AI Product Masterclass in 45 Minutes from Canva's Co-Founder | Cameron Adams
48:01

AI Product Masterclass in 45 Minutes from Canva's Co-Founder | Cameron Adams

Peter Yang 29.06.2025 2 065 просмотров 49 лайков обн. 18.02.2026
Поделиться Telegram VK Бот
Транскрипт Скачать .md
Анализ с AI
Описание видео
Today, I want to share a new episode with Cameron Adams. Cameron leads product for Canva where people are creating 1B+ designs every month. We talked about how AI will transform design and why he believes in prototypes over PRDs and coaches over managers. I also loved hearing Cam’s stories about Canva’s rubber ducky easter egg and what a bike accident taught him about persistence. This episode is brought to you by Vanta — Join 9K+ companies like Atlassian who use Vanta to manage risk and prove security in real-time. Get $1000 off at https://vanta.com/peter Timestamps: (00:00) Will AI democratize design? (06:32) How Canva runs AI evaluations for its product (10:17) Moving beyond chat boxes to visual AI interfaces (22:23) Why Canva skips PRDs and goes straight to prototypes (29:08) The rubber duck Easter egg that became a cultural icon (33:13) Keeping craft and quality as a priority as the team grows (35:32) Why Canva gives everyone executive coaching, not just managers (39:23) How AI is blurring boundaries between PM and design (43:26) The bike accident story that almost derailed Canva's launch Get the takeaways: https://creatoreconomy.so/p/canva-co-founder-ai-playbook-most-popular-design-tool-cameron-adams Where to find Cameron: LinkedIn: https://www.linkedin.com/in/themaninblue/ Website: https://www.canva.com/ai-assistant/ 📌 Subscribe to this channel – more interviews coming soon!

Оглавление (9 сегментов)

  1. 0:00 Will AI democratize design? 1194 сл.
  2. 6:32 How Canva runs AI evaluations for its product 714 сл.
  3. 10:17 Moving beyond chat boxes to visual AI interfaces 2019 сл.
  4. 22:23 Why Canva skips PRDs and goes straight to prototypes 1251 сл.
  5. 29:08 The rubber duck Easter egg that became a cultural icon 795 сл.
  6. 33:13 Keeping craft and quality as a priority as the team grows 452 сл.
  7. 35:32 Why Canva gives everyone executive coaching, not just managers 690 сл.
  8. 39:23 How AI is blurring boundaries between PM and design 766 сл.
  9. 43:26 The bike accident story that almost derailed Canva's launch 872 сл.
0:00

Will AI democratize design?

