Anthropic Overhauled Claude Design Two Months After a Token-Burning Launch Nearly Made It Unusable

Since Anthropic quietly released Claude Design in April as a research preview, the tool has racked up over one million users in its first week and a reputation for burning through subscriptions in minutes. Wednesday's update is the company's attempt to fix both the perception and the underlying mechanics.

The Token Problem Was Real

According to VentureBeat, a PCWorld reviewer burned through 80 percent of his weekly Claude Pro allowance in roughly 25 minutes, producing exactly three variations of a single webpage prototype. The product was functionally unusable for the people paying the most for it.

The April launch generated the kind of first-week traction most product teams dream about. It also generated immediate, documented complaints about token consumption. Anthropic took two months to ship a fix.

The June 17 update, according to both VentureBeat and Engadget, makes Claude Design more token-efficient and adjusts how usage limits are shared across Claude Design, Claude Code, and standard chat. Anthropic says most users will hit their limits less often as a result. The company has not published specific figures on how much the token consumption actually dropped.

What the Update Actually Changes

The headline technical feature is the rebuilt design system import. Users can now pull design systems directly from a GitHub repository, design files, or raw uploads. Once imported, Claude builds with those components and auto-corrects its output against the design system before the user sees the result, according to VentureBeat.

For enterprises, a new admin role can lock down a single approved design system, meaning every asset Claude produces conforms to company brand guidelines without requiring manual review. That represents a meaningful shift from "AI that makes mockups" to "AI that enforces brand compliance at scale."

The Claude Code integration is the other significant change. Claude Design can now ingest a local codebase, generate assets that match existing front-end components, and then hand the design directly to Claude Code for implementation, according to Engadget. Developers who want to skip Claude Design entirely can now type `/design` into their terminal to create and edit designs directly from Claude Code.

Other changes in the update include a more flexible import tool that can build design systems from GitHub and raw files, a refined image editor with finer control over element positioning, and a new shortcut to access Claude Design from the sidebar of Anthropic's desktop app or by navigating to `claude.ai/design` in a browser.

Context: This Is One Piece of a Much Larger Push

VentureBeat's coverage offers a more useful framing. The outlet places Wednesday's Claude Design update inside a ten-week product sprint that also included additional model releases, ten agent templates for financial services, a multi-year alliance with DXC Technology to embed Claude inside bank and airline IT infrastructure, and Claude for Small Business with QuickBooks and PayPal integrations.

Engadget covers the Claude Design update accurately but does not situate it within this broader pattern. The VentureBeat framing is more complete: Anthropic is not building a collection of standalone tools. It is building a platform where Claude is embedded in the systems where work actually happens.

The Fair Concern About Anthropic's Usage Limits

The strongest criticism of Anthropic's subscription model is not just that the tool was slow. Max plan subscribers reasonably believed they were buying meaningful access, and the token consumption rates made that promise hollow. A product that costs more and delivers less access than advertised is a legitimate consumer protection issue, not just a PR problem.

That concern now has legal weight behind it. According to Engadget, The Wall Street Journal reported Monday that a Washington D.C.-based resident has sued Anthropic, accusing the company of misleading consumers about the limits of its Max plans. No charges have been filed in any regulatory proceeding, and the lawsuit's allegations are unproven. The timing of Wednesday's token-efficiency update, though, comes two months after a documented overconsumption problem and days after a reported lawsuit, which is not a coincidence.

Anthropic's stated position is that Wednesday's changes will result in most users hitting their limits less often. The company has not published specific before-and-after token consumption benchmarks, which would be the most direct way to validate that claim.

What Comes Next

The update's structural changes—shared usage limits, the admin approval layer, the sidebar shortcut—read as preparation to move Claude Design out of preview status entirely, as Engadget noted. Whether the token-efficiency improvements are sufficient to resolve the pending lawsuit's allegations will depend on facts the company has not yet made public.