Canva's AI assistant can now call various tools to make designs for you

At a glance:

• Canva AI 2.0 lets users describe design tasks in plain language and have the assistant call the right tools automatically.
• The assistant builds editable designs using layered elements, giving users full control to tweak the final product.
• New integrations with Slack, Gmail, Google Drive, Calendar, and Zoom let the AI gather context from emails, files, and meetings.

A design assistant that plans and executes

Canva's latest AI upgrade shifts the platform from a static design tool to an agent-like assistant that can interpret a user's intent, select the appropriate creative tools, and assemble an editable design in one go. Instead of manually choosing templates, dragging in assets, and adjusting layers, users can now type a description—"a minimalist Instagram post for a coffee shop opening"—and the AI will call the relevant tools, generate options, and return layered designs ready for fine-tuning. This approach mirrors the broader industry push toward AI agents that handle multi-step workflows, but Canva's focus remains squarely on the creative and publishing pipeline.

Layers, context, and integrations

The new system builds designs using layers, preserving the flexibility that professional designers expect. Users can modify text, images, colors, and layouts without starting from scratch. To make the assistant more context-aware, Canva is rolling out integrations with Slack, Gmail, Google Drive, Calendar, and Zoom. With permission, the AI can read emails, conversations, files, and meeting notes to better understand brand guidelines, recent projects, or upcoming events. A web research skill is also being added so the bot can browse the internet for up-to-date information when generating content.

Scheduling and code improvements

Beyond one-off design tasks, Canva AI 2.0 introduces scheduling. Users can set repeatable tasks—such as generating a weekly social media graphic—to run in the background. The assistant will produce a draft for review before publishing, keeping humans in the loop for final approval. On the technical side, the AI code generator now imports HTML, and users can describe the kind of spreadsheet they want to create using natural language prompts, broadening the assistant's utility beyond pure design work.

Faster, cheaper models

Canva says it has significantly improved the efficiency of its AI models. The Lucid Origin image-generation model is now five times faster and thirty times cheaper to run, while the 12V image-to-video model is seven times faster and seventeen times cheaper. These performance gains are intended to make high-quality creative output more accessible, especially for small businesses and teams that rely on Canva for daily workflows.

Enterprise momentum and the road to IPO

While a large portion of Canva's revenue still comes from individual users and small teams, its enterprise business is growing rapidly—up 100% year over year, according to co-founder and COO Cliff Obrecht. He noted that many small businesses start and end their day in Canva, using it for a wide range of workflows. Obrecht also emphasized that Canva integrates well with other AI platforms like Anthropic, Google, and OpenAI, allowing users to pull content from Canva into their agentic workflows before returning to Canva for the final mile of editing, collaboration, and deployment. The company, most recently valued at $42 billion per PitchBook data, is targeting a public listing next year.

The competitive landscape

Canva's move comes as competitors double down on AI-driven assistants. Adobe this week launched a Firefly AI assistant capable of using its suite of apps to complete tasks, while Figma last month added support for AI agents through an MCP server. Despite these advances, Obrecht believes Canva's strength lies in its end-to-end creative environment, where users can both generate and publish content without leaving the platform. As AI agents become more capable, the battle may shift from who has the best model to who owns the final step of the creative workflow.