We have been doing AI since before it was cool. Cameron, a co-founder and chief product officer of Canva, but how do you think about like the simplicity and approachability when it comes to AI? Two weeks later, we started getting all these reports from our customers about rubber ducks and we were thinking like what the hell are you people talking about? How do you keep that uh culture alive? value called empower others which means giving people the power to make choices and to come up with decisions and push the product in the way that they want to push it. It was a busy Saturday night in the city. I was coming down a hill and this taxi just came out of nowhere and t-boned me and I was just like collapsed on the road. Ended up in the ambulance, had to go to hospital, got stitches, but we still needed code shipped for the launch on Tuesday. Okay, welcome everyone. My guest today is Cameron, a co-founder and chief product officer of Canva. Canva is the most popular design platform in the world with I think over a billion designs created a month. We're super excited to chat with Cameron about uh Canvas's AI features, maybe how AI will democratize the design and also Canvas's unique product culture that prioritizes coaches over managers. So, welcome Cameron. Thank you, Peter. It's awesome to be here. to start, can you describe Canva AI at a high level and kind of what kind of products exist there? Yeah, so Canva itself is a design tool for pretty much designing anything and we've built it up over a decade now to be able to cover a whole sway of visual content from presentations to social media graphics to t-shirts to videos and even websites. And I think in the last few years, you've seen the massive rise of generative AI. Um, and we have been doing AI since before it was cool. And this year, we kind of integrated all of our AI features into the one interface. So, you can now access Canva AI from the homepage. And it gives you access to all of the power of Canva to help you generate any visual content that you need, whether that's a social post or a presentation or a document that you need to work on. Uh we've even integrated interactive code widgets that you can include in Canberra as well. So you can access all of these things from the same AI conversational interface. Um, and it's been awesome for helping people really get from zero to one quickly. When they have an idea for a design that they want to create, they can just converse with Camber AI, get it created, um, pretty much get it to the 80% that you most often get out of AI systems, and then you can lean on the rest of Canva for the final mile. So when you want to tweak some of the text or change some of the colors or upload an image that you specifically want to have in your presentation, you can do that through the rest of the Canva tool and then publish it exactly as you would through any other Canva design. Yeah, I like I think the final mile is actually really important. Like when I generate an image or something on Chat GPT, like I have to keep prompting AI to update it. But I really love how Cam AI you can just like go into the editor afterwards, tweak some text or like add some images. I think it makes a huge difference, you know. Yeah, I think prompt box is definitely a great interface for quickly getting an idea together, but then for drilling into the details becomes really cumbersome and that's where you really need the power of a tool like Canva to be able to just go in and do the exact thing that you want. Mhm. So, um, you know, people watching this are a lot of like product leaders and product builders and maybe I would love to hear if you can talk a bit about behind the scenes about how this stuff works like for example for the text to design feature like what actually happens when I submit a prompt like does it use some uh predefined model or do you think can share? Yeah. So, when you're submitting a design so I mean with Camber AI generally there's a lot of different things it can do. So it can generate a flat image for you. fully editable design for you. Uh it can generate a document. It can generate a code snippet um that you can use as a widget on a website. And when you're kind of figuring out which of those things you need to create, the AI needs to conduct itself to the area which it thinks you're going to go in. So if you're looking to generate an image, that kind of takes a different path to if you want to generate a design or bit of code. Um so the first thing that our product needs to do is figure out your intent. Um figure out which part of Canva you actually want to need and which model that you might be interacting with. Um say you're generating an image. So you just want to generate, you know, just a normal image. Um the AI will figure that out. uh and that will then start conducting your prompts over to our image model. Uh our image model is actually based on a company that we acquired last year called Leonardo and they have a foundational image model that they have developed uh which is super high quality um focuses a lot on creativity and also being able to take particular styles that you might pass to it and use that as inspiration for the images it generates. So we take all that context and then start piping it over to that image foundational model uh called Phoenix which Leonardo has developed. If you're looking at creating a design, we actually funnel that over to a different model that we've worked on and design is our bread and butter here at Canvas. So uh we have our own model that we use to generate designs based off prompts. Uh I should say that underlying this kind of the fundamental conversational aspects and the more kind of textual LLM bits that you get out of the chat interface we rely on open AI for. Um so we interact with their GBT via an API. Um so that kind of underlies the conversational aspects but then we're passing the different inputs and outputs over to the particular models that you need to work with to generate the visual content that you need uh out of Canva. Got it. And like there's so many diverse use cases for Canva even just focus on the design like how do you evaluate whether it's AI products doing it job or not? Like do you have some uh manual
6:32

How Canva runs AI evaluations for its product

evals or like you know user focused evals that you run? Yes. So it's been a real progression. So we've had something like Canberra AI in the uh Canva interface for probably about 3 years now. Um, we started off with something called magic design, which you went to specifically when you wanted to create a design and we've kind of graduated that up into a more general model that can take pretty much any input now and figure out where you need to go and what you want to create. Uh, over that time we've gradually opened up the kind of amount of things that you can create with. Uh, and I think that scoping in the early days really helps you nail down uh the quality of what you're producing. So in the early days we focused a lot on social media templates because that was where we saw a lot of people needing to quickly get some inspiration and get a design out of it. Y um so we focused a lot on building quality for social media type queries. Um from there we've kind of moved outwards through all the design types that you can create in Canva. So now you can create a presentation which is kind of more complex than a social media template because you need this multi-page narrative experience. Uh we've also worked on full uh documents. So we have Canva docs which focus a lot on more lengthy textual documents but still with strong visual elements throughout them. Um and again it's a different style of creation. It's a different type of narrative that you need to produce. Um so we work specifically on those different dock types and what we need to pass into the AI models to get decent outputs and then of course tuning the outputs. uh based on evaluations that we do to make sure that you're getting quality output from it. Um so I think if you kind of start with a smaller problem and then gradually work your way out as you uh want to uh eat up more of the world, it enables you to control that quality that you're getting out of the AI and also to figure out the different paths that people take through there. For us, figuring out those paths to different designs or different jobs that people need to do enable us to tailor the experience a lot more so that we know if you're heading down a design path, we can frame the conversation in a certain way or we can pass it off to a different model to get you a different output versus if you're looking at an image or you want to do a video or maybe you want to um create some code. Got it. and um I'm sure you like manly tested the product yourself. Uh but how do you like track whether like a user has moved on accepted the AI generated design? Is it like when the user actually moves to like the editing stage that that's kind like a good generation or like how do you have some sort of bar for this? Yeah, for us there's a couple of activation points. There's uh either downloading a design or an image. Um, so that's a very strong signal that they've accepted it or yes, moving into the Canva editor. And that's kind of a key one for us because moving into the Canva editor means they're happy with what they've got out of the chat interface, but they want to start uh editing in certain ways. Um, so once they've moved into the camera editor, we can start uh capturing a lot more data about what they're working on and helping them tweak the design itself. So it's probably those two main activation points for us. Got it. Okay, that makes sense. And one of the best things about Canva is just like how approachable these tools are, right? Like um like you know when I use another design software like there's like auto layout and all this kind of stuff that like just restricts me as a non-designer from using this stuff. So canvas is like just like very approachable. But how do
10:17

