Anthropic as we speak launched Claude Design, a brand new product from its Anthropic Labs division that enables customers to create polished visible work — designs, interactive prototypes, slide decks, one-pagers, and advertising and marketing collateral — by means of conversational prompts and fine-grained modifying controls. The discharge, accessible instantly in analysis preview to all paid Claude subscribers, is the corporate's most aggressive enlargement past its core language mannequin enterprise and into the appliance layer that has traditionally belonged to corporations like Figma, Adobe, and Canva.
Claude Design is powered by Claude Opus 4.7, Anthropic's most succesful usually accessible imaginative and prescient mannequin, which the corporate additionally launched as we speak. Anthropic says it’s rolling entry out progressively all through the day to Claude Professional, Max, Staff, and Enterprise subscribers.
The simultaneous launches mark a watershed for Anthropic, whose ambitions now visibly prolong from basis mannequin supplier to full-stack product firm — one that desires to personal the arc from a tough thought to a shipped product. The timing can also be important: Anthropic hit roughly $20 billion in annualized income in early March 2026, in accordance with Bloomberg, up from $9 billion on the finish of 2025 — and surpassed $30 billion by early April 2026. The corporate is in early talks with Goldman Sachs, JPMorgan, and Morgan Stanley a couple of potential IPO that would come as early as October 2026.
How Claude Design turns a textual content immediate right into a working prototype
The product follows a workflow that Anthropic has designed to really feel like a pure artistic dialog. Customers describe what they want, and Claude generates a primary model. From there, refinement occurs by means of a mixture of channels: chat-based dialog, inline feedback on particular parts, direct textual content modifying, and customized adjustment sliders that Claude itself generates to let customers tweak spacing, coloration, and format in actual time.
Throughout onboarding, Claude reads a crew's codebase and design recordsdata and builds a design system — colours, typography, and elements — that it robotically applies to each subsequent undertaking. Groups can refine the system over time and preserve a couple of. The import floor is broad: customers can begin from a textual content immediate, add photographs and paperwork in numerous codecs, or level Claude at their codebase. An online seize software grabs parts straight from a dwell web site so prototypes appear to be the actual product.
What distinguishes Claude Design from the wave of AI design experiments which have proliferated up to now yr is the handoff mechanism. When a design is able to construct, Claude packages all the things right into a handoff bundle that may be handed to Claude Code with a single instruction. That creates a closed loop — exploration to prototype to manufacturing code — all inside Anthropic's ecosystem. The export choices acknowledge that not everybody's subsequent step is Claude Code: customers also can share designs as an inner URL inside their group, save as a folder, or export to Canva, PDF, PPTX, or standalone HTML recordsdata.
Anthropic factors to Good, the schooling know-how firm identified for intricate interactive classes, as an early proof level. The corporate's senior product designer reported that essentially the most advanced pages required 20 or extra prompts to recreate in competing instruments however wanted solely 2 in Claude Design. The Good crew then turned static mockups into interactive prototypes they may share and user-test with out code assessment, and handed all the things — together with the design intent — to Claude Code for implementation. Datadog's product crew described the same shift, compressing what had been a week-long cycle of briefs, mockups, and assessment rounds right into a single dialog.
Why Anthropic's chief product officer simply resigned from Figma's board
The launch arrives towards a backdrop that makes Anthropic's declare of complementarity with current design instruments tough to take solely at face worth. Mike Krieger, Anthropic's chief product officer, resigned from the board of Figma on April 14 — the identical day The Data reported Anthropic's subsequent mannequin would come with design instruments that would compete with Figma's main providing.
Figma has collaborated intently with Anthropic to combine the frontier lab's AI fashions into its merchandise. Simply two months in the past, in February, Figma launched "Code to Canvas," a function that converts code generated in AI instruments like Claude Code into absolutely editable designs inside Figma — making a bridge between AI coding instruments and Figma's design course of. The partnership felt like a mutual guess that AI would make design extra important, not much less. Claude Design complicates that narrative considerably.
Anthropic's place, based mostly on VentureBeat's background conversations with the corporate, is that Claude Design is constructed round interoperability and is supposed to satisfy groups the place they already work, not substitute incumbent instruments. The corporate factors to the Canva export, PPTX and PDF help, and plans to make it simpler for different instruments to attach by way of MCPs (mannequin context protocols) as proof of that philosophy. Anthropic can also be making it attainable for different instruments to construct integrations with Claude Design, a transfer clearly designed to preempt accusations of walled-garden ambitions.
However the market learn the indicators in a different way. The structural rigidity is evident: Figma instructions an estimated 80 to 90% market share in UI and UX design, in accordance with The Subsequent Net. Each Figma and Adobe assume a educated designer is within the loop. Anthropic's software doesn’t. Claude Design isn’t merely one other AI copilot embedded in an current design utility. It’s a standalone product that generates full, interactive prototypes from pure language — accessible to founders, product managers, and entrepreneurs who’ve by no means opened Figma. The enlargement of the design person base to non-designers is the actual aggressive risk, even when the skilled designer's workflow stays anchored in Figma for now.
