Anthropic Unveils Claude Design, Disrupting Figma

Anthropic has launched Claude Design, a new product from its Anthropic Labs division. This tool allows users to generate designs, interactive prototypes, slide decks, and marketing collateral using conversational prompts and fine-grained editing controls. The release, available immediately in research preview to paid Claude subscribers, signifies Anthropic's significant expansion beyond its core language models into application-layer software. This move directly challenges established players like Figma, Adobe, and Canva.

Claude Design is powered by the newly released Claude Opus 4.7, Anthropic's most capable generally available vision model. This launch marks a watershed moment for Anthropic, demonstrating its ambition to evolve from a foundation model provider into a full-stack product company. The product facilitates a natural creative conversation, where users describe their needs, and Claude generates an initial version. Refinements are possible through chat, inline comments, direct text editing, and custom adjustment sliders.

A key feature is Claude's ability to read a team's codebase and design files during onboarding to build a custom design system, which is then automatically applied to new projects. The tool supports various input methods, including text prompts, image and document uploads, and web captures. Its unique handoff mechanism packages designs into a bundle that can be passed to Claude Code for implementation, creating a closed loop from idea to production code within Anthropic's ecosystem. Export options include internal URLs, folders, Canva, PDF, PPTX, and HTML.

Early users like Brilliant and Datadog have reported significant efficiency gains, compressing traditional design cycles into single conversations. However, the launch arrives amid strategic shifts, with Anthropic's chief product officer resigning from Figma's board shortly before the announcement. While Anthropic asserts Claude Design prioritizes interoperability with existing tools, its standalone nature and accessibility to non-designers present a clear competitive threat to market leaders like Figma.

The underlying Claude Opus 4.7 model shows marked improvements in software engineering, instruction following, and vision capabilities. It accepts higher resolution images and performs significantly better on coding benchmarks compared to its predecessor. Notably, Anthropic also announced Claude Mythos Preview, a more powerful model with advanced cybersecurity capabilities, which is being kept behind a vetted-access program due to its potential risks. This dual-track approach to model releases is unprecedented, with Mythos being used to identify vulnerabilities and inform future public releases.

Regarding data handling for enterprise clients, Anthropic states that design-system representations are stored, but source files are not uploaded or retained. Code linked by users is not used for training. Direct connections to GitHub are being added. Claude Design is off by default for Enterprise customers, requiring administrator enablement. Pricing is included in existing Pro, Max, Team, and Enterprise plans, with additional usage subject to existing limits. The API pricing for Opus 4.7 mirrors its predecessor, at $5 per million input tokens and $25 per million output tokens.

Anthropic is transparent about Claude Design's limitations, including basic collaboration features and rough editing experience, with no firm general availability date set. The company intends to let user feedback guide the product's readiness. This launch is part of a broader trend of major AI labs moving into application building, with Anthropic now offering a suite of integrated tools including coding, knowledge assistance, and design. This comprehensive ecosystem aims to support users throughout their entire product development lifecycle.