Moving beyond chat boxes to visual AI interfaces

you think about like the simplicity and approachability when it comes to AI? Like you know we're starting with like a chat box, right? We can just generate stuff. But do you think it's going to evolve or the combination of things? Yeah. It's been really interesting to see people I think evolve alongside the chat box as well. Uh back in, you know, two years ago, you can put a prompt box in front of someone and they would be scared out of their wits as to what they want to type in and just be like frozen with paralysis. I think we've all gotten a lot more used to the prompt box as a great way of starting a conversation and starting to explore an idea, but particularly in the visual realm, we've seen that a lot of people don't have the vocabulary that they need to express their idea or it just gets incredibly complicated and the way that you would describe something would probably take like five pages of text to describe it. Um, so I think we're particularly in our realm, we're moving towards augmented interfaces that enable you to move from the chat into something that's a bit more visual and start to direct the AI a bit more. Good example of this is when you're working with that image generation model, you can now supply a reference image. Uh, and from that reference image, you can pull a bunch of different things. So firstly, you might be interested in the style of the image. So say the photo that you're uploading has this nice golden glow to it. It's a bit retro and nostalgic. You can pass that in to the image generation model and then tell it to generate images with that same style as well. That would probably be something that's really hard to describe and capture in an accurate way purely through text. So giving people a visual way to do that is really important. Another way that you can use images to kind of structure your output is through um structure reference. So you might have an image that is the exact kind of layout that you want, but the to style is totally incorrect. Yeah. So through the Canva image model, you can upload an image and say this is the structure I want, but I want a totally different style. Um, and that's a really great way of passing in uh image data without having to type it because like telling you what's on my desk, how the laptops oriented, what the windows behind me like look like, the artwork on the wall, like being able to capture that in a text prompt is just going to be incredibly cumbersome. Yeah. Um, then I think there's lots of little ways as well of editing any visual content. When you're working with a presentation, telling it to change this one word on the 15th slide that's in the bottom right corner, again, really cumbersome to do through a chat interface, but just being able to do that by pointing and clicking just as you normally would in a presentation is super valuable. And that's where we've seen the progression from an AI chat interface through to the more normal editing interface that you get in Canva being really valuable. Yeah. Um, and we've actually worked on that. um with the conversational interface that we have with Canva AI because when you're generating you can uh have a full screen conversation once you've started creating a visual artifact it actually creates this window off to the side where you can see the visual artifact be it a presentation or a social post you can continue conversing but you can also interact with the visual design over on the right so you can click on piece of text you can get it to change some colors and then you can go fully into the camera editor which kind of pushes the conversation off the screen and you're in full editing mode where you get to interact with everything. All right, let's do a little demo of Canva AI. It is available right from your homepage. It's actually part of this omni box that we now have which enables you to look through your designs, look through the millions of templates that we already have in our content library, or use Canberra AI, which is our new conversational interface, which enables you to describe your idea and bring it to life in whatever visuals you need. Today, I'm going to start with a bit of a demo of creating an image. So, one of the brilliant things with creating an image is that you can now actually reference an existing image that you have when you're generating new content. So, say I had this photo of my office. I wanted to use this photo to match the style. And there are two different things you can do when you match the style. You can use the colors and the feel of your image, or you can use the arrangement and the layout of your image. I'm going to select the second one which is match the design and I want to say can you please bring my office to life with pop plants and some more vibrant and engaging colors. I'll get Canva AI to kick that off. It will start looking at the image reference that I have uploaded and it will start thinking about how to turn that into something which matches the prompt which I've just put in. We will wait a second for it to generate those images. And there you go. Uh if you actually see the reference image that I uploaded, you can see this desk. I've got a nice anacronistic telephone here, monitor, window, etc. And