Claude Design launched April 17, 2026 as an Anthropic Labs product that lets you collaborate with Claude to create polished visual work like designs, prototypes, slides, one-pagers, and more. It runs on Claude Opus 4.7, currently available in research preview to paying subscribers.

The immediate reaction in design Twitter was predictable: either panic (Figma is dead) or dismissal (it's just a slide maker). Both camps are missing the point.

---

What Claude Design actually is

It is a conversation-first creative tool. You describe what you need, you collaborate back and forth with the model, and you get a finished visual artefact. No canvas. No layers panel. No component libraries. No shift-click.

Opus 4.7 was specifically trained to be "more tasteful and creative when completing professional tasks, producing higher-quality interfaces, slides, and docs." Anthropic also gave it substantially better vision, with higher resolution image understanding. Whether this makes Claude Design's outputs genuinely good -- rather than just competent -- is something you can only answer by using it, and I haven't been able to yet (research preview, paying subscribers only). But the underlying model capability is real.

---

What Figma actually is

Figma is a precision instrument. It gives designers a shared canvas, a component system, a handoff layer, version history, and a collaboration model that teams have built entire workflows around. Its real power is not that it makes design easy -- it is that it makes design collaborative and consistent at scale. Variables, auto layout, design tokens: these things exist because production design is a discipline with constraints, not a creative conversation.

Figma has also been adding AI features. First Draft generates initial designs from prompts. Make Designs turns selected layers into variations. Figma AI can rename layers, generate copy, summarise comments. These are, if we are being honest, mostly tidy utilities. They help you move faster inside Figma's existing model. They do not change what Figma fundamentally is.

---

Where the comparison actually matters

Claude Design and Figma are not competing for the same job.

Figma's job is: take a design system and a brief and produce production-ready, developer-handoff-quality work. That is a craft. It takes training, it takes component thinking, it takes understanding constraints.

Claude Design's job appears to be: take an idea and produce something polished enough to communicate, test, or present. That is a different job entirely, and it is a job that currently has no good solution.

Right now, if a product manager wants to mock up a new feature, they either: bother a designer, fumble through Figma themselves, or produce something embarrassing in Google Slides. If a founder needs a pitch deck, they spend three hours in Canva arguing with templates. If a strategist wants to show a client what a one-pager could look like, they describe it in a document and hope.

Claude Design, if it works well, eats all of those cases. None of them were going to Figma anyway.

---

The real threat to Figma

Here is the uncomfortable truth for Figma: Claude Design does not need to beat Figma at design to hurt Figma's business.

Design tools succeed when stakeholders believe designers need them. The client who sees a polished prototype and thinks "why does it take two weeks to get this?" -- that perception shapes budgets, timelines, and headcount. If Claude Design produces genuinely good-looking output fast, the pressure on the time and cost of traditional design processes increases. That is a problem that sits one layer above the tool.

It also puts pressure on what Figma charges for. Figma Professional is 12 dollars per editor per month. Organisation tier is 45 dollars. The bet Figma is making is that teams need the full collaboration and component infrastructure. Claude Design, potentially sitting inside an existing Claude Pro subscription, does not need to win a feature comparison to feel more accessible.

---

What I would watch for

A few things will determine whether Claude Design is genuinely important or an interesting demo:

Output quality. "Polished" is doing a lot of work in Anthropic's announcement copy. Opus 4.7's design capability needs to produce things that look considered, not just competent. The difference between the two is the whole game.

Iteration. Conversation-based tools succeed or fail on how well they handle the second and third request. "Make the header bigger" or "use a darker blue" are the real tests. The first output is marketing; the revision loop is the product.

Export and handoff. If Claude Design produces outputs you can only view inside Claude, it stays in the "useful for comms" bucket. If it exports clean, developer-friendly code or Figma-importable files, that changes the conversation entirely.

Who actually uses it. If Claude Design ends up mostly in the hands of non-designers producing internal comms and pitch decks, it is a productivity tool with a nice story. If designers start using it for first-pass exploration before moving to Figma, that is structurally more interesting.

---

The bottom line

Claude Design is not a Figma killer. It is something potentially more disruptive: a tool for everyone who needed a designer but did not have one, could not afford one, or did not want to bother one for something that would change three times anyway.

Figma's position in production design workflows is not under threat this week. But the assumption that visual work requires design tools -- and by extension, designers -- is being tested. That is a slower, more interesting disruption than a direct feature fight.

If you have access to the research preview, try it. The question worth answering is not whether it beats Figma. It is whether the output is good enough to change someone's mind about what good enough looks like.

---

Claude Design launched 17 April 2026 as an Anthropic Labs product, currently in research preview for paying Claude subscribers. It runs on Claude Opus 4.7.