Hi everyone. It is Tuesday, February 17th. Welcome to another Tech Live stream where we cover all the moves in the AI space. There's a lot of them. I'm Georgia. Now, there's a question that has consumed every software CEO that we've spoken to over the last few weeks. How do you survive when AI can do what your product does? Most of them don't have a great answer yet. Some are embedding AI features and hoping that's enough. Others are cutting costs and praying the storm passes. A few are in outright denial. But then there's Dylan Field who decided to partner with the company most likely to disrupt him. Figma today announcing, just announcing code to canvas, a new feature built in partnership with Anthropic that lets users take code generated in AI tools like Cloud Code and convert it into fully editable designs right inside of Figma. Now, if you've built a working interface by prompting an AI agent, you can now bring that output directly into Figma's canvas where your team can refine it, compare options side by side, and align on what the product should actually look and feel like. Now, essentially, Figma and Field, they're running toward AI. They're not running away from it. And it's not that they're immune to disruption. It's that they're betting they can be a part of the new workflow that this disruption creates. Dylan Field, CEO Figma, the guy making that big bet, joins me now. And I'll remind the audience too, please get in your questions and comments. Dylan, thank you for joining us.
Thanks for having me. Good to see you.
So maybe let's start. How did this partnership come together? Did you call Anthropic or did they call you?
We've been exploring all sorts of stuff with Anthropic historically because this is actually just an evolution on what we already have. We have a way already to go from Figma design to code through dev mode MCP and that lets you read Figma files, bring them to code. We thought it'd be really cool given that workflows are changing. We're seeing now folks do things like before it was linear. You'd go from brainstorm to design to code. Now we're seeing people hop around between different stages at different times. But you still need to get the big picture. You need to zoom out and actually consider different possibilities, be divergent in your thinking, and also be able to directly manipulate data and visuals. Like if I prompt that I want to go change something from a 12 pixel width to a 14 or 16 and I'm trying to resize icons via prompting, it's just the wrong tool for that job. And being able to do all this in a shared canvas where you can see the bird's eye view and bring your teammates in to productively collaborate and expand the ideas with you. We think that round trip and going all the way back and forth is extremely important.
Right? So you're kind of combining the tools and I will note there was this sort of post that got a lot of attention. And it showed how Anthropic was actually using Cloud Code and its marketing tools and it named Figma before this partnership sort of became official. So what does this partnership change?
Yeah. Well, now our users that are in Cloud Code and they're in Figma can now bring their work from Cloud Code into Figma and with that they can get all the benefits of the design canvas which is infinite and able to really be a place where you can explore all these possibilities because I think right now what happens is a lot of people that are working first in code, be it developers, designers or anyone that's an AI builder, they might be someone who is trying to make progress fast. And just because you're going fast doesn't mean you're going fast to the right place. And so instead of having that tunnel vision, how do you actually zoom out and really see the possibilities? Because now we're in this world where people are trying to really embrace taste. That was the discourse this last weekend. And for us at Figma, we've been talking about how design, point of view, and craft, these are the differentiators. And we've been talking about that for years. And I am so excited that everyone else is now seeing this. But that taste, that craft, that ability to then select from this infinite possibility space and then pull in the right thing to build it really requires side-by-side comparison and the ability to think through problems visually. And most people are visual thinkers. So being able to go from code to the design canvas and then back again and implement it and then ship to production. I think that's something that's really exciting and just a start too that roundtrip experience will only get better from here and with more partners.
So at the crux of sort of the selloff in software that we've been seeing in public markets is this idea: do you need another step? Do you need that round trip? And actually there's lots of comments and questions already in the YouTube live stream getting at this sort of existential question for Figma and many other software companies. And I guess someone put it this way. What are your views on AI replacing the whole design framework rendering Figma and other companies pointless?
Yeah, I think I obviously have a different point of view than the hypothesis there. So my point of view is that we're now in an era where you can make things very quickly. You can build things very quickly. So it's not enough just to go build them. You have to build the right thing. Good enough, it works is not enough. You really have to focus on
Right. I guess that's the distinction designers for everyone for all companies.
In a world where you're an engineer and you're able to build things very quickly and now you're considering what should I build? Well, welcome to the world of design. And if you're a company that is trying to quickly go build and innovate, you really need to have a point of view. This actually goes more back to marketing in some ways, but also it goes through your entire brand and product process too. You have to have a point of view that is different than the rest of the world that you'll hold on to. And I think the requirements have gone up, the stakes have gone up. If you're trying to build software, there will be exponentially more software in the world. This is already a trend that's been happening. We've already been on that exponential. It just happens to be going asymptotic now and vertical. So, right. You know, I'm actually quite excited. I think the future of software is more exciting than ever before.
Right. Dylan, you bring up a really good point. We've seen a lot of this over the last three years. There's been these sort of scares. We've seen really great image generation models. What is it about this moment and I'm talking in the last like six weeks or so where these AI coding and agent tools have been better. It feels like a big sort of step change. I don't know does it feel like that to you? What's different about this moment that is causing so much fear in the market thinking that AI agents are threat? You're obviously seeing that as an opportunity but what's different right now?