the images that have been generated directly reference that. Uh, so we have the desk, we got the monitor, window, chair here. But it has restyled it with the exact prompt that I put in and made it a much more lovely office environment. Uh, second thing I want to demo with Canberra AI is actually creating a design. So I will click on design. For me, I should me should mention that you can just have a fully conversational uh interface with Camber AI, but if you know exactly what you want to do, you can click on the thing that you want to do. So, I know I want to create a design and I'm going to say, can you create an Instagram story ad for my DJ business? I'll kick that off. Uh Canva AI will go off and have a think. It knows an incredible amount about designs and communicating your content through visuals. Uh and it is putting all that together with the prompt that I have put it in. Uh and it's going to create a fully editable design that I can then take as a starting point. I can use it exactly as is or I can change it to what I want it to be. So, here are the designs it's created. Uh, if I click on this one, it brings it up. I can change any of the colors in there. It will restyle that based on the color palette which I've chosen. Uh, I can also open it in the Canva editor. So, I'll click that off. It opens it in the Canva editor. Everything on the page is editable. So, I can change this text to DJing. Uh, I can, uh, move the image around. I can make it bigger. Put that in. Done. I can move this text down. I can go search for some other elements. Let's say I want some headphones. Uh, kind of like those ones. I'll drop those in. Give it a bit of rotate. Put them there over there. And I can change this design however I want. Back in Canva AI, um, if I wanted to make some wholesale edits to this, I don't quite like the vibe of the designs it's generated, I can say, can you please make these designs just use blue because my DJ name is the man in blue and I want my ads to represent that. So, I'll kick that off. It will go off. Take a look at the design designs again. use that style information which I've just passed onto it. I want it to be a lot more blue. Here you can see it generating the designs already. It brings those in and it gives me some uh blue designs which I can continue editing or use just as is. So as you can see there quite blue. I'll take this one. I will use that in the Canva editor because I want to tweak uh the image size. I will u just boost up that image size there. I want this text to be a bit bigger. So I'll just go ahead and directly say bigger. Uh move that over and edit the text with our awesome DJ. Brilliant. There you have it. So it's a really great example of Camber AI enabling you to creatively brainstorm. get a bunch of designs put in front of you to kickstart your imagination and then using the normal Canva editor to go the very last mile and add those little finishing touches that you need in order to use the design wherever you want to send it. Very last thing I am going to demo is uh Canva code. So within Canva AI you can now get it to code for you. Uh, and Canva code is awesome for prototyping really interactive ideas. And we've seen so many Canva users bring their websites and their presentations to life with an interactive widget that they've made in Canva Code. Can you make me a dragable version of the solar system planets from left to right and I have to order them in the correct order. I will send camera AI off to help me code that. It will have a little think just identify that I'm creating an interactive coding session and it will start coding it for me. All right, there we have it. Some dragable planets that I can put into the interactive. So, where does Earth go? Third rock from the sun. Uh, Mercury, hottest planet. Venus, quite cloudy. Jupiter, it's the biggest. Saturn has rings. Oh, nice rings. Um, Uranus, I think that's over there. Neptune's there. Let's check the order. Not quite right. Try again. All right. Well, from those examples, I think you can see that we have taken a particularly visuallyoriented route to interacting with AI in Canva. as you'd imagine because Canva is all about visual content, but we're really focused on bringing all of your visuals to life from designs to images to even creating a doc or getting Canva AI to code for me. It's a pretty exciting leap in our product. Uh, and we were super excited to demo it a couple of months ago at Canva Create. And if you haven't tried it out, please dive in. That's awesome. Yeah, it's like different tools for different uh use cases, right? Yeah, the chat is not a great tool for everything. So, yeah. Um, so I' like to switch gears a little bit. I think um talk about like how you guys built this product behind the scenes a little bit and uh like your typical like your product development process, right? Like um I guess at a typical company you have uh I don't know like some sort of PRD and then you do like a bunch of like tickets and all this stuff, right? Do you guys especially now that you know you're building these AI products? I'm curious if uh you and the product team are like prototyping more or like streamline some of this process. Yeah, we've always been very visual thinkers at Canva. I think it's
22:23

