CEOInterviews.AI
Start App
Melanie Perkins
CEO & Co-Founder, Canva

Melanie Perkins Canva's CEO on its big pivot to AI enterprise software | Decoder

📅 Apr 20, 2026 Decoder with Nilay Patel and The Verge 60 MIN 2808 VIEWS 139 SEGMENTS · 2 SPEAKERS
The last time Canva CEO Melanie Perkins was on Decoder, the company was starting a big push into enterprise. Now, she's leading it through a total reinvention, going, in Canva's words, "from a design platform with AI tools to an AI platform with design tools." But there's a lot of competition in that AI enterprise space. Not only is Canva competing with design software like the Adobe Creative Suite, but also it's competing with AI companies, like Anthropic and Meta, that are launching their own AI design platforms. So we talked a lot about whether Canva really is the right platform to bring t...

Questions asked in this interview

12
  1. 0:00Who do you think your biggest competitor is?
  2. 5:28Did you automate the creation of your own deck announcing the AI tools or did you make it all by hand?
  3. 12:53So is it actually clicking around Canva or is there some other way of the AI addressing the tools?
  4. 18:59How do you think about who owns the interface in a world where the core tool set might be usable somewhere else entirely that also has access to all that data and all that information that the company might have generated?
  5. 24:14The last time I asked you how you make decisions, you said you had a process called decision decks where you literally made Canva documents with all the pros and cons and you mocked up the products. Is that still the process?
  6. 26:57... really curious about, you know, just in that context, decisions and structure, how you made the decision to say, 'Okay, we're going to do Canva 2.0 and we're going to lean heavily into AI the way that we're going to lean into AI,' right?
  7. 32:17So can you make Canva with Canva Code?
  8. 37:06And I think a bunch of designers are going to be very unhappy about this, right?
  9. 45:58I just, the reality is Mark Zuckerberg exists and he's very, very, very competitive, right?
  10. 48:06I kind of thought that's what you would say, but it's just I see it coming, right?
  11. 50:04When you say your own models, are you actually, are you like in the fight for GPUs or are you training them on someone else's cloud? How's that working?
  12. 55:42Do you see that as your, I will say your competitor, retrenching?
Neil Patel 0:00 ↗
Who do you think your biggest competitor is?
Melanie Perkins 0:02 ↗
I shouldn't have opened that question up, should I? I should have let you just go.
Neil Patel 0:05 ↗
You walked right into this.
Melanie Perkins 0:06 ↗
I know. I know. There's a lot going on. And it told me I had a massage booked. And I was really surprised about that because I didn't actually know until I read that in my Canva doc. And then I was like, 'Oh, I think there's like a bug here.' And then I realized that my partner...
Neil Patel 0:19 ↗
The bug is it's booking self-care for you whenever it wants.
So, can you make Canva with Canva code?
Melanie Perkins 0:25 ↗
Yeah, we do make Canva with Canva code. Not deployed. We have many incredible engineers that actually make it sound to go out to hundreds of millions of people.
Neil Patel 0:39 ↗
Hello and welcome to Decoder. I'm Nilay Patel, editor-in-chief of The Verge, and Decoder is my show about big ideas and other problems. Today I'm talking with Melanie Perkins, founder and CEO of Canva, the popular design tool. I always enjoy talking with Melanie. She was last on the show a couple years ago, just as the AI revolution was coming to the world of art and design. At the time, Canva had escaped a lot of the criticism leveled at competitors adding AI tools. Something Melanie attributed both to how much Canva users love the product and also the fact that the company is making huge inroads into the business world. Well, now it's 2 years later and it's safe to say that AI is all over modern design software and a lot more people have had a lot more feelings about AI in general. But Melanie and Canva are pushing even more aggressively into integrating AI. The company just announced a big new update that allows people to simply tell Canva what to make and have it go through various data sources like Slack and email to build presentations, documents, and other design materials. Those projects arrive as regular old Canva files, which you can edit at will. You'll hear Melanie come back to that idea several times that having the output of the Canva AI system be regular old Canva files is a big deal so that users can refine that work and make it what they actually need. The idea here, to borrow the company's tagline, is that Canva is quote moving from a design platform with AI tools to an AI platform with design tools. I'll let you all sit with that for a minute. Obviously, I dug into all these ideas with Melanie and how she's thinking about Canva's relationship to the AI model providers, the cost of the token required to automate an app like Canva in this way, and the kinds of pricing that might lead to for users. That's the same idea as a lot of other big AI players aiming for corporate dollars. And so Melanie and I talked a lot about whether Canva is the right platform to bring everything all together. Unsurprisingly, she thinks it is, not least because she runs Canva using Canva. There's a lot in this one. Like I said, I always enjoy talking to Melanie. Okay. Canva CEO Melanie Perkins. Here we go. Melanie Perkins, you are the founder and CEO of Canva. Welcome back to Decoder.
