Canva adds Perplexity Computer connector for editable assets

Canva plugs into the emerging desktop-agent workflow

Canva has launched a connector for Perplexity Computer, extending the design platform into a fast-growing class of AI tools built around desktop agents rather than standalone chat windows. According to the supplied item, Perplexity subscribers can now add the Canva connector and use the Computer platform to turn prompts and data into editable Canva assets.

That combination matters because it shifts AI-assisted design work from simple text generation toward an action-based workflow. Instead of only answering questions or summarizing documents, an agent can move from instruction to output inside a live creative toolchain.

The key phrase in the item is editable assets. That suggests the output is not merely a flattened image or a throwaway mockup, but something users can continue to modify inside Canva. In practical terms, that could make the connector more useful for marketing teams, social-media managers, educators, and small businesses that already rely on Canva as a lightweight production environment.

Why agent connectors matter

AI platforms are increasingly competing on access to tools rather than model performance alone. A connector is valuable because it lets an AI system act on software a user already depends on, whether that means creating, organizing, or transforming material in-place.

In this case, the connector joins two patterns that have been developing in parallel. Canva has been layering AI into design creation and editing. Perplexity, meanwhile, has been pushing further into agentic workflows in which a desktop AI can take structured actions based on instructions and available data.

The result is a workflow where design generation becomes part of a broader task pipeline. A user might assemble information, summarize a brief, convert it into a presentation draft or social asset, and then keep editing from there. The supplied item does not detail every supported format or limitation, but the core proposition is clear: use data and prompts in Perplexity’s desktop agent to produce Canva-native work.

What this says about the next phase of AI software

The bigger story is not only that one connector launched. It is that creative software is increasingly being treated as an endpoint for autonomous or semi-autonomous AI actions. That changes how software vendors think about interfaces. The important question becomes less about whether a model can generate text, and more about whether it can complete work inside the tools where real output lives.

For Canva, that is strategically sensible. The company has a large installed base among users who want speed, templates, and collaboration without the complexity of traditional design suites. Making those assets accessible through an external agent can help keep Canva central even as AI assistants become a first stop for task initiation.

For Perplexity, the value is similar. Connectors make its desktop agent more practical. They let the platform claim usefulness beyond search and answer generation by showing it can produce a tangible deliverable in another application.

The likely use cases

Even with limited verified detail in the supplied item, the immediate use cases are easy to infer from the described capability. Teams that work from structured data could generate charts, presentations, or branded graphics more quickly. Individuals could turn rough prompts into editable campaign assets. And because the assets remain inside Canva, they can be reviewed and polished rather than accepted as final AI output.

• The connector is available to Perplexity subscribers, according to the supplied item.
• It links Perplexity Computer with Canva.
• The stated outcome is turning data and prompts into editable Canva assets.

That last point may be the most important. In AI software, editability is often the difference between a novelty and a workflow. If users can keep control after the agent does the first pass, connectors like this are more likely to become standard infrastructure rather than one-off demos.

This article is based on reporting by 9to5Mac. Read the original article.