Figma CEO Says AI Commoditizes Code, Shifts Software Edge to Design

Figma CEO称AI使代码商品化，软件竞争优势转向设计 随着人工智能降低生产门槛并使编程沦为通用商品，软件差异化已完全转向人类的品味、品牌认知与设计。 Dylan Field剖析了不断变化的技术格局，认为随着AI模型使软件工程走向商品化，设计与品牌认知已成为产品差异化的真正战场。他回顾了历时16个月、最终阻止Adobe以200亿美元进行收购的艰难反垄断审查历程，分享了Figma如何凭借强劲的独立增长实现反弹，并通过体面的离职补偿方案重塑公司文化。Field还驳斥了“SaaS末日论”，认为基础模型的简单进展无法取代精细且专业化的工作流工具。他最终坚称，尽管AI降低了生产准入门槛，但人类的付出、勇气与品味仍是唯一能够提升创造力上限的力量。 Dylan Field坚称，AI编程模型的飞速进展正在使软件工程商品化，从而使设计、品牌塑造和独特视角成为产品价值与差异化的主要来源。 (0:00) 在Anthropic推出Claude Design后，Figma实现了48%的营收同比增长，打破了网络上关于AI引发“SaaS末日”的普遍论调。 (12:09) Field批评反垄断监管机构使用滞后的市场定义阻止了Adobe 200亿美元的收购要约，并将其与亚马逊收购iRobot被否决后导致iRobot走向崩溃的情况进行了类比。 (3:16) 为防止质量下滑并维系信任，Figma实施了一项内部政策，强制要求任何使用AI起草的公司文件必须明确标注“由AI撰写”。 (24:45)

Figma CEO Dylan Field argues that advancing AI coding models are commoditizing programming, shifting competitive differentiation in software toward design, taste, and brand identity. While AI lowers the technical barrier for production, Field maintains that human judgment and craft remain necessary to break outside statistical averages and raise the ceiling of creative work. Commoditized Code and Value Differentiation (0:00) Field's central thesis is that 'design is the new code.' Because foundation models increasingly automate end-to-end coding tasks, programming alone can no longer serve as an enduring competitive moat for software products. App store figures reflect an exponential surge in software production even as consumer app usage has plateaued, illustrating that software supply is outpacing user attention. Under this dynamic, product success shifts toward higher-order human inputs: brand identity, strong points of view, and distinct emotional resonance. While AI platforms easily produce outputs centered on statistical training distributions, Field asserts that human creators differentiate products by demonstrating care, non-consensus thinking, and intentional craftsmanship. The Adobe Block and Post-Merger Realignment (3:16) Figma initially agreed to a $20 billion acquisition by Adobe, believing its scale would protect it from being dissolved into the legacy conglomerate. However, antitrust authorities across the United States, the European Union, and the United Kingdom's Competition and Markets Authority intervened, resulting in a grueling 16-month review before the deal was abandoned. Field argues that regulatory bodies made backward-looking assessments based on legacy market definitions, failing to recognize how fast-moving AI shifts expanded future competitive landscapes. Following the merger's collapse, Field called an all-hands meeting to address the internal fatigue caused by prolonged ambiguity. To reset organizational momentum, Figma introduced a voluntary exit package offering several months of severance with a six-month rehire cooldown. Only a small fraction of the roughly 1,100 to 1,200 employees accepted, allowing Figma to refocus as an independent enterprise pursuing a standalone path to going public. Foundation Models Versus Specialized Application Workflows (12:09) When Anthropic launched Claude Design following an earlier partnership with Figma, social media commentary claimed Figma was obsolete. Field dismissed the 'SaaS apocalypse' framing as an overreaction, noting that public discourse repeatedly mistakes raw foundation model capability for finished product solutions. Figma subsequently reported 48% year-over-year revenue growth in the following quarter, showing no measurable business loss from