Australia's Saas-Y graphic design service Canva is requesting candidates for developer jobs to use AI coding assistants during the interview process.
Discovered in 2012, Canva started out as a web-based tool for DIY designers and has since grown rapidly into an enterprise Adobe's more serious competitor. The company, which is still private, currently serves more than 220 million users and has recurring revenues of $3 billion a year in 2024.
Canva's hiring process previously included interviews focusing on the fundamentals of computer science. Meanwhile, candidates had to code using only the actual human brain.
The company is currently looking for candidates for FrontEnd, BackEnd, and Machine Learning Engineering roles to demonstrate their skills with tools such as Copilot, Cursor and Claude during technology interviews.
AI tools are essential to stay productive and competitive in modern software development
His rationale for this change is that almost half of Canva's front-end and back-end engineers use AI coding assistants every day, and is now a predictable behavior, and the tool is “essential to stay productive and competitive in modern software development.”
However, Canva's old interview process “suggested candidates to solve coding problems without the exact tools they use at work,” admitted Newton. “This rejection of AI tools during the interview process meant that candidates were not truly valuing how they function in their actual roles,” he added.
Candidates had already started using AI assistants during interview tasks – and sometimes they used Subterfuge to hide it.
“We decided to fight this reality and not try to police the use of AI, but to embrace transparency and cooperate with this new reality,” Newton wrote. “This approach provides a clearer signal about how they actually perform when they join our team.”
Canva's existing engineers did not like the change.
“The initial response was concerns that it would simply replace the foundations of strict computer science with what one engineer called a “vibe coding session,” Newton writes.
Canva pointed out that he will tackle these concerns and continue testing basic computer science skills in a variety of ways.
The company piloted the recruitment process where candidates are seen who are expected to use their preferred AI tools, solving what Newton described as “the kind of challenge that even AI support requires authentic engineering judgment.”
“These problems cannot be solved with a single prompt. They require repetitive thinking, clarification of requirements, and good decision-making.”
The pilot tested candidates for the following skills:
- Understand when to effectively utilize AI
- Ability to break down complex and ambiguous requirements
- Make healthy technical decisions while using AI as a productivity multiplier
- Identifying and fixing AI-generated code issues
- Ensure that AI-generated solutions meet production standards
Newton said successful candidates “not only embrace what they have generated by encouraging AI,” but instead demonstrated their ability to selectively use coding assistants to improve their output.
“Interestingly, candidates with minimal AI experiences often struggled because they lacked the decision to identify subtle suggestions as subtle, not because they couldn't code, but because they lacked a ruling that guided AI effectively,” Newton said.
“We believe the future belongs to engineers who can seamlessly blend human creativity and judgment with AI capabilities,” Newton added.
He rated early testing of Canva's new recruitment process as “promising.”
“Our AI-assisted interviews find it more attractive to both candidates and interviewers, providing a strong predictive signal about candidates' performance,” he writes. “Most importantly, we help identify engineers who can use AI thoughtfully and effectively. They're like people who want to build a future in visual communication.”
This suggests that the experience of AI coding assistants has become an important developer skill, at least if you need gigs in Canva. ®
