Canva just made the gap between AI-generated images and usable finished designs a whole lot smaller.
Magic Layers, Canva’s tool that turns flat images into fully editable layered designs, is now available directly inside ChatGPT and Google Gemini. Canva says the integration launched June 7, and the X post pushing it live has already racked up 1.4 million views.
Here’s what that actually means.
What Magic Layers does
When you generate an image with AI, you get a flat file. Want to change the tagline? Pull out the background? Swap the font? You’re stuck starting over or jumping into Photoshop.
Magic Layers solves that. It takes any image and separates it into individual editable elements. Text becomes live text. Objects become selectable layers. Backgrounds become their own thing you can swap, remove, or regenerate.
Canva’s own numbers tell the story. Magic Layers was used over nine million times in its first four weeks after launching in March. That’s not a niche feature. That’s people hitting a wall with flat AI images and looking for a way through it. I’ve covered using Canva with AI tools before, and the demand has only grown.
How it works inside ChatGPT
The workflow is dead simple. Generate an image in ChatGPT the way you normally would. Then just prompt something like “turn this into an editable Canva design.” ChatGPT hands it off to Magic Layers, and within seconds you’ve got a fully layered Canva file open and ready to edit.
No download. No re-upload. No switching tabs and losing your flow.
The same flow works in Google Gemini. Generate with whatever model you’re using, prompt @Canva, and get a layered design back.
What you can actually do with it
A few real examples from Canva’s announcement:
Run a local running club and want a poster? Generate a scene in Gemini, prompt Magic Layers, pull the sky into its own layer, drop in your club name and date, change the font to match your brand. Done in five minutes.
Redecorating your living room? Generate three different aesthetics, use Magic Layers to isolate the sofa, the rug, the lamp as separate objects, move them around, swap in the actual furniture you’re eyeing. You just built a mood board and a design brief in one go.
Building brand assets? Generate a product image, unlock it, pull out the background, swap colors to match your brand, resize for every platform. That’s a single sitting instead of a back-and-forth with revisions.
The bigger picture
This isn’t just a ChatGPT thing. Magic Layers is built on Canva’s proprietary foundation model, the Canva Design Model, which powers integrations with Claude and Microsoft Copilot too. The same MCP-based connector infrastructure runs across all of them.
The message is clear: Canva wants to be the design engine for every AI assistant, not just one. If you’re generating images anywhere, they want the finish line to be a Canva file. This follows OpenAI putting an app store inside ChatGPT – the platform play is accelerating on both sides.
Availability
Magic Layers inside ChatGPT and Gemini is available now to all users. No special tier, no waitlist. If you can generate an image in either platform, you can prompt it into Canva.
The old workflow was generate, download, re-upload, edit. The new one is generate, edit, ship.
