TLDR
- Canva is shifting generative AI from one-off image creation to an integrated “creative operating system” that streamlines designing, editing and publishing in one place.
- Its design model generates fully editable, layered outputs, reducing repeated prompting and blending AI suggestions with direct human control.
- Canva is expanding into business use cases, focusing on brand consistency at scale while staying model-agnostic and investing in faster, lower-latency AI.
Online design giant Canva is betting that the next phase of generative AI in design will be less about producing static images and more about making it faster and more intuitive for people to create, edit and publish content in one place.
“We’re really focused on continuing to build out Canva as the platform and the creative operating system in the AI era,” said Danny Wu, the company’s head of AI products, in an interview with The AI Innovator.
“What I mean by that is really bringing together all the different steps, all the different tasks, all the different things you might do while you’re creating a design and actually leveraging AI to not just streamline the process, but to actually enable all of our users to just better achieve whatever design tasks they want to create.”
At the end of 2025, Canva said it served 260 million people a month globally who use its free and subscription-based platform to create and design content. The Australian company said 95% of the Fortune 500 use its services. Wu said Canva was created to make design easier for users, combining on one platform a scattered set of design tools that require different skills.
Wu has been at Canva for about a decade and oversees AI initiatives across teams. He said AI is central to Canva’s core mission of making design accessible to as many people as possible, including users who never pay for the product.
“We intentionally want to (offer), and have had, just as good of a free product as we possibly can, and actually pack as much value as we can,” Wu said. This is to ensure “Canva is not something that’s just accessible to people who … have a lot of money and can afford to pay.”
AI everywhere in design
In the generative AI era, the vision is to weave AI into the entire platform. Recently, it introduced several AI initiatives. These include its Creative Operating System that serves as the design foundation and a free studio-grade creative app called Affinity that combines photo, vector and layout tools in one place.
Another AI initiative is the Canva design model, which lets users edit an AI-generated image without having to re-prompt the model to fix the text, objects or photo effects. That’s because the design model creates results that are “layered and have things broken apart,” Wu said.
“This last mile and the combination of AI and direct editing is something unique and unparalleled that we offer,” Wu said. “You don’t have to keep on re-prompting AI. … It’s a system where the AI parts as well as the human control and the human creativity actually mesh together and work in harmony instead of being on two separate tracks.”
Wu said the approach is possible because Canva trained its models on millions of professionally created templates that are already separated into editable components. “That allowed us to train the design models to also generate designs, not images,” he said.
The design model can also offer feedback and suggestions during the creative process. “It can give you design advice as well,” Wu said.
Coding for non-technical people
During the interview, Wu demonstrated how a single prompt could generate multiple poster designs for a fictional vegan food festival in Sydney, with each result immediately editable inside Canva’s editor. He contrasted that with chatbot-based image generation workflows that require repeated prompting.
“You keep prompting and prompting and it comes out worse,” he said.
Wu also demonstrated Canva Code, another AI feature that enables users with no programming experience to generate simple applications and websites using natural language prompts.
“You don’t have to be a programmer, you don’t have to be a software engineer,” Wu said. “You can actually just describe what application, what website, what tool that you want to create.”
In Wu’s demonstration, Canva Code generated a multi-page website within minutes for the same fictional food festival, complete with a vendor application form and color preferences. (Canva can host the website.) He said this feature has seen unexpected uptake among educators.
“We’re finding that a lot of teachers are actually finding this really helpful for creating tailored and interactive and engaging educational materials, like quizzes, like games,” he said. “That’s been surprising to us, but in a really, really good way.”
Wu said Canva is focused on improving performance as it adds more AI-powered features, especially reducing latency. “Something else that we are absolutely focused on is really making our AI as fast as we can without trading off quality,” he said.
Next up: the last creative mile
Beyond individual creators, Wu said Canva is increasingly focused on business use cases, particularly brand consistency.
“If you’re a business, you have a brand style, you have guidelines. The generic AI outputs of just any presentation aren’t really suitable for you,” he said. One of Canva’s next goals is to make it easier for companies to create designs that fit their brand style at scale.
There’s a “huge gap today between AI that’s useful for individuals and SMB use cases versus generative AI that is actually far more immediately useful for businesses,” such as a real estate agency that’s using AI to scale its business and communication, Wu said. “You want to use your color schemes, your logos, your brand guidelines.”
On the technical side, Wu said the company is investing heavily in its own AI research while remaining model-agnostic, combining internally developed systems with models from partners including OpenAI, Anthropic, Google and Black Forest Labs.
“We have a strategy of being model agnostic,” he said. “We evaluate and we look for the best models in terms of quality, in terms of speed, latency and control.”
What’s next in design
Looking ahead, Wu said Canva is working toward making AI-driven design feel closer to real-time collaboration.
“We do see a lot of pain points with AI today, especially around the latency and around how long things can take and how it’s maybe a bit of a mystery box. You don’t necessarily know what you get until you’ve waited a bit,” Wu said.
Meanwhile, Canva will keep working on making its platform better. “One of the things that we’re really excited about is just how much we can push the speed of AI so that we can push the productivity of all of our communities,” he said.
When asked about Canva’s long-rumored IPO, Wu declined to comment but said the focus remains on advancing Canva’s mission of making design easy and accessible to all.
The company has a two-step plan: First, build one of the world’s most valuable companies. Second, do the most good that it can. “From the beginning, we’ve believed that success and impact go hand in hand. The more we grow, the more good we can do – and that’s what drives everything we build,” according to Canva’s website. The company also said its two founders have pledged much of their wealth, equal to about “30% of Canva’s value,” to doing good.