Yeah. Well, I guess I've kind of been in the world in my head of course model capabilities will continue to improve. Of course, we're on an exponential for a while now. And that's why I thought it's really important that we make sure that as models get better, Figma gets better. And also to me it's very clear that's why every company, every designer, every engineer, anyone building products that are software-based has to really think about do they get better as models get better. And about this moment I mean I think that we're continuously seeing exponential play out. And it'll be super interesting where it goes from here. You know that things will get better. You knew new capabilities will come online. You just don't know when exactly. But the world is moving from closed systems to open systems and that's a really foundational point to understand if you're in the software business.
What do you need to understand about it? Embracing that open model.
Well, I think that in the past in closed systems you could really model in your head and simulate all the potential players and sort of the different ways that technology would evolve. And now I think with an open system you can know what is possible and what you're going to be likely to see and you really have to think through what is the experience of the future and you have to have that point of view. That point of view is so critical now. And that's why I think that there's all the more need at a time when you can go build things so quickly to really be thinking about the bigger picture and to be going through the design process. And you know this is nothing new. In the past when we've had technology leaps, we've come to the same conclusions. This is not the first time that this has been explored. So if you look to history, I think it informs the present too.
And that's a lot of sort of the through line that I hear too is you need taste whether it is image design or writing or even coding. But so let me ask you there was this post that kind of went got a lot of attention on SaaS. It was this long post about where software is heading and it kind of laid out an argument that you need one of these three things: either proprietary data, regulatory lock-in, or you need to be embedded in a transaction. What does Figma have that can't be replicated? Maybe it's none of those things, but what is it?
I think we've got a number of modes, and I think that everyone should not only be looking for modes, but they should be thinking about what new modes can they build as well. But I also think you shouldn't be sacred about trying to lock your users into one workflow. And that's what the bet is today that we are talking about is that we're trying to make it so that wherever you start, whether it be a napkin sketch or a shower thought or design canvas or code, you should be able to go anywhere else that you need to go. And so you should be able to start anywhere, go everywhere. And I think it's really key that you embrace that philosophy, which is really the Unix philosophy, as you work in order to make it so that your customers are able to get the most benefit possible.
Right. It reminds me what Allegatsi told us a few weeks ago. He said there's no more lazy software investing. You need to look for the companies that are actually embracing these tools. Which is sort of the message we're getting from you. A few really great questions from Torque and Trends on YouTube. They ask, 'What is the biggest misconception people have about designing products with AI today?'
I think that the biggest misconception, by the way, I love this question. I would say it's that the first prompt is the final output. Whether it's iteration through design like we're talking about today, where you're able to go from code to design canvas, or it's even just image generation and working with media. I mean we bought this company Weevy, now we call it Figma Weave, and they've created a workflow-based system where you can use node-based editing in order to transform different model outputs and then actually create a pipeline that gets you to an amazing solution at the end of the pipeline. And I think of AI output almost more like clay that you then use to shape, and less like 'okay I've done a prompt and I got a one-shot and oh my god it's so good let's go.' And so I think people need to embrace this as a part of the process but not assume that a single prompt is the process itself in order to get somewhere great.
Right? The idea that you're starting from a point that has already taken so many iterations and so many drafting. And I think you're talking about integration. I know that Figma has been going down this road very much expanding into slides, sites, marketing assets. What happens though when the AI agent becomes the integration layer?
Well, there are many things that can mean. Some people talk about for example AI agents as what will use software in the future and I think you know there's verifiable tasks and there's non-verifiable tasks and anything in the world of design and creation is inherently non-verifiable. And so I think there's sort of different categories there but also there's the auditability. And when agents are working on your behalf, I think that if you're trusting agents today, you are very brave. You know, they're not flawless as it stands right now. And so my assumption is that at the very least, even under the current paradigm of thinking, which I'm not sure is quite correct, you will need to be able to audit agents and that'll have to be human readable. But the other thing that I think is really interesting is the world that people are not quite talking about yet, which is one where agents can work alongside you. I mean, in Figma, we have a multiplayer canvas. You can invite humans in and you can work alongside them and you can see what they're up to. You can see their cursors moving around the screen. Well, what if all software it was the case that you could actually work alongside agents, see what they're doing in real time and actually really grok that in a more intuitive way about what they're up to.
I mean I hear you on that, Dylan, and I feel like that's the version of AI agents we've been sold over the last few years, but what we're seeing this year is so much more powerful. We're talking about, you know, like OpenClaw where you have an agent on a totally separate device, a Mac Mini, and you just let it do its thing. I've seen this firsthand as well. I don't have the time or the patience to sit next to my agent while it's doing all of these things. I hear your argument as well on security. That's a huge concern, but it's only getting better, isn't it? Isn't there a world in which, you know, you're not sitting alongside your agent, but you're giving it more and more trust because that's making you more productive. So where does that leave, you know, an interface?