Inside Claude Opus 4.7, the mannequin Anthropic intentionally made much less harmful
The mannequin powering Claude Design is itself a big story. Claude Opus 4.7 is Anthropic's most succesful usually accessible mannequin, with notable enhancements over its predecessor Opus 4.6 in software program engineering, instruction following, and imaginative and prescient — however it’s deliberately much less succesful than Anthropic's strongest providing, Claude Mythos Preview, the mannequin the corporate introduced earlier this month as too harmful for broad launch as a consequence of its cybersecurity capabilities.
That dual-track strategy — one mannequin for the general public, one mannequin locked behind a vetted-access program — is unprecedented within the AI {industry}. Anthropic used Claude Mythos Preview to determine 1000’s of zero-day vulnerabilities in each main working system and net browser, as reported by a number of shops. The Challenge Glasswing initiative that homes Mythos brings collectively Amazon Net Providers, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Basis, Microsoft, Nvidia, and Palo Alto Networks as launch companions.
Opus 4.7 sits a deliberate step under Mythos. Anthropic acknowledged in its launch that it "experimented with efforts to differentially cut back" the brand new mannequin's cyber capabilities throughout coaching and ships it with safeguards that robotically detect and block requests indicating prohibited or high-risk cybersecurity makes use of. What Anthropic learns from these real-world safeguards will inform the eventual purpose of broader launch for Mythos-class fashions. For safety professionals with respectable wants, the corporate has created a brand new Cyber Verification Program.
On benchmarks, the mannequin posts robust numbers. Opus 4.7 reached 64.3% on SWE-bench Professional, and on Anthropic's inner 93-task coding benchmark, it delivered a 13% decision enchancment over Opus 4.6, together with fixing 4 duties that neither Opus 4.6 nor Sonnet 4.6 may crack.
The imaginative and prescient enhancements are substantial and straight related to Claude Design: Opus 4.7 can settle for photographs as much as 2,576 pixels on the lengthy edge — roughly 3.75 megapixels, greater than 3 times the decision of prior Claude fashions. Early entry associate XBOW, the autonomous penetration testing firm, reported that the brand new mannequin scored 98.5% on their visual-acuity benchmark versus 54.5% for Opus 4.6.
In the meantime, Bloomberg reported that the White Home is making ready to make a model of Mythos accessible to main federal businesses, with the Workplace of Administration and Finances establishing protections for Cupboard departments — an indication that the federal government views the mannequin's capabilities as too essential to depart solely in personal fingers.
What enterprise consumers have to learn about information privateness and pricing
For enterprise and regulated-industry consumers, the info dealing with structure of Claude Design will likely be a crucial analysis criterion. Primarily based on VentureBeat's unique background discussions with Anthropic, the system shops the design-system illustration it generates — not the supply recordsdata themselves. When customers hyperlink a neighborhood copy of their code, it’s not uploaded to or saved on Anthropic's servers. The corporate can also be including the power to attach on to GitHub. Anthropic states unequivocally that it doesn’t prepare on this information. For Enterprise clients, Claude Design is off by default — directors select whether or not to allow it and management who has entry.
On pricing, Claude Design is included at no further price with Professional, Max, Staff, and Enterprise plans, utilizing current subscription limits with optionally available additional utilization past these caps. Opus 4.7 holds the identical API pricing as its predecessor: $5 per million enter tokens and $25 per million output tokens. The pricing technique mirrors the strategy Anthropic took with Claude Code, which launched as a bundled function and quickly grew into a significant income driver. Anthropic's reasoning is easy: the easiest way to study what folks will construct with a brand new product class is to place it of their fingers, then construct monetization round demonstrated worth.
Anthropic can also be being clear concerning the product's limitations. The design system import works finest with a clear codebase; messy supply code produces messy output. Collaboration is primary and never but absolutely multiplayer. The modifying expertise has tough edges. There isn’t a normal availability date, and Anthropic says that’s intentional — it would let the product and person suggestions decide when Claude Design is prepared for prime time.
Anthropic's guess that proudly owning the complete artistic stack is definitely worth the threat
Claude Design is essentially the most seen expression of a pattern that has been accelerating for months: the key AI labs are shifting up the stack from mannequin suppliers into full utility builders, straight getting into classes beforehand owned by established software program corporations. Anthropic now affords a coding agent (Claude Code), a knowledge-work assistant (Claude Cowork), desktop pc management, workplace integrations for Phrase, Excel, and PowerPoint, a browser agent in Chrome, and now a design software. Every product reinforces the others. A designer can discover ideas in Claude Design, export a prototype, hand it to Claude Code for implementation, and have Claude Cowork handle the assessment cycle — all inside Anthropic's platform.
The monetary momentum behind this enlargement is staggering. Anthropic has obtained investor affords valuing the corporate at roughly $800 billion, in accordance with Reuters, greater than doubling its $380 billion valuation from a funding spherical closed simply two months in the past. However constructing an utility empire whereas concurrently navigating an AI security popularity, an impending IPO, rising public hostility towards the know-how, and the diplomatic fallout of competing with your individual companions is a balancing act that no know-how firm has tried at this scale or pace.
When Figma launched Code to Canvas in February, the implicit promise was that AI coding instruments and design instruments would develop collectively, every making the opposite extra invaluable. Two months later, Anthropic's chief product officer has left Figma's board, and the corporate has shipped a product that lets anybody who can kind a sentence create the type of interactive prototype that when required years of design coaching and a Figma license. The partnership might survive. However the energy dynamic simply modified — and within the AI {industry}, that tends to be the one type of change that issues.