Why Canva skips PRDs and goes straight to prototypes

the nature of our product and the way that we think and how we think about the world. Um so we attract visual thinkers to the company and it's really part of our product development process. Uh often we'll go to uh kind of product build without a textual description of what the product actually does because writing a bunch of paragraphs in a document for us doesn't capture what we want to create. But really critical to our product development is having a very signed off set of visuals related to the product. So, we always have very detailed mockups. We're talking about mock-ups constantly. We're interrogating them. We're making sure that they actually fit the experience that we want to ship. Uh, and before, you know, pretty much any line of code is touched, we need to have that visual artifact um that we know uh we want to ship to users and that we're certain is going to work. Um, for probably the first decade of Canva, there was always a really big delta between static mockups and prototypes. Um, and I've always been a huge fan of prototypes because I have both a design background and a computer science background. So, I kind of utilize both of those things to create very uh interactive prototypes to test out a bunch of assumptions, how the product is going to feel, how people interact with it. And that's been really important. But not everyone on the team has had those skills. We have a lot of designers who just u mainly work in the visual realm uh and then they have to work with an engineer to kind of bring a prototype um more to life. That's always been a friction point because um the iteration cycle between two people is always going to be slower than just one person. Uh and then engineers often come with the whole framework of the product with them. So they're thinking about how they're going to productionize it which slows down prototyping. you end up with uh kind of slower turnaround times. So figuring out ways to move beyond those mockups uh into something that people are actually interacting with has always been the dream of product development. Um and we've hacked around it in various ways. Like you can go to your design tool and do a couple of transitions and you click around which gives you a slightly better indication of what the product's going to be like but is still not as high fidelity as going to code. So what's been amazing in the last couple of years has been seeing the rise of for one of a better word vibe coding. Um and for anyone to be able to prototype an experience be that a product manager or a product designer or even engineers who want to quickly design an idea or code an idea without having to go through the full production cycle. Yeah. Um, and the turnaround time that we're now seeing with ideas and getting it in front of users, testing it with them, making sure that their experience matches what we were thinking when we first came up with the idea, uh, has been incredible. Um, so we're now moving prototyping further and further up the stream in terms of product development. Um, and we're pretty much in some cases almost skipping static mock-ups and just going straight from idea to prototype, getting that in front of someone, seeing whether it works, um, and then using that to inform the product development spec. I love that. Um, and I find that like at a lot of companies, um, like you can have a bunch of internal feedback loops, uh, for a long time before you even show a product to a real customer. You talked about getting in front of users multiple times. I'm curious, do you encourage your team to like just go talk to users of like a prototype or like is there like channels to do that or uh before they even come to you maybe like get some user validation? We're super optimized for user testing and research. So we want all of our product designers and our product managers to be putting stuff in front of potential customers uh as early as possible. Um, and we're now seeing that more frequently with prototypes because with static mockups, you kind of have to like handhold people through it of like, okay, this mockup is doing this. You're now on this screen. What would you do in your land here? What would you click on? Like it becomes very manual and quite biased to be honest because you've got someone guiding someone through the prototype. If you can actually hand them code and say, "Right, here's the new app. Just click around, see what you do. " if you wanted to create a video, how would you go about that? How would you know when you've gotten to a final state of a video? And just having them click around like it's a real app, truly observing their behavior. Um, and then being able to iterate really quickly on that, like seeing what people are doing, the next day, coming back with a prototype that's fixed the issues that you were seeing and retest your hypothesis. It's a really powerful form of product development and really exciting to see it rolling out across all of our teams and like if a user has some feedback of like you know I don't want I don't like this or I don't like that you can just like tell the AI to update the prototype, right? Exactly. You know, can you adjust for this? Yeah. Um do you spend a lot of time going back to the vision piece? You said like uh canvas is very visual, right? So like uh the product team used to present like a visual vision to get everyone excited. Yeah. Yeah, but you know like inevitably as you go through this like feedback loop process like you know people want different things and you learn more things. So do people like update the vision mock too uh along the way or like h how does that work like you know what I mean? Yeah. I think centralizing around that visual asset is super important for us and if you don't update it, you kind of end up with a lot of confusion towards the end of like what were you actually aiming for. Yeah. Um so making sure that the prototype whether it be static mockups or an actual coded prototype is kept up to date with those latest learnings is really important. And I think it's becoming easier and easier to do that. You saw it become pretty easy as components came in and people were able to drag and drop more of the experience and now with um coding being accessible to more people it is easy to make those updates. So it's not so much a lot of maintenance. It's just quickly getting it up having that prototype you know ready to interrogate by either the internal team or by external customers um and making sure that prototype represents what you actually want to ship to production. Got it. Okay. And um another unique part of your culture, I was talking to John, who I think is on your team. He told me to ask
29:08

