodology. The company stated that production AI models would be trained on three categories of data: publicly available design resources with appropriate licenses, datasets licensed specifically for AI training, and synthetic data generated by Figma's internal teams. The company explicitly committed to not training on user-created files stored in Figma without opt-in consent.
The May 7, 2025 announcement marked the production release of the AI features. According to Figma's blog post, the release includes three primary capabilities.
The first capability, called "Generate," allows users to describe a UI component in natural language and receive a generated design. Users can specify elements such as "a login form with email and password fields, a remember me checkbox, and a submit button" and receive a styled component matching Figma's design system conventions.
The second capability, "Edit with AI," enables users to select existing design elements and request modifications through text prompts. Examples provided by Figma include requests like "make this button more prominent" or "adjust the spacing to follow 8-point grid."
The third capability, "Auto Layout Suggestions," analyzes design compositions and recommends responsive layout configurations. The system identifies groups of elements that could benefit from Figma's Auto Layout feature and suggests appropriate settings.
Figma stated that the AI features are powered by a combination of proprietary models and partnerships with external AI providers, though the company did not disclose specific technical partners.
Key Claims and Evidence
Figma's announcement included several specific claims about the AI system's capabilities and limitations.
On accuracy, Figma stated that the Generate feature produces designs that conform to established UI patterns approximately 85% of the time in internal testing. The company acknowledged that generated designs require human review and refinement before production use.
On training data, Figma provided documentation stating that no user files were used in training without explicit consent. The company published a list of licensed datasets used for training, including design resources from Unsplash, Google Fonts, and several commercial design asset libraries.
On performance, Figma claimed that the Generate feature produces initial results in under 5 seconds for typical component requests. More complex multi-component generations may take up to 30 seconds.
On privacy, Figma stated that prompts and generated outputs are not stored or used for model training unless users explicitly opt in to a feedback program. The company's privacy documentation specifies that AI processing occurs on Figma's infrastructure rather than being sent to third-party AI providers for the core generation features.
Independent testing by The Verge confirmed that the Generate feature produced functional UI components for common patterns such as navigation bars, card layouts, and form elements. The publication noted that generated designs sometimes required manual adjustment to match specific brand guidelines or accessibility requirements.
Pros and Opportunities
The AI features address several workflow inefficiencies that designers have identified in user research. Creating initial component layouts and establishing consistent spacing are tasks that consume significant time in early design phases.
For design teams with established design systems, the AI tools can accelerate the creation of new components that follow existing patterns. The Edit with AI feature's ability to apply consistent modifications across multiple elements reduces repetitive manual work.
Developers who use Figma for prototyping but lack formal design training may find the Generate feature useful for creating initial layouts that follow established UI conventions. The system's adherence to common patterns provides a starting point that can be refined.
Design educators have noted that the AI features could serve as learning tools, demonstrating how experienced designers approach common UI challenges. The Auto Layout Suggestions feature in particular exposes the logic behind responsive design decisions.
For Figma as a business, the AI features provide differentiation in an increasingly competitive market. The company's large user base provides a distribution advantage for AI features that competitors may struggle to match.
Cons, Risks, and Limitations
The AI features introduce new considerations for design teams regarding originality and intellectual property. Generated designs, while based on licensed training data, may produce outputs that closely resemble existing published work.
Designers have expressed concern that AI-generated components may encourage homogenization of interface design. If many teams use similar prompts, the resulting designs may converge toward common patterns, reducing visual distinctiveness.
The 85% accuracy rate cited by Figma means that approximately 15% of generated designs require significant correction or regeneration. For time-sensitive projects, the unpredictability of AI output quality may offset efficiency gains.
Accessibility compliance remains a human responsibility. Figma's documentation notes that generated designs are not guaranteed to meet WCAG guidelines and require manual accessibility review. Color contrast, focus states, and screen reader compatibility must be verified by designers.
The AI features require paid Figma subscriptions for full functionality. Free tier users receive limited generation credits, which may create pressure to upgrade for teams that find the features useful.
Some design professionals have questioned whether AI-assisted design tools may reduce demand for junior design positions. Entry-level tasks such as creating initial component layouts have traditionally served as training opportunities for new designers.
How the Technology Works
Figma's AI design generation system combines multiple machine learning approaches to produce and modify visual designs.
