A conversational design tool powered by Claude Opus 4.7 brings prototyping, brand system automation, and direct handoff to Claude Code, positioning Anthropic against Canva and Figma in the enterprise visual work market.
Anthropic announced Claude Design today, its first product from a newly-formed internal division called Anthropic Labs. The tool lets users create designs, prototypes, slides, and one-pagers through conversation with Claude, and ships immediately to Pro, Max, Team, and Enterprise subscribers as a research preview.
The product is built on Claude Opus 4.7, Anthropic's most capable vision-enabled model, and treats visual creation as a dialogue rather than a canvas exercise. Users describe what they want in text, upload reference materials including DOCX, PPTX, XLSX files, existing codebases, or captured website elements, and Claude generates and iteratively refines the output through follow-up prompts and granular controls.
The onboarding flow is the quiet innovation. When a team sets up Claude Design, the tool analyzes their codebase and existing design files to extract a design system: typography, color palettes, component patterns, and brand assets. Every subsequent project automatically inherits those elements. Aneesh Kethini, a product manager at Datadog quoted in Anthropic's announcement, described the shift: "What used to take a week now happens in a single conversation."
Claude Design outputs to multiple destinations. Finished work exports to Canva, PDF, PPTX, standalone HTML, or internal shareable URLs scoped to the organization with private, view-only, or edit permissions. There is also direct handoff to Claude Code, meaning a static design can be passed to Anthropic's coding agent for conversion into working frontend code.
The Canva integration is notable for what it signals. Melanie Perkins, CEO of Canva, provided a quote in the announcement expressing excitement about deepening the collaboration. Canva has been building its own AI tools aggressively, but choosing to integrate with Claude Design rather than compete directly suggests Anthropic is positioning this as infrastructure for design work rather than a Canva replacement.
The Brilliant case study included in the launch materials frames the productivity claim more concretely. The education company reported that pages requiring 20 or more prompts in other AI design tools needed only 2 prompts in Claude Design. Whether that efficiency generalizes beyond curated case studies is something the market will decide over the next few months.
Enterprise organizations must enable Claude Design through admin settings before users can access it, and usage draws from existing subscription limits with an "extra usage" option for heavier workloads. The pricing model suggests Anthropic is not treating this as a standalone premium product but as an extension of the Claude subscription, which has implications for how quickly adoption can scale.
The positioning question
The launch puts Anthropic in competition with established design AI tools including Figma's AI features, Canva's AI Magic Studio, and Adobe's Firefly-integrated products. Each competitor has design-native heritage that Anthropic lacks. What Anthropic brings instead is the coding workflow integration. The ability to generate a design, then pass it directly to Claude Code for implementation, closes a handoff gap that has been a persistent friction point in design-to-dev workflows.
Anthropic Labs itself is a new organizational signal. Until today, Anthropic's product surface area has been tightly coupled to the core Claude model and Claude Code. The Labs division suggests the company intends to expand into adjacent productivity categories where a capable foundation model provides natural leverage.
The rollout is gradual over the course of launch day, which is standard for research previews. Initial user reports will emerge over the coming week, at which point the accuracy of the Brilliant-style productivity claims becomes testable. For now, Anthropic has staked out a position: design work, like coding, is a collaboration between humans and models, and the tool should be the conversation.