The rubber duck Easter egg that became a cultural icon

you about the duck, the duck that flows by when people make I think when people make 100 designs or something. Yeah. So, uh can you talk about that and why you built it? Yeah. The duck is a really strong cultural moment for us. The duck actually floats by on our upload indicator every time you've uploaded 100 images. uh and it's this little Easter egg in the product that has come to represent um a whole bar of excellence and user engagement that we expect from our product. Um it was put into the product in the very early days. It was like six months after Canva launched. Um and it was actually a process between me and one of our front-end engineers because for a long time Canva didn't have an upload indicator. You drag an image in not much would happen. the app would churn a bit and then 5 seconds later your image would appear on the page. So, we needed some way to give people feedback that the action they' taken was actually being processed and that it would result in something. So, I designed this little upload indicator and I wanted to be slightly fun. So, I like said, "Can you make it fill up like water and kind of wobble around and wave as it gets to 100%. " Yeah. And I took it to one of our front-end engineers, uh, Patrick Lee, and I said to Pat, "Can you do this? " this and he's like, "Yeah, it looks pretty simple. I can do it. " So, he went away and coded it, came back a couple of hours later. It was this nicely wobbly, watery upload indicator and it filled up to 100%. Uh, we kind of high-fived each other, sent it off to production, and I didn't think too much more about it. Two weeks later, we started getting all these reports from our customers about rubber ducks. And we were thinking like, "What the hell are you people talking about? There's no rubber ducks in Canva. " Until someone in the office saw it. It was the little rubber duck that floated by on the upload indicator and it was this little tiny touch that Pat put in. He saw my upload indicator which was, you know, wobbly and watery and he's like, "It's kind of cool, but I just want to add in a little extra touch of whimsy. " And he put in the rubber duck as a bit of an Easter egg. He didn't tell anyone about it, and he shipped it off. Nice. And the reaction we got from people was amazing because a very small subset saw it. But when they did see it, it created this incredible connection particularly amongst our most rabid fans because they're the ones uploading the most images. And when they get to see this, they feel like an insider because they're doing something special. They've interacted with the product so many times that they're being rewarded for it. Yeah. And they also got this connection with us as people because we're all people making products that we want other people to use and they're a way of expressing ourselves. And the little rubber duck was a bit of personality that show shone through. So I use that as an example for a whole bunch of reasons. Um first is like really putting love and care into the product like thinking about all these moments in which you can make it better and more human. Part of the success of Canva has been our approachability, the feeling that we've built this product just for you and that we are people who care deeply about our craft and care deeply about helping you deliver on your craft and your goals as well. Um, and I think little touches like the rubber duck, they're cute, they're fun, they're whimsy, they let you know that it's not some monolithic um, robot that's just delivering product to you via the internet. It is actually a team of people in Sydney, Australia and all over the world creating a product that they love and they want you to feel the love as well. Um, and that kind of permeates our brand. Um, and makes people fall in love with Canva. And if you go on Twitter and search for Canva love, uh, you get an incredible number of people who feel that passion and that fervor and bring that to the product and then refer people in. like that word of mouth, that organic growth is actually something that still powers Canva today. Um, and it's I think that emotional part of productled growth isn't given enough attention.
33:13

Keeping craft and quality as a priority as the team grows

attention. Yeah. How do you keep that uh culture alive? uh you know as you scale the team like uh this craft and this taste and this like care like people giving a about things like how do you know how do you encourage this because it doesn't maybe doesn't help you achieve some OKR or like what whatever you're trying to do right how do you encourage this I think as with any part of culture you have to do it across a whole bunch of different levels so you need to talk about it a hell of a lot you need to hero it you need to point it out to people constantly as with any message at an organization, you need to say it at least 10 times before it gets through someone's skull. Um, so there's that aspect of it, leading by example. So, anytime we dive into product or we're working with a team or a design or an engineer, pushing them to include those moments and to think about how they can make the product, you know, 50% better than someone else would make it um is also something that you need to do. Um, and then we have another value called empower others, which means giving people the power to make choices and to come up with decisions and push the product in the way that they want to push it. Um, so empowering them to have that fun, have that whimsy, think about customers and this delightful experience they're going to push out to customers. Yeah. Uh, I think all of those things together make up culture and make up all these touch points that people feel when they're building product to Canva and know that they have to meet that bar. Uh, and that also they're given the power to meet that bar. Yeah, I think that's really important. Like I'm all about like I'm fond of growing metrics, but nothing gives me more happiness than just like a customer on Twitter saying like this guy really cared like he clearly cared, you know? Yeah, exactly. So, yeah. Is there any Easter eggs in the AI product or you want to keep it a secret? Uh, we'll keep it a secret, but yeah, there are a couple. Okay. All right. Hopefully people can find it. All right. Well, another unique part of the culture is uh like I heard like it's like very it's a very flat organization and uh it sounds like you want to give everyone executive coach chain. Maybe you want to have fewer managers. Maybe talk about that too. Yeah. Yes. So we had some pretty pivotal experiences with
35:32