Melanie Perkins 2:39 ↗
Thank you so much for having me. It's great to be here.
Neil Patel 2:41 ↗
I am very excited to talk to you again. It's been a couple years. You were last on the show in 2024. We talked about AI and design and the feelings people have about AI and design and I was looking at that interview again just to prepare for this one and a lot of the themes are all the same and then the facts surrounding those themes have changed so dramatically in the past two years and on top of that you have big news that I really want to dig into. So let's just start at the start. The last time you were on the show, I said, 'What is Canva?' And you said Canva is an online design platform. And your news this week is that I believe the company is changing its own conception of itself. Tell us about that change and what led into it.
Melanie Perkins 3:20 ↗
So there are some things that are changing and there are many things that remain the same. So our mission is still to empower the world to design and we're going to be doing that very much over the years to come. But something that we've always believed is that we should take the latest and greatest technology. We should build the latest and greatest technology and put that into our community's hands and enable them to achieve their goals. And what is the latest and greatest technology has certainly changed over the last few years. And so obviously AI is at the center of that change. But obviously the latest and greatest technology today is AI. And so we're really excited to be doubling down into that space.
Neil Patel 3:57 ↗
All right. But I'm looking at a press release that says we're moving from a design platform with AI tools to an AI platform with design tools. That seems like more than bringing the latest and greatest technology, right? It seems like a rethinking of what Canva's product is. Unpack that a little bit for me.
Melanie Perkins 4:14 ↗
Okay, so let's get into it. So when we launched Canva for the very first time, one of the huge innovations that we had was moving from pixels where everything was very granular and required deep expertise to be able to move anything around to be able to design anything to objects where you could lay out a design. You could just have ideas for different objects. You could search our stock photography library, our illustration library. You could drag it onto the page. You start with a template or start from scratch. You could collaborate and design. And now what we're really excited about is with AI, we're moving into the concept layer. So you can just take an idea, you can write it in, and then something can get created for you. But very importantly, you can still move into the canvas object editor and lay things out, collaborate, edit away. And so we're really excited about bringing it to this third tier of concept editing, which we think will be extraordinarily exciting. So it's our biggest launch ever and it becoming the system where work happens end to end. But still very importantly with design at its core. I was going deep the other day into the definition of design and to design is to mark an idea and really to mark an idea is at the essence of design. So we're really excited about bringing new tools and capabilities to be able to do exactly that.
Neil Patel 5:28 ↗
So I have to ask you I'm looking at the presentation about all of this. It was obviously made in Canva. I know you told me last time that the whole company works in Canva. Did you automate the creation of your own deck announcing the AI tools or did you make it all by hand?
Melanie Perkins 5:42 ↗
So, what's really cool about this new product release, it can be one-shot generation and that is awesome. But the really exciting thing is it's actually also iterative. So, it can lay out pages. So, for example, you can take huge passages of text and then you can just lay that out with Canva AI. So, you can actually be your companion, your creative partner as you're going through the process. So, we didn't do it to just one-shot generation for the entire deck, I have to say. But what we were able to do is like to use it for all the fine-grain edits, the laying out of boxes, and that sort of thing. So, it really helped with the deck. But I think that's the exciting thing is that I think one-shot generation is sort of like AI 1.0 and being able to do iterative agentic orchestration is really 2.0. So, we're really excited about that. And then turn it into the press release doc. And it's really great at helping to create like that first draft for us and then we can use that to iterate, to collaborate because I think we both certainly know and everyone knows that one-shot generation might be a helpful starting point but that really is the draft to then be able to iterate and refine from there.
Neil Patel 6:49 ↗
Yeah, I'm curious for this. You know I'm just looking at pictures of the interface. It looks like a chatbot. You can ask it all kinds of questions as you showed off. Make me a content plan, do a bunch of stuff for me out on these platforms. You can connect directly to the platforms and have it publish for you. That feels like in particular the cutting edge of marketing, right? Is automating the creation of assets and the publishing to platforms and collecting the data and iterating through that, but the interface is still a chatbot and it feels like maybe that's going to be the interface for everything forever. Did you experiment with other kinds of interfaces or is it just the open-ended text boxes, the end-all be-all of AI?
