Magician for Figma

Read this first: Magician is gone

Magician for Figma is discontinued. The plugin, its subscription and the company behind it no longer exist as an independent product, and I would not install anything currently trading under this name. This review exists to document what it was and to warn readers off the domain. The strongest evidence is in front of me: at time of writing, magician.design — listed in our catalog as the product's homepage — serves Indonesian online-gambling content, complete with slot promotions and a betting app download. A design tool's domain running gambling spam means the domain changed hands and nobody with a stake in the brand is maintaining it. Whatever a third-party page claims about Magician's current pricing or features in 2026, treat it as stale data scraped from a dead product.

What Magician was

Diagram, a small AI design startup founded in 2021 by Jordan Singer (Dealroom company profile), built Magician as a Figma plugin that put generative AI on the canvas. It had three core capabilities, sold as "magic spells": icon generation from a text description, producing editable Figma vectors rather than raster images; UX copywriting, where you select a text layer and the plugin generates headlines, button labels, error messages or product descriptions; and image generation dropped directly onto the canvas for mood-boarding and mockup filler. Figma itself profiled the plugin as a case study of its text review API, which gives a sense of its standing at the time — it was one of the first AI tools designers actually used daily, and Diagram was the kind of studio Figma paid attention to.

That attention ended the way these stories usually end: Figma acquired Diagram (the deal is dated July 2023 in some sources and February 2024 in others — I could not resolve the discrepancy and have not asserted a date), the team moved in-house to build AI into Figma itself, and the standalone plugins were phased out. The capabilities did not disappear so much as migrate: icon generation, copy suggestions and image generation are now native Figma AI features under Figma's own billing and governance. Review sites that still list Magician with a "$5/month" price are describing a product that has not been sold for a couple of years; a few aggregator pages quote other numbers entirely, which is what happens when nobody is left to correct the record.

My testing notes

I cannot test the current product, because there is no current product, so these notes describe what I observed of Magician while it was live and what I see now. First, during its life the plugin did one thing well: filling mockups. Generating a set of editable vector icons from a prompt saved the usual trip to an icon library, and the copy generator was a decent placeholder engine — good enough for prototype fidelity, not good enough to ship. That was always the honest character of the tool: it accelerated the dull 20% of early-stage design and did nothing for the hard 80%. It was a $5-to-$6-a-month convenience, not a design platform.

Second, observing the aftermath: the domain situation is the detail I would act on today. magician.design resolving to gambling content is not just a curiosity — for anyone who bookmarked the site, followed old links, or has security tooling that allowlists the domain from earlier use, it is an active hygiene issue. I would remove the domain from internal documentation and warn design teams that old references to it lead somewhere they do not want to go. This is also a cautionary tale about aggregator directories in our own industry: a listing with a price, a tagline and a screenshot can outlive the product by years.

To be fair to the original tool, one more thing from its live era. It was unusually good-natured software: prompt-driven, immediate, no configuration, no onboarding. A designer could generate a dozen icon variations in the time it takes to search an icon library, discard eleven, and tweak the twelfth. That immediacy is why people loved it, and why its disappearance stung more than the monthly price ever justified.

What took its place

The natural question is where a designer should go instead, and the answer is unusually clean here: Figma's own AI features occupy the exact slot Magician pioneered. After the acqui-hire, the Diagram team's work fed into Figma's native generation tooling — the same three jobs, now under Figma's governance, billing and model updates rather than a plugin author's. Practically, that means icon and image generation, copy suggestions and layout assistance arrive with your Figma seat or Figma's AI add-on structure instead of a separate card charge, and they are maintained against Figma's release cadence. For teams with procurement requirements — data handling, model training opt-outs, admin controls — buying the capability from the platform owner is also a much easier conversation than buying it from a two-person plugin studio.

The broader lesson from Magician's arc is about plugin economics. It was genuinely one of the first AI tools designers used in anger; Figma's engineering blog ran a case study on how its author used the text review API; aggregators still carry its listing years later. Then the team was acquired, the plugin was wound down, the domain lapsed, and it ended up in the hands of a gambling operator. That sequence — useful tool, acqui-hire, abandonment, domain decay — is common enough that I now check three things before recommending any plugin that charges a subscription: who owns the company, what happens to the plugin if they are acquired, and whether the domain is actually maintained. In Magician's case all three answers are now bad, which is why this review leads with the warning rather than the feature list.

Where it falls short

Everything, in the present tense. There is no support, no roadmap, no billing entity and no functioning homepage. The historical product also had real limits that are worth recording: the generated icons and images were prototyping-grade and required manual refinement before anything production-ready, the plugin was a panel inside someone else's platform rather than a workspace of its own, and output consistency across generations was uneven — a common failure of early diffusion-based tools. For production assets, the sensible reviewers of the era said to use dedicated image tools, and they were right.

There is also an ecosystem risk that has now fully materialised: a plugin's life depends on two companies — its maker and Figma. When both moved on, users simply lost the tool one day to the next. Anyone choosing design plugins today should ask what happens to their workflow if the maker is acqui-hired.

Verdict

Magician for Figma is rated 1.2 on the ToolSpotter rubric, and the score reflects a discontinued product whose domain now serves gambling spam: the historical functionality was a genuinely useful prototyping aid and the price was trivial, but there is nothing to buy, no support, and a reliability picture that could not be worse. The capabilities live on inside Figma's native AI features. If you arrived here looking for AI icon and copy generation in Figma, use the built-in Figma AI tooling — and if your team's documentation still links to magician.design, clean that up today.

Sources

• https://magician.design
• https://00011000.com/en/tools/magician
• https://allbestapps.net/ai-app/magician-for-figma
• https://www.figma.com/blog/how-magician-uses-figmas-text-review-api
• https://app.dealroom.co/companies/diagram_
• https://andrewcmcguire.com/ai/acquisition/figma-diagram
• https://thesmartfares.online/blog/diagram-ai-review
• https://digitalbydefault.ai/blog/magician-figma-ai-review-2026
• https://rightaichoice.com/tools/magician
• https://toolmango.com/tools/magician
• https://lazyrobot.ai/blog/magician-for-figma-vs-canva-ai-2026