When Anthropic launched Claude Design in April 2026, the product hit one million users in its first week. Then the reviews started coming in, and they were not entirely positive. One PCWorld reviewer burned through 80 percent of his weekly Claude Pro allowance in roughly 25 minutes. For enterprise teams hoping to scale the tool across a department, that was a dealbreaker.

On June 17, Anthropic shipped a major Claude Design update that addresses the token problem and, perhaps more importantly for business teams, adds the kind of brand controls and design system integration that enterprise deployments actually require.

What Was Broken

Claude Design is a visual creation tool built into Claude that generates slides, prototypes, one-pagers, and marketing assets through plain conversation. The core concept is strong: describe what you need, Claude builds it, you refine through dialogue. But the early implementation had a real problem — each design operation was consuming tokens at a rate that made the product impractical for sustained daily use.

For individual users experimenting with the product, this was annoying. For companies considering rolling it out to marketing, product, or creative teams, it was a barrier to adoption.

What Got Fixed

The June 17 update makes two key changes to the token situation.

First, Claude Design now draws from the same usage pool as Claude’s chat, Cowork, and Code products, rather than a separate and smaller allocation. Most users will notice they have significantly more headroom for design work without worrying about a separate quota running dry mid-project.

Second, Anthropic says it has reduced the average token consumption per design turn while maintaining output quality, and that error rates have dropped sharply. The practical result is that the same weekly plan should support meaningful design work, not just a handful of experiments.

The Enterprise Addition That Matters More

The token fix is a relief. The design system feature is what makes Claude Design usable for companies that actually care about brand consistency.

Teams can now import design systems into Claude Design from a GitHub repository, a design file, or a raw upload. Once a design system is loaded, Claude builds within those constraints, checks its output against the system, and self-corrects before showing anything to the user. The result is that generated assets come out with the right fonts, colours, component styles, and spacing — not generic AI-flavoured output that needs to be reworked.

For larger organisations, a new admin role can approve a single standard design system and lock it down across the team. No one else can modify it. Every asset Claude produces for that organisation will conform to the approved guidelines automatically. This is the kind of governance that enterprise IT and brand teams require before they will sign off on AI tooling at scale.

Developer Integration

Anthropic also tightened the connection between Claude Design and Claude Code, which matters for teams where designers and developers work in the same pipeline.

A new /design-sync command lets developers pull a design system directly into Claude Code from the terminal. A companion /design command lets developers create, edit, and sync design projects without leaving the terminal environment. The intent is to close the loop between a design created in Claude Design and the code implementation, reducing the manual handoff that typically costs time and introduces inconsistencies.

Expanded Export Options

Claude Design can now export to PDF and PowerPoint in addition to its existing formats, and it has added integrations with Adobe, Base44, Canva, Gamma, Lovable, Miro, Replit, Vercel, and Wix. Teams that live in Canva for social content or Miro for collaboration can pull Claude-generated work directly into those environments.

The update also includes finer control over individual design elements, proper layout tools for dragging, resizing, and aligning objects, and what Anthropic describes as hundreds of stability fixes — suggesting the April launch was more of a rough preview than a polished product.

What This Means for Business

The original Claude Design launch was intriguing but hard to act on at scale. The token burning problem meant teams had to ration usage. The lack of design system support meant outputs needed heavy reworking before they were on-brand. And without admin controls, IT teams had no way to govern what was being produced.

This update changes the calculus. If your team produces a high volume of internal documents, sales decks, one-pagers, or marketing assets, Claude Design now has the fundamentals that enterprise deployment requires: it works within your brand standards automatically, it doesn’t drain your token budget on a single project, and administrators have the controls they need to keep outputs consistent.

The comparison with traditional design tools is still evolving, but the direction is clear. A tool that generates an on-brand slide deck from a written brief — without a designer in the loop — can compress turnaround times significantly for content that currently lives in a long queue. The question for most teams is no longer whether the technology is capable. It’s whether your people are set up to use it well.

For organisations that haven’t yet put together an AI tools strategy, tools like Claude Design are a useful place to start. The value is concrete, the output is tangible, and the learning curve is low enough that non-technical team members can adopt it without significant training.

---

Enterprise DNA put together a free field guide on exactly this: the full Claude ecosystem, Claude Code, and how to roll agents out without breaking things. Get the guide.