Why Canva gives everyone executive coaching, not just managers

coaching early on uh ourselves as founders uh and we found it really eye opening and really I think collaborative and collegiate rather than being top down driven uh and command driven um and being able to democratize that style of coaching within Canva was something that we were really keen on. Um, so we figured out this system where you have a coach at Canva and it's strongly driven around growth and achieving goals and having conversations about that and kind of collaboratively setting those goals and understanding where someone is, how they want to improve, what skills they want to grow in, and laying out a pathway for them to do that. So we call it coaching on purpose because it isn't managing. It really is about thinking about the person holistically. um giving them a framework within which they can operate to think about the different skills they have, giving them uh plans quite regularly where they can identify the skills that they want to work on both personal and professional. Yeah. Um and then giving their coach their tools to help that person work through that and then create a plan for them that can help them hit those goals over the next six months or 12 months. and like most companies just reserve this stuff for like either execs or like high potential employees, right? But are you saying you give this coaching service to all employees or Yeah. So we um we've structured the company in a way that we uh specifically train up our coaches. Um so we have quite a few coaches across Canva. We've got uh I can't remember the last number, over 500. Wow. Um and they're embedded in the organizational structure. They're your senior product managers, your product designers, your principal engineers, like people of all stripes and they are coaching people within their teams. Um, so they're deeply embedded. They're normal canvanorts who we uh kind of give the skills of coaching to. We have a whole bunch of programs. They're constantly working on their own skills as coaches and how they can help their coaches go on their own uh journey to growth. Um and we give them a whole toolkit and a whole framework to work within to do that. Have you thought about uh including AI in some of this uh coaching or like you know training like for example I have a AI coach I just tell you about my life and then it gives me some advice for good or bad you know have doing that. Yeah that is a fantastic suggestion. I think individually lots of our coaches have um started to figure out their own ways to use AI tools in their coaching relationships. Um and I'm sure a bunch of coaches are also using AI to coach themselves as well in the in between times where they don't have facetime with their coach. Um but it's not yet something we've integrated fully into the coaching program. But it's a great uh prompt for me to uh go and move that along. Yeah. Uh especially like with like you know chat between memory and some of this stuff it learns about you over time for better or for worse and it can tell you some pretty surprising things about yourself. That's what I found. Yeah. Do you specifically go into a coaching session with it or it's just stuff that pops up along the way as you're going through your work? Um like I found this really good prompt. I think it's from Toby Shopify about like, hey, uh, you can ask it like, hey, tell me something about myself that I may not know about that that you think I should know or something like that. And then it tells you some pretty interesting things. Yeah. I won't ask you to divulge what your deep secrets are. Yeah. No, no. Um, okay. Cool. Let's see. How are you? Uh how how do you think uh we talked about this a little bit before but how do you think AI like
39:23

How AI is blurring boundaries between PM and design

other companies are like you know I want all my employees to use AI right but like how do you think this will change how your teams function you think people will like start doing each other's jobs a little bit more or like how do you think about it this for canvas product or yeah I definitely think it bleeds roles together a bit more and just makes it a lot more collaborative so we're seeing greater conversations between PMs and engineers designers and engineers, uh, designers and PMs. There's a lot more shared ownership, um, and just a lot more working together on the core problem rather than working on these little bits that normally take a lot of time to like iron out and polish in your craft. You're actually focusing on the bigger problem and solving the bigger problems, particularly for your customers. And I think that's one of the beauties of AI is it really helps teams come together um and speak more of a shared vernacular language. Um so you know particularly orienting around prototypes like everyone being able to see exactly what you're talking about is a massive leap forward and there's been you know over the last 20 years of my career that moment where someone actually sketches on a whiteboard the concept that everyone is talking about is this breakthrough moment in a lot of meetings where you're just kind of sitting there talking and like missing each other. Yeah. And visualizing those key concepts is actually a really powerful skill. It goes, you know, 10 times more for a prototype. Being able to prototype an idea that actually works rather than just a scribble on a whiteboard or some static mockups is the thousand% moment. and everyone being able to do that in a meeting, whether that's an engineer or a designer or a PM or even just a subject matter expert that you brought in, then being able to do that just moves the conversation forward so much more. Um, and being able to centralize around that prototype then enables you to have better conversations about the real problem that you're solving. Um, where you actually want to get users to and then rolling it out and getting real information rather than just sitting in your chair and positing what might happens. Uh it's really powerful and we've seen a lot of energy coming from our product teams off the back of AI because they're all getting involved a lot more um and just getting reinvigorated with the problem spaces that they're tackling. Yeah, it's like way more fun to like just like make a prototype and like show some users. Maybe you can invite the user into like the whiteboarding or prototyping sec session, right? It's like it's way more fun than trying to write a PRD around this stuff. Yeah. May maybe the PRD will finally die. I don't know. Maybe we just should write some edge cases, some notes around that prototype. That that's it. You know, exactly. You need a PR. Yeah. Maybe we should start a website death to prd. com which is all about new product development processes. Yeah. Yeah. Because like so much time going to like my philosophy is like if I'm going to spend a lot of time crafting something internally, like I want to spend the time crafting something that the company sorry that the customer can actually see. Like I don't want to craft some sort of internal strategy document that like nobody no real customer will ever see like you know yeah it's all I think it's we we've dealt with these layers of abstraction and you hack around them in ways to try and get it to work when really you just want the shortest path from idea to something in front of your customers as possible. And exactly you know we've built up product processes based around the tools that we have at our disposal. And when you have something that enables you to go from your brain to shift code to customers, embrace it and put it into hyper mode and see what you can do even faster to generate better ideas. Yeah, I love that you said that. I think that's what that's all that matters, man. Like all the other stuff around like OKR planning and like who cares? Just like get your idea from the customer, you know? Exactly. And um just to wrap up, I would love to hear some of uh some stories about your
43:26

