The drag handle Anthropic isn't shipping. And the vision that relies on it never happening.
Figma is the design tool most designers use. Canva caters to everyone else. Claude Design is Anthropic's new entry trying to do it all. Anthropic could offer the granular editing that would make Claude Design competitive with Canva – they don't, and I think they probably won't. The missing drag handle isn't an oversight. It's strategic discipline, and a bet that AI must be the entire product.
Three days before Anthropic launched Claude Design, the company's CPO resigned from Figma's board. Figma's stock fell 7% and Canva celebrated publicly on launch day. After three weeks of using the tool, I keep finding myself back in Figma for design control and in Canva for the drag handle. But I think I see why both are missing.
What's missing
Across three weeks of work in Claude Design, the gap was consistent. Nothing beats Claude Design on generation. The first-pass creative is above what incumbents can produce.
The product is iterating fast. Embedding fonts in image exports required five re-prompts for me two weeks ago, but worked first time when a colleague tested it yesterday. Anthropic is shipping fixes on a tight cadence; this tool is a moving target.
What we aren't able to do, and I'm not expecting to see introduced, is element-level direct manipulation – drag-and-drop. Right now, we can adjust individual elements' parameters like colours, fonts, sizing. An impressive feat considering the technical limitations that have restricted competitors when it has come to AI-generated creative. You cannot, however, drag, move, or remove the elements on the canvas. This always requires re-prompting and accepting whatever output comes back, often times with new issues to match the fixes.
The most basic features that have shipped with every design tool for decades are conspicuously absent. That's the gap I think will hold, it's a structural choice.
Three explanations
There are three reasons a feature might be missing from a shipped product. They couldn't build it. They forgot to build it. Or they chose not to.
"Forgot" feels impossible. The UI itself is self-evident – excluding these elements is more of a task than leaving them in. Their absence is intentional.
The "couldn't" answer needs unpacking, because what can't these companies do? The underlying output is technically trickier. Image generation today defaults to raster (a flat grid of pixels rather than a structured document of editable layers). Producing separable, editable layers from a generative pipeline is closer to current research than to a production-ready system. Acknowledging the limitations outlined above, Design does offer componentised assets – that is, users can modify individual elements rather than the whole image – which is already a true leap forward. So this isn't a case of "couldn't." In reality, they have already delivered the hardest part – these absences show restraint, not inability.
To achieve true element-level control, the truth feels closer to "could, at meaningful and potentially unnecessary additional cost." Canva have already launched similar features and are touting them accordingly, but their offering doesn't wow users like Design does. Figma, on the other hand, with a Claude connector and a comparatively negligible AI moat to defend its position, may have already accepted its fate. The moonlight creators in its base have somewhere shinier to go, while the serious designers who want full manual control stay.
That leaves the third reason: they chose not to. Why?
A drag-and-drop editor would feel like Canva+. Every screenshot of Claude Design with a rotate handle and a traditional layer stack docked on the side would say "this is a manual design tool with a chatbot bolted on". That isn't the company they're building, and they're right in not wanting it to be.
Apple held this kind of line with hardware design for years (no headphone jacks, no SD card slots, no replaceable iPhone batteries). Admittedly, MagSafe is back, and SD cards returned to the Pro models. But most of the time, the market criticises but ultimately respects the discipline, even when the result is frustrating.
So the 'missing' drag handle is a strategic commitment.
Why intent makes strategic sense
Anthropic's brand is coherent and deliberate. They shipped Claude Design with export-to-Canva as a launch-day feature and an endorsement from Canva's CEO, Melanie Perkins, baked into the announcement itself. "...bring ideas and drafts from Claude Design into Canva, where they instantly become fully editable and collaborative designs ready to refine, share, and publish." The public framing: a chain of specialised tools, with the final editing surface explicitly handed off. Read more cynically, the partnership buys Claude Design time to outgrow that editing surface entirely.
With the release of Design, Claude's relationship to these established design tools has shifted from symbiote to predator. With this shift, their CPO's conflict of interest with Figma becomes too clear to ignore. These are not companies that intend to coexist. The market read it the same way, reflected in Figma's 7% stock price drop on launch day. Canva is private, so there's no equivalent signal there, but Perkins' endorsement is exactly what an incumbent says when hoping a collaboration is real and not a stepping stone to its irrelevance.
The handoff runs one direction too: Claude Design exports to Canva, Canva doesn't feed back into Claude Design. Genuine collaborations are two-way streets. Whether the partnership survives as Claude Design matures is an open question. Either reading lands at the same place: Anthropic's strategic case rests on the AI being the entire product.
The cost is real, and the user pays it
The discipline is impressive. The bet is defensible. Neither provides today's user with a usable end-to-end AI workflow.
Figma's stock has already walked back the launch-day panic, dropping as low as $16.86 before recovering to $20.66, slightly above launch-day close. Investors are pricing in what the user already knows: Claude Design hasn't closed the loop.
Claude Design is the shiny new apex predator of content generation. By definition, it shouldn't compete with tools lower in the food chain. Forcing users back to those incumbents for menial work isn't an accident. It's the apex declining to fight in arenas where human labour still trumps AI grunt.
Canva isn't sitting still. At Canva Create 2026, COO Cliff Obrecht framed the company's shift as a pivot from a design platform with AI tools to an AI platform with design tools. Bold framing for what is, in practice, augmentation of the old toolbox rather than replacement. Claude Design still owns the first-pass wow factor.
The question is whether you're willing to be a user through the research preview. I'm stuck in the middle for now, the perfect user who keeps paying for two incomplete products. I'll be dropping one the moment my workflow can tolerate it, and I'm sure there are a lot of people in the same boat, waiting for the true winners to declare themselves.
Claude's current capability falls shy of the end-to-end creative vision. Acknowledging that would undermine the ethos of the entire product. Better to ship a tool that's stunning at one half of the workflow and absent at the other, and let the user fill the silence by reaching for Canva and friends.
This silence is where this strategy will live or die.
I am the user. I had a launch to ship.
For the time being, my creative gets finalised in Canva. But only just.
Counter-readings, comments, and any insider context welcome: hello@greatworkeveryone.com.