Melanie Perkins 7:28 ↗
So, I think that's where, you know, I was going through those three tiers of pixels, objects, and concepts. I think that's what's really exciting to me is that in most chat bots out there, you kind of like in a chat and you kind of go backwards and forwards asking for the same thing and it will regenerate the entire thing over and over again. It's kind of annoying. But with Canva AI, you've got the ability to have the conversational editing which is extraordinarily powerful and brings completely new capabilities. But then you have the normal Canva that you know and love where you can just drag and drop, you can collaborate, you can do all your iterative editing, you can go and change a word here and update that and not having to prompt to do that. So it actually helps to make complex things simple by bringing it all together into one spot. So rather than having to... you'll see in the interface there's kind of AI, it's like a brand new tab inside the editor and so you can go there, you can dictate into your phone, you can do it on the fly, get that first pass and then it will lay it out just in the normal Canva that you know and love and then you can just edit that as you would typically do. After a lot of experimentation that was where we landed that it just... it's so powerful to be able to dictate, you know, for everyone's different accessibility needs, even accessibility needs on a day-to-day basis like sometimes now I can just be like talking to my phone, ask it to generate something and you can just do that on the fly. But then that creates a normal Canva design that you can collaborate, you can edit, you can use our 100 million plus stock photos and illustrations and drag and drop and design that. So you know the really the huge opportunity is this end-to-end workflow of being able to take an idea and turn that into a finished usable work in one seamless platform. Yeah.
Neil Patel 9:04 ↗
Yeah. Can I ask just a question about the relationship of the AI to the tools in Canva? And I'm gonna basically just do personal tech support with you. So I used Canva this week. My daughter's having a detective themed birthday party and so we took photos of all our friends and we're going to make wanted posters and so I was like I'm going to talk to Melanie and I better use Canva to do this, right? It seemed very natural. So I could ask you very tech support questions. And it just in the version of Canva that I was using, it was clear that the AI tools operated in some places and not others. Right? They weren't seeing the whole Canva tool palette and very simply is background remover, which I believe is like one of your most popular tools. It's everyone's most popular tool. I could do it in some parts of Canva, not the other. I couldn't look at my sort of like finished layout and say, 'Actually, can you just go ahead and remove the background from this photo?' to get to where I needed to be and then asked the question is the new Canva AI can it address the whole set of tools is it using Canva as a whole or is it still narrowly sliced?
Melanie Perkins 10:05 ↗
You hit the nail on the head with what we were doing with Canva AI 2.0. So it actually goes across the entirety of Canva. And it means so... you using Canva AI 1.0. I'm very excited to get your hands on Canva AI 2.0. We'll have to get you into the million. Though I guess by the time this podcast out we maybe we'll have to make a special exception because it will help you easily with exactly that. And so you can say for your example for your wanted posters, create me the wanted poster and you can upload the photos and it can actually orchestrate all of the different tools in Canva to be able to create that on the fly without you having to go to the different spots. But you can still go and like edit the different particular parts, the element editing if you want. But it actually is able to orchestrate it and then create a layered file in Canva's standard format.
Neil Patel 10:54 ↗
I think I understand how the user will see it. I think architecturally I'm very curious how you build the product that way because it doesn't seem like there's like some industry standard way of saying now you can use the software. Like about half of the attempts I see are we're just going to take screenshots of everything and very slowly click around and there's an infinite number of variations on that approach. There's the MCP approach which everyone was really high on and seems to have arrived at whatever point it's going to arrive at and now maybe half the industry is back at well we should just do APIs. What approach did you land on?
Melanie Perkins 11:31 ↗
So I think the reason we've been able to make so much progress in this space firstly was the decade of investment in this interoperable format. So being able to have this design format that spans presentations and whiteboards and docs and videos, the full gamut has been a really powerful part of why when we launched the foundational model, the design foundational model, it actually is able to create across all of these different formats and is that layered file. Which means that you can operate at a full design level, you can operate on a page level, you can operate on a like a photo level or text. And so the huge investment in that space is why we've been able to bring this to life with Canva AI 2.0. And there's an extraordinary amount of complexity behind the scenes. You know, we've had hundreds of people working on this project for some years to get to this point in time. But I think that the really important part is, you know, one of our engineers described it as an orchestra because there's so many tools and systems under the hood that need to talk together to be able to bring that thing to life. So when you say, I want to create a wanted poster for my daughter's birthday party, it will then be able to go and use background remover, it will be able to go and use all the different tools to be able to assemble that. But from a user standpoint, they just get to say what they want and then we go and do the hard work to achieve that goal.
