Workforce Management
How Canva is Getting the Humans Ready for AI
Korn Ferry APAC President Esther Colwill recently spoke with Canva CTO Brendan Humphreys about how they are preparing their employees for the impact of AI.
December 04, 2025
Canva’s Chief Technology Officer Brendan Humphreys was employee number 13 when he joined Australian technology company Canva in 2014. Canva’s journey has been remarkable, powered by 5,000+ employees across the globe who are united in its mission to empower the world to design.
Canva’s visual communication platform is user-focused, accessible to everybody, and at last count has more than 400 designs created every second by a global community of some 260 million monthly active users.
Brendan recently spoke to Korn Ferry’s Asia Pacific President, Esther Colwill, and an audience of tech CEOs and board members, on how Canva is ‘getting the humans ready for AI’.
Esther Colwill: Can we start with a significant milestone that led to the success story that Canva is?
Brendan Humphreys: I think the biggest milestone in Canva’s history was when our co-founder Melanie Perkins had the prescient insight that graphic design was hard and inaccessible. You had to use many different tools to get from an idea through to a published artifact, and each one of those tools has a steep learning curve with its own toolset, and then there is an annoying handoff between the tools.
So, Mel’s observation, that we could condense those six or seven steps with different tools into one unified platform, simplified around very easy-to-use concepts, and importantly, a vast library of content, was the foundation of Canva. Our mission remains to democratise design, to empower all of us, whether in personal or business life, to create stunning visual graphics, which improve our communication.
Esther Colwill and Brendan Humphreys speak about the impact of AI
Colwill: In the last few years, we've seen the shift in workflows, not only around design but incorporating AI. Within your team at Canva, what mindset shifts have the team had to make as AI has started to adjust your workflows?
Humphreys: Canva was doing AI way before it was cool. Our first AI teams started in 2016, and we were bringing AI features to market in our product as early as 2017. Users wouldn't have recognised it as AI, the AI that we all know now as this disruptive technology, but it was still AI-powered technology.
We've been on that journey to bring AI power to our users in our product ever since, and we've built a very powerful platform to do that. Internally, though, we face the same challenges that many face. Our ways of working increasingly look like they need to be adapted to this new AI reality. The tools that are on the market, whether in the visual AI space or large language models or agentic AI, seem to present once-in-a-generation disruption to how we all do work.
We're obviously not immune to that yet we have one advantage in that we have very solid technical expertise around these tools, and we've been tracking the development of the tools very closely. But at the same time, some of the tools are kind of frustrating in that they're very different from the AI tools we were promised.
Colwill: What do you do at Canva to counter the frustrating downside of AI?
Humphreys: We put the tools into the hands of users (employees). We give them a very strong mandate to experiment, to understand right at the coalface of their work how these tools might be applicable. Time and space are hard to give, you've got to pull people out of their routine and give them license to experiment, to fail, and to see what works, but importantly to find that ‘aha’ moment when they realise their jobs have changed. Maybe there is an orchestration that they are manually doing in their business that could now have an agentic solution? We are very much about empowering the workforce, and creating the space for them to experiment, to find that intrinsic motivation rather than a top-down ‘go use AI’ mandate.
That has implications for how you train a workforce, because it's very hard to pick one tool and then build a whole class of education around that tool. The approach we've taken is more permissive because we let people experiment with different tools, and then we have very rapid, iterative cross-pollination of best practice and cautionary tales.
Colwill: I think there's something there that we all identify with, especially the ‘aha’ moment when you've tried something with AI, and, conversely, where you feel really let down when the AI response is deeply wrong. I like the idea of permissive licensing and giving everything to your users rather than the top-down, thou shalt use AI approach.
Humphreys: At Canva, the bottom-up approach gives people space to experiment and to fail and to see what works. We recently ran an “AI Discovery Week”, and the entire company was given license to step away from their normal jobs. We provided in-house training and licensing tools and said, "Go have fun, see what you can do," and then on the last two days there was a hackathon, which was, "Okay, show us what you've learnt." As a tech company, we run hackathons regularly and we use them productively to feed interesting ideas into the product. It's taken on extra importance in the age of AI because it's a very loud and clear signal to the entire organisation.
Colwill: Many organisations are struggling with escalating IT costs, from their core ERP to a multitude of additional applications. The cost benefit of AI is conceptual to some extent. I've certainly heard board members say they’ve still got so much other infrastructure and fixed costs, yet everybody's telling them to pour money into AI as well. So, when and how do you think the cost benefit and value of AI tools play into that concern?
Humphreys: It's really tough. Anyone who's weighing up whether to shore up their data warehouse or invest in AI, should, in my view, always shore up their data warehouse, because the AI is only as good as the data that's exposed to it.
The way that I approach ROI is by looking at task transformation. If I can see a task that used to take hours now taking minutes, with a degree of accuracy, and I see enough of that, then I know that I'm getting some significant material benefit for the investment that I'm making. Humans are essential in guiding the tools, assessing the output and then ultimately owning the output.
Colwill: Can you reflect on any decisions in the AI journey that you might do differently now?
Humphreys: When GPT-4 was announced, there was a huge amount of hype about these tools delivering on an exponential curve of improvement, and if that did follow, we were going to need fewer employees. We took stock and looked at pausing hiring, particularly graduates, because they don't have any domain expertise and therefore were potentially easy to replace with AI. I think on reflection, that was just hype.
I think there is real value to be had from these tools, but I do not think that there is wholesale displacement of human capital now. It could come, but I don't think it's happening yet.
We see that AI in the hands of a skilled engineer turns that engineer into a super engineer so, we need more of those people. It is a story of wanting to accelerate our hiring, not reduce it.
Colwill: Well, as part of a hiring firm, I’m happy to hear that from a human impact perspective, there's a long way to go. And, frankly, as a mother of teenagers, I’m really happy because there are concerns about the interns and grads and how they learn and whether AI is going to make their jobs disappear.
Humphreys: The universities have responded well to AI and we see a lot of AI literacy in the grads entering our business.
The grads are hungry for AI, however the challenge with the grads is that they come into the industry with zero domain knowledge. They don't have any knowledge that's hard fought in a business setting. We're looking at longer mentoring with senior engineers and we're seeing senior engineers are really starting more and more to be predominantly reviewing code, whether that's work generated from AI, or by graduate engineers (who are also using AI). Senior engineers have always had this review responsibility, but in the age of AI coding, this has become even more important.
Colwill: Finally, any tips for leaders?
Humphreys: I think the one tip that I would pass on to leaders is the importance of leadership modelling. Use the tools yourself to really build a good understanding of the possibilities and limitations. I use AI tools every day and now can’t imagine not having them at my disposal, for research, briefing, writing, summarising, and yes, occasionally, even coding. That immersion has helped me build a first-hand understanding of the revolution our employees face.
Transforming Your Workforce for the Future
Are you ready to help your workforce meet the demands of the future? Find out more about how Korn Ferry can help.
Related insights
