When support leaders talk about AI, the conversation often starts with automation.
But the companies seeing the biggest impact from AI aren't necessarily the ones automating the most tickets. They're the ones figuring out how to make AI and human agents work better together.
That's the approach Canva took as its support operation scaled alongside the company's growth.
With 260 million monthly users and a global support team of about 700 specialists across BPO and internal teams, Canva faced a challenge many digital companies encounter: how do you keep delivering high-quality support as your customer base grows rapidly?
The answer wasn't replacing agents with AI. It was strengthening operations first and then using AI to amplify the team already in place.
Start with operational clarity
I’ve seen AI initiatives fail not because it’s a bad fit or wrong use case, but because they're layered on top of messy operations.
Before expanding into AI, Canva invested in improving visibility across its global support organization, which includes internal teams and BPO partners across multiple regions.
Workforce management tools gave leaders a clearer view into demand patterns, staffing coverage and performance across channels. That operational foundation made it possible to identify where AI could deliver meaningful value.
When we spent time observing support workflows, including with Canva's BPO teams in Manila, one insight stood out: Agents weren't just answering questions. A significant portion of their time was spent on the work surrounding each interaction.
That included tasks like:
- Summarizing conversations
- Logging issues for product teams
- Documenting reproduction steps for bugs
- Categorizing and routing cases
These processes are essential for maintaining quality and collaboration, but they were also creating administrative overhead that slowed down resolution times.
Instead of trying to automate the entire customer interaction, Canva focused on reducing that friction.
Introducing AI as a copilot
Canva introduced AI tools designed to assist agents directly within their workflow. The initial focus was on helping agents handle interactions more efficiently rather than replacing them entirely.
In the early stages, agents used AI to summarize past interactions, surface relevant knowledge articles, generate draft responses and automate documentation tasks that previously required manual effort.
The results were solid:
- 15.5% efficiency increase across chat teams in the first half of the year
- 50% reduction in BPO vendor onboarding time (from 2-3 months down to one month)
That last one was particularly important for Canva. Most of their support operation runs through BPO partners in the Philippines. When they bring on a new vendor, it used to take 2-3 months of training before agents could perform at Canva's quality standards. With copilot, they cut that time in half. Across multiple BPO providers this was significant.
Moving to targeted automation
As adoption grew and confidence built, the team began identifying opportunities for targeted automation. Canva introduced AI agents to handle certain high-volume requests where AI could reliably resolve issues end-to-end.
Rather than pursuing full automation everywhere, Canva focused on specific use cases where the technology performed well. When a situation requires human support, interactions are handed off with full context so agents can step in quickly.
Within a couple of months, this approach delivered a 33% increase in automation while maintaining quality metrics.
A new model for scaling support
The Canva team isn’t the only one rethinking customer support. It’s changing in front of our eyes and the most successful teams treat AI as an extension of their workforce, not a replacement for it.
The organizations seeing the strongest results tend to follow a similar pattern:
- Build operational visibility first
- Use AI to augment agents before replacing workflows
- Automate targeted use cases where AI performs well
- Orchestrate AI and human capacity as one unified system
Canva's experience illustrates how that model works in practice. AI didn't remove the need for great support teams. It simply made those teams far more effective.
And as digital products continue to reach global scale, that combination of strong operations, skilled agents and well-designed AI will define the future of customer support.
John Wang is CTO and co-founder of Assembled. Prior to Assembled, he helped build products at top tech companies and has spent the last several years focused on helping support teams leverage AI and modern workforce management to deliver better customer experiences at scale.