The bike accident story that almost derailed Canva's launch

early days. I want to hear this one story I kind of know about already, but I want the audience to hear it, which is um I think you were launching Canvas first version back in 2013, right? And like it wasn't making a million designs back then. It was still a small company and maybe talk about like talk about what happened that night before. Yeah. So, we hadn't launched yet. We had been working on Canva for about a year. We'd gone through all those product development processes of brainstorming mockups, doing user testing sessions, focus groups. Uh we eventually got prototypes in front of people. Uh and then for the last six months, we were very much head down coding the vision that we knew we wanted to push out. Um I had I was probably doing less designing in the last three months of launch and a lot more coding because we knew what we wanted. We just needed to get it out the door. Uh, and we're working late the weekend before we were launching. It was a Saturday night. Uh, I was on my way home. Uh, as bike. I commute everywhere. Yeah. And it was a busy Saturday night in the city. I was coming down a hill and this taxi just came out of nowhere and t-boned me. I flew over the hood, blacked out, woke up on the road. A bouncer from a nearby pub had come over and had stopped traffic. It was like four lanes of traffic that was steaming down on me and I was just like collapsed on the road spitting blood from my mouth. Ended up in the ambulance, had to go to hospital, got stitches, but we still needed code shipped for the launch on Tuesday. So on Sunday, I was back in the office not looking as good as I do now and back coding. And fortunately we managed to uh get all the code done by the Tuesday morning which is when we had a press embargo on and everything launched fine. But it was it's a memory I love thinking back on because the whole team rallied around me. They were the ones who turned up at the accident site, called the ambulance, got me to hospital. Um and of course I wanted to support my team. So I was in the next day making sure that I was working alongside them uh and getting the product out the door. That's incredible. Yeah. And that Tuesday is when we got our very first user. It was a slow, slow journey. We only got about 100 people that first night, but word about the product grew. We had,000 people by the end of the week. We had 5,000 next Fortnite, 25,000 people by the end of the month. And hockey sticks just grow really slowly in the early days. What do you think was the thing that getting a little bit floss walk over the last question like what do you think was the thing that really motivated you to actually get up and work Sunday? Do you think it was the team or like the grand vision or like all the above? Like I'm trying to figure out like what motivates like people to like do really hard things, right? And like not give up. I mean I think primarily it was the vision. We want to empower the world to design and we truly believe that putting design in the hands of people who haven't been able to access it before lets them have an amazing positive impact on their lives and their world. Um so getting that product out has always been our inspiration. But as you start heading towards that vision, you do build up an amazing team of highly skilled designers and engineers working with you to get this thing out the door and realize that vision and create value for customers. So you build up this team who rely on you, who you need to inspire, and who also inspire you. So it kind of operates on two levels. It's like we need to get this vision shipped, but we also need to do it for an amazing team and be there in the trenches. next to them and make sure that we're doing it together and not let them down. Yeah. Uh so yeah, it operates on many different levels. Yeah. I think that's the core of everything. Like you just got to find people they enjoy working with to go after something hard and everything else will hopefully happen around it. Yeah. You can't have a great product without a great team. Got it. So uh so where can people find Canva AI? Well, I guess it's pretty obvious. Just go to campa. com, right? Just go to the homepage. Camber AI is right there. It's kind of our omni box where you can search, you can find templates, uh, and you can start generating images and designs, uh, all from the one interface. Super exciting. Thank Thanks so much, Cameron. This has been an awesome conversation. Yeah, really great to chat to you, Peter. Thank you.

Ещё от Peter Yang

Ctrl+V

Экстракт Знаний в Telegram

Транскрипты, идеи, методички — всё самое полезное из лучших YouTube-каналов.

Подписаться