The Generate feature uses a large language model to interpret natural language prompts and extract design requirements. The system identifies requested components, their relationships, and any specified styling attributes. This semantic understanding is then passed to a visual generation model.
The visual generation component produces vector-based designs rather than raster images. Figma's system outputs native Figma objects including frames, shapes, text layers, and Auto Layout configurations. This approach allows generated designs to be immediately editable using standard Figma tools.
The Edit with AI feature operates on selected Figma objects, analyzing their current properties and the user's modification request. The system generates a set of property changes that are applied to the selected elements. For complex modifications, the system may restructure the layer hierarchy or add new elements.
Auto Layout Suggestions uses a classification model trained on examples of well-structured Figma files. The model identifies groups of elements that exhibit patterns consistent with responsive layouts and recommends appropriate Auto Layout settings based on the spatial relationships between elements.
Technical context (optional): Figma's AI system operates on the company's vector graphics format rather than pixel-based representations. The generation models output structured data describing Figma objects, which is then rendered by Figma's existing graphics engine. This architecture allows AI-generated designs to maintain full editability and scalability, distinguishing the approach from image-generation AI systems that produce flat raster outputs.
Why This Matters Beyond the Company
Figma's AI release represents one of the largest deployments of generative AI tools to a professional creative workflow. The company's reported 4 million user base means the features will receive extensive real-world testing across diverse design contexts.
The design tool market has consolidated around a small number of platforms, with Figma, Adobe XD, and Sketch serving the majority of professional UI designers. AI capabilities are becoming a competitive differentiator, and Figma's release may accelerate feature development at competing platforms.
The training data transparency that Figma provided in response to the 2023 concerns establishes a precedent for AI tool developers. Other companies introducing AI features to creative tools face similar questions about training data provenance, and Figma's approach of publishing licensed dataset sources may become an industry expectation.
For the broader software development industry, AI-assisted design tools represent another step in the integration of generative AI into professional workflows. The pattern of AI augmentation rather than replacement that Figma describes mirrors approaches in code generation, writing assistance, and other knowledge work domains.
The release also tests user acceptance of AI in creative fields where human judgment and originality are traditionally valued. Designer reactions to Figma's AI features may indicate how other creative professions will respond to similar tools.
What Is Confirmed vs. What Remains Unclear
Confirmed:
- Figma released three AI features: Generate, Edit with AI, and Auto Layout Suggestions on May 7, 2025
- The features are available to paid plan users with limited access for free tier users
- Figma stated that training data did not include user files without consent
- The company published a list of licensed datasets used for training
- Generated designs are native Figma objects, not raster images
Remains unclear:
- The specific AI model architectures and external partnerships powering the features
- Long-term pricing implications as AI features mature
- How the 85% accuracy metric was measured and what constitutes a "correct" generation
- Whether Figma plans to offer opt-in training on user files in the future
- The computational infrastructure costs and how they affect Figma's business model
What to Watch Next
User adoption metrics will indicate whether the AI features provide sufficient value to change design workflows. Figma has not announced plans to publish usage statistics, but third-party surveys of design professionals may provide insight.
Competitor responses from Adobe and other design tool vendors will shape the market. Adobe has integrated AI features into Photoshop and Illustrator, and expansion to Adobe XD would directly compete with Figma's offering.
The design community's assessment of generated design quality will emerge through published reviews, social media discussion, and conference presentations. Config 2025, Figma's annual conference, will likely feature sessions on AI feature usage.
Regulatory developments regarding AI training data may affect Figma and competitors. The European Union's AI Act and proposed U.S. legislation on AI transparency could impose new disclosure requirements.
Enterprise adoption patterns will indicate whether large design teams integrate AI features into production workflows or treat them as experimental tools. Enterprise customers represent a significant portion of Figma's revenue.
Sources
- The Verge - Figma AI Update (May 7, 2025): https://www.theverge.com/2025/5/7/figma-ai-design-tools-announcement
- Figma Official Blog (May 7, 2025): https://www.figma.com/blog/ai-features-may-2025/
- TechCrunch - Figma AI Features (May 7, 2025): https://techcrunch.com/2025/05/07/figma-ai-design-generation/
- Figma Developer Documentation: https://www.figma.com/developers/api