Neil Patel 12:53 ↗
I'm just curious what bet you made there because that it feels like the industry has not coalesced on a strategy. So is it actually clicking around Canva or is there some other way of the AI addressing the tools?
Melanie Perkins 13:06 ↗
I won't go into technical detail there because I think that we have had a few breakthroughs that made this all possible.
Neil Patel 13:16 ↗
I understand. That's why I'm asking. The other thing that I'm very curious about...
Melanie Perkins 13:21 ↗
You have good questions. I will give you that.
Neil Patel 13:23 ↗
That's why everybody comes back. It's surprising. But it turns out it's a fun game to play to see if I can get through the media training. Melanie, the other question I have is who the model providers that you have doing this are, right? Because we are hearing every single day that token use rates for agentic software are through the roof. We're watching Anthropic have to modify its pricing. There's all kinds of stuff happening in that world and you're launching an agentic AI product that just from the interface alone makes you want to use it a lot. But it's not like... and you know Anthropic literally has in Claude there's like a usage meter and it will tell you like you're done now or pay us more money. Are you going to have like a token usage meter in Canva in the same way?
Melanie Perkins 14:09 ↗
Yeah, so you ask so many questions in that very short perspective time.
Neil Patel 14:15 ↗
Although there's more to come. Don't worry.
Melanie Perkins 14:18 ↗
We have been investing in the areas that we really need to. So becoming domain experts in design has been a really critical part of our research strategy. But then partnering with incredible companies that are spending billions of dollars to build the best in their own areas and then bringing that technology onto Canva is also a key part of the way we're approaching this. Being experts in design because that's where we really need to specialize because there isn't great technology in that space and then we've got a 100 person research team working very specifically on these problems themselves. On the AI credit front we have different tiering available for each of the different packages. So in free you get a certain number of... you get limited credits and then in pro we get much more generous and then in our business package you get far more generous and enterprise more so. But actually for the first million users we're giving everyone an AI pass which we're really excited about. So it's a $100 monthly pass we're going to be giving everyone in that first million so they can just go completely wild and test out all these new products. So we're really excited about to see how that is used and see where it takes them.
Neil Patel 15:26 ↗
Don't worry, we're gonna... I want to come back to pricing because I have a lot of questions about it, but first I just kind of want to understand the product a little bit more. The last time you were on the show, you were making the inroads to enterprise, right? You had sort of relaunched for enterprise and we talked a lot about how what you needed to do for enterprise was not necessarily product focused, but just workflow focused. You needed user authentication systems and management systems and dashboards and all that stuff. And you built it out and that seems to be going really well. I think the numbers I have here are you're at 4 billion in annualized revenue, 500 million of which is enterprise. So in two years you've grown. Is that the part of the business that's growing the most?
Melanie Perkins 16:03 ↗
The whole company is growing very rapidly, but yes, enterprise has been growing extremely rapidly. We grew by 100% over the last year, 95% of Fortune 500 companies and getting really deep footprints with thousands of people at companies now, which is extraordinary to see. We think that this with Canva AI 2.0, you know, it will radically change that. It will be a huge step change again and becoming the system at the center of work. Really bring things together. I think a lot of people can relate right now. It feels like there's a lot of fragmented systems, things that are kind of in lots of different places. Being able to have that all in one platform, having that with all of the work and all the designs and presentations and documents all in one place and with connect is being able to go even further and pull in context and information from your Gmail or your Slack we think is going to be a huge step change for the way work gets done.
Neil Patel 16:52 ↗
So that's the part I'm really interested in. The idea that a company is just a collection of disparate databases that are not well organized or managed and that there's truth in those databases if only we could read them all at the same time. That's a big part of like the AI thesis in general. Like you hear it all over the place. I, you know, I work with a bunch of cranky reporters. I don't think they put all their ideas in the databases, but like I get it, right? There's a sense that there's a lot of opportunity in the disparate data sources in a company and you can bring them together to platform and then take action on it and achieve some results. Is Canva the right tool to do that work? Right. You're kind of like right up against Claude or you're right up against I don't know like Oracle whatever like whatever big enterprise business process automation vendor is going to say AI will connect all your databases and then there's Canva and I'm wondering if you want the whole opportunity or just the design opportunity.
Melanie Perkins 17:53 ↗
Well, to me, design, as we've just talked about before, is like bringing creativity and productivity together and being able to do that in a way that we think is pretty extraordinarily powerful. I mean, I had my own experience of this the other day, which actually blew my mind. I had to answer a whole bunch of questions and they were going into like all sorts of different questions over the last decade and then I was able to just type it into Canva AI 2.0, know each of the questions and it was able to construct answers based on all of my designs and all of my documents from the last decade and it kind of blew my mind that this is the only place that actually has this information about me. And so being able to have that full visual suite from docs to sheets, whiteboards, presentations, all of that context. And then I guess the other thing is that when you think about it, most things kind of end up in a design format of some description at the end of the process. And so being able to have all of that context right there beside the AI tools, we think is pretty powerful.
Neil Patel 18:59 ↗
Yeah, I think the thing I'm curious about is where the primary interface for that lives and you're obviously making the case that it should be Canva. It clearly is for the CEO of Canva inside of Canva, but I could bring the CEO of Slack on here and they would happily tell you that that is Slack or I don't know, Microsoft will tell you that they're going to force feed Copilot to you wherever you are using a Microsoft product and that's where that should be. There's a lot of ideas about this and one of the things that makes that messier in my mind is all of these products can now talk to each other in very specific ways. So Canva itself is a plugin for the other chat bots and the usage of that plugin is very high. How do you think about who owns the interface in a world where the core tool set might be usable somewhere else entirely that also has access to all that data and all that information that the company might have generated?
Melanie Perkins 19:54 ↗
The key focus for us is always like how do we empower our community the most? How do we help them to achieve their goals? So, we're already embedded in organizations and businesses all around the world. And you know, when they're creating a design today in Canva, it's like quite a manual process. You have to go to all these different fragmented tools, collect all the information. And so being able to have that just inside the design tools we think will make a great deal of sense because it means that you're not... it's just like cutting down manual and busy work which is kind of always a thing that we're doing for our customers. Like when we launched, you know, some years ago like 2019 we launched background remover and the whole point of that was you click the background remove button and then the background was removed and like that reduced a lot of manual work. And then again like with this release it's kind of the same sort of thing. It's like there's a lot of manual work to go and collect all the information, collect all of the context all in different places. And so having that just there where you're designing, we think makes a lot of sense where you've already got like huge repositories of your images across your company, where you've already got all your brand templates, where you're already doing the collaboration, we think that makes a lot of sense. But really, we just want to be putting the tools that help to reduce busy work in the hands of our community and helping them to achieve their goals with less clicks.
Neil Patel 21:10 ↗
A few weeks ago, we had the CEO of Okta on the show, Todd McKinnon, and he was like, 'The future of Okta is managing agent permissions because this is a security nightmare, and I will sell kill switches to every enterprise that has agents running rampant over its networks and databases.' And so, I hear what you're saying. It's like, okay, Slack is gonna have a bunch of agents that can go talk to Canva's database of images. Canva will have a bunch of agents that can go talk to Slack's database of conversations. Something else is going to happen over here. Does that seem like a workable picture of a company of the future where all of these tools are sort of accessing one another independently or do you think it will naturally land on just one?
Melanie Perkins 21:52 ↗
I think there will be... I think the cool thing is for consumers there's going to be choice about how they want to have their work stack set up. I think it's a really exciting time in technology because there's just so many new possibilities for the way work gets done to reduce fragmentation. You know, we've got a quarter of a billion people using Canva today. So, we think there's a huge opportunity to make AI simple and accessible just like we did with design. But very importantly helping to empower people to achieve their goals and to communicate their ideas. So, we think we're pretty excited about what we're going to be able to bring out into the world.
Neil Patel 22:26 ↗
Yeah. How does it work for you? What's the dynamic inside of Canva? Like you're obviously on the bleeding edge of this technology and you obviously have your own tool. How does it work for you?
Melanie Perkins 22:36 ↗
In what sense? Sorry.
Neil Patel 22:37 ↗
Do all of your tools have AI access to all the other tools or do you work only in Canva and let Canva AI go talk to all the other tools? What's your setup?
Melanie Perkins 22:45 ↗
Oh yeah, since getting connectors, like so obviously Canva's always had all my designs and my presentations and my documents, but being able to get connectors and being able to pull in information has been pretty astonishing. So for example, being able to say, 'Hey, create me a plan for my next week and how I can optimize my time.' And it being able to go and read my calendar and then create me a document about my upcoming week. It was like there's a lot going on and it told me I had a massage booked and I was really surprised about that because I didn't actually know until I read that in my Canva doc and then I was like oh I think there's like a bug here and then I realized that my partner...
Neil Patel 23:23 ↗
The bug is it's booking self-care for you whenever it wants.
Melanie Perkins 23:27 ↗
And so I think it's really cool because there's a lot of things that would be like very very manual like going and doing a calendar audit and that all of a sudden can actually happen inside the one thing and it can actually create the presentation or it can create the document. And then you can have people collaborating on that. People talk about AI slop and I think the AI slop is often it's like one-shot generation. You kind of just take that and you put it somewhere else. I think that what Canva has really... it's really exciting about with Canva is that that is really just the draft. That's the starting point and then you can use it to iterate. You can use that, you know, through manual editing or you can use that through being able to iteratively edit through kind of AI inside the editor itself and like to refine it to really be able to clearly articulate your idea. So, we're pretty excited about the possibilities that it unlocks.
Neil Patel 24:14 ↗
I feel like it's time for the Decoder questions because you've talked about how much you use Canva internally. The last time I asked you how you make decisions, you said you had a process called decision decks where you literally made Canva documents with all the pros and cons and you mocked up the products. Is that still the process?
Melanie Perkins 24:29 ↗
That is still the process. Prototyping has become a very key part of it. So often now there's like a workable prototype before anything gets launched. I think the really fun thing about... I don't know if I talked to you about the complex decision-making framework.
Neil Patel 24:42 ↗
No, this is new. This is definitely new.
Melanie Perkins 24:44 ↗
Okay. Well, this for anyone that needs to make complex decisions, I find this extremely helpful is like splitting it out into like what are the goals, then what are the options, what are the pros and cons for each of the options. But it's really fun because like now we have a template inside Canva which is the complex decision-making framework doc and you can literally just dictate using dictation through Canva AI and it will actually go and fill out this template. So there's a lot of really exciting ways of like how you can take your ideas and like the thoughts in your head and then have that distilled in a way that other people can see and understand which I guess is the essence of design. Like if you think about design in the sense of it previously, you know, design can sometimes be thought of as like making things look pretty, but really design is about expressing ideas and being able to communicate that effectively and being able to turn something from an idea into reality. And so we think all these new tools really help to facilitate that. I use Canva code all the time. I used to do a lot of mockups and now I use Canva code to create prototypes all the time for every idea that I have which is pretty powerful because it takes the idea far further than it could before.
Neil Patel 25:47 ↗
The last time I asked you about structure I asked the other Decoder question is how the companies are structured. Last time you were about 4,000 people and you described your structure as a very centralized product team and then lots and lots of local teams and the metaphor you used was a cupcake and you said we work on the cupcake and we make the cupcake bigger and all the local teams work on the icing. Is that still a metaphor?
Melanie Perkins 26:12 ↗
Yeah, that's a fair metaphor. That one's been around for... the cupcake and the icing is actually so applicable in so many different ways. These small empowered teams is really at the essence of how we get things done. We're very much a goal-oriented structure. So for example with Canva AI 2.0 we really brought everyone together across the company to achieve that goal and bring Canva AI 2.0 out into the world. We do like show and tells every week so everyone can share and get deep context on what's happening. I think that like goals has really been at the essence of how we've achieved anything over the last decade is like being able to rally around goals and have different team formations in order to achieve that.
Neil Patel 26:51 ↗
How many people is Canva now?
Melanie Perkins 26:54 ↗
Oh, latest stat about 5,000.
Neil Patel 26:57 ↗
Okay. So, you've been growing. I'm really curious about, you know, just in that context, decisions and structure, how you made the decision to say, 'Okay, we're going to do Canva 2.0 and we're going to lean heavily into AI the way that we're going to lean into AI,' right? That's a lot of people. It's a big decision. I imagine that there was a decision-making slide or a deck and then this feels like it sort of inherently is a top-down decision like we're all doing this like Melanie says we're all doing this. Walk me through that decision and walk me through any structure changes you had to make in order to accomplish it.
Melanie Perkins 27:29 ↗
Yeah, absolutely. So, I'm going to take us back to 2011 and to a deck that we had which was called Canva Chef before Canva was even called Canva. And when you... the first slide when you go into it, it's like what do you want to chef up today and then you could type into a search box and the idea was you could type whatever you wanted and then you'd pop into the editor and you could collaborate and you could have the editing tools. If we share it with you after this, you'll see it's like bizarrely similar to what we're launching today. So I guess this has been the dream for a really long time but the technical ability to do this has been hard. I'd say in 2017 we had this document we called it getting smart and we're like in the future there's going to be search-driven design rather than going to the buffet and getting something it'll be able to happen on the fly. Like a chef cooking something up from the raw ingredients. And now it feels like we can actually do that. Back in October, we've been researching into this space for some years with the foundational model that was a huge step. The design foundational model that was a huge key piece, but then in October, there was a significant breakthrough in the company that meant that we could actually do it. So as soon as we saw that, we were all like, 'Oh my goodness, this is really exciting and groundbreaking for what Canva can unlock.' And so that was when we really started to go all in and realized that that technology needed to be pushed as far as it could go, which is what we're launching today.
Neil Patel 29:02 ↗
Did you send out an email? Did you send out like a Canva deck being like, 'We're doing this now. Decision made.' How did that...?
Melanie Perkins 29:11 ↗
That is a great question. So there was a team working on it already and then we really bolstered up that team. So then we were like, okay, we need to get every single person that can possibly help bring this to life onto the project. And we started the weekly show and tells and we, you know, turned it into more of a centralized AI team with then hundreds of people that went from a smaller team to then many hundreds of people to bring it to life. Everyone working on the different parts that needed to become part of this orchestra. So yeah.
Neil Patel 29:44 ↗
Yeah. I'm curious like that part, that structure part seems really interesting to me. Right. We have a software tool, a standard sort of deterministic software tool with a select box and a text arrow, all that stuff and we're going to build an AI that can use that tool. Now we've got to take all the engineers we had and point them at that problem. Did you have to rethink your product team or did you just make it bigger? The team that was working on that part.
Melanie Perkins 30:09 ↗
I think a little bit of both. Once the team had this big breakthrough and we all saw it in action we're like my goodness this is the coolest thing ever and then we had...
To figure out who could actually help from across the company, and I think that's the goal-oriented structure I was mentioning before. When there's a goal, you need to figure out who are the people that can help bring this to life. And then everyone, that's where we were doing the weekly show and tell, so everyone can get a really clear understanding of where everything is at, all the pieces that needed to be orchestrated to come together. I think Canva already being interoperable meant that there was a lot of these things that had already been built and that then could just come together in an exciting way. We do something called the Canva Jigsaw that we've been doing different variations of since the earliest days, which is like it's often a goal and then all the pieces that need to be worked on independently to be able to bring that to life. And that was exactly what we had at the center of this project.
Neil Patel 31:12 ↗
You're fundamentally the software CEO. I think that's a fair description. I think you make software. The nature of software development itself seems to be going under some kind of existential crisis. One of our designers here at the Verge Media described all software development now as calibrating yourself to a database and just like talking and seeing what happens and maybe that'll turn your brain to mush. Are you using as much Claude Code or Cursor to make Canva as it seems like every other company is racing to do?
Melanie Perkins 31:42 ↗
Yeah. So I use Canva Code really extensively from the perspective like...
Neil Patel 31:47 ↗
When you say Canva Code, that's your own coding product. You're not using Claude Code or...
Melanie Perkins 31:52 ↗
Because it's so cool. So what you can do now, rather I used to create mockups all the time. Like anytime I had an idea, I'd create a mockup. And now anytime I have an idea, I can use Canva Code. But with this latest release, you can actually go in and edit the text. So you can actually create something, you can edit the text, you can drag and drop, you can move things around. We've been really investing heavily on the AI front and upskilling our team.
Neil Patel 32:17 ↗
So can you make Canva with Canva Code?
Melanie Perkins 32:19 ↗
Yeah, we do make Canva with Canva Code, not deployed. We have many incredible engineers that actually make it sound to go out to hundreds of millions of people. But we use it for prototyping all the time.
Neil Patel 32:34 ↗
Yeah, I think that's the question I'm asking is more about those folks and how you think about the costs associated with those folks. The nature of software engineering is changing in some big meaningful way due to, in particular, the coding tools that are available. Are you rethinking how that works inside of Canva as you ship new versions of Canva? Because every other software CEO I talk to, their minds are exploding. They don't quite know how it's going to go, but they know it's definitely going to change forever.
Melanie Perkins 33:00 ↗
I think one of the things that we've invested really heavily in is continuously upskilling our team and systems. So we've taken a very intentional approach to give all of our team access to all of the latest and greatest tools. So we actually have not selected a winner. We have just given them everything and it's been very intentional because we want everyone to be playing with the latest and greatest and to be upskilling all the time. We need to be upskilling every one of our systems. The way we build product is completely different today. The way we do QA is completely different today. Every system and process inside the company has had to have an AI-native transformation. And every specialty inside the company has had to have an AI-native transformation. What a designer does today, what an engineer does today across every single part of the company. So it's been a huge area of investment on the tooling, on giving our team time on the specialties. We've had this focus on AI everywhere and then AI impact and now AI native because we really want to be rethinking everything in this AI era.
Neil Patel 34:04 ↗
There's some sort of rebalancing of power between product managers, designers, and engineers because AI lets them all do each other's jobs. Where have you landed on that inside of Canva?
Melanie Perkins 34:15 ↗
I think we're all here to build the best experience we can. And I think having really siloed expertise has never been the best way to build product. In fact, great PMs often think about the design perspective. Great engineers often think about things from a design perspective. So really it's about the team that is there to just create the best thing possible. Having people in their separate siloed isolated lanes and saying that's my territory was never a great way to build product. And so I think that with AI it's really leaning further into that, is everyone thinking about what is the best product experience that we can build. And everyone will bring different skills to the fore. A designer will obviously have a certain expertise, a PM will have certain expertise, an engineer will have certain expertise. But we've always thought of it as a bit of a team sport where the best idea should be winning. And everyone should be collaborating to create the best outcome that they possibly can for our community.
Neil Patel 35:08 ↗
I understand the positive case for AI, why you made the decision. I understand the product promise of just tell this box what to make and it will make you a first draft and you can go on from there based on the data you have. There's a pretty significant downside to AI, particularly as it relates to brand, right? There's polling here in the United States at least that basically is just bad vibes around AI. Like the last NBC News poll that we are constantly citing is AI is polling under ICE in terms of favorability and just above the war in Iran. That's not a great place for AI to be as a concept that is favorable or not favorable. People are voting against data centers in their communities. AI is widely associated with job loss and maybe now you're going to cause some enterprise job loss because social media teams don't need to be as big as they needed to be anymore. There's a lot of layoffs being blamed on AI across the board. You're leaning into AI with Canva, right? You're rebranding the whole product as having AI in it. How do you think about that downside risk that people don't like it? The more they're exposed to it, the more they're saying, 'Wait, stop. I don't want this around me.'
Melanie Perkins 36:13 ↗
I think it's kind of like any tool, it will be whatever you want it to be. And so if you want it to help empower people, if you want it to help deliver better experiences for your customers, if you want it to uplift students and to give them great quality education materials, it can do that. And I think that's...

69 more exchanges in this transcript

Sign in free to read the rest of this interview. No card required.

Sign in to read the full transcript

Cite this transcript

APA, MLA, BibTeX
APA

Perkins, M. (2026, April 20). Melanie Perkins Canva's CEO on its big pivot to AI enterprise software | Decoder [Interview transcript]. Decoder with Nilay Patel and The Verge. CEOInterviews.AI. https://ceointerviews.ai/interview/848172/

MLA

Melanie Perkins. "Melanie Perkins Canva's CEO on its big pivot to AI enterprise software | Decoder." Decoder with Nilay Patel and The Verge, 20 Apr. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/848172/.

BibTeX
@misc{perkins2026_848172,
  author       = {Melanie Perkins},
  title        = {Melanie Perkins Canva's CEO on its big pivot to AI enterprise software | Decoder},
  howpublished = {Interview transcript, Decoder with Nilay Patel and The Verge. CEOInterviews.AI},
  year         = {2026},
  month        = {apr},
  url          = {https://ceointerviews.ai/interview/848172/},
  note         = {Speaker-attributed transcript with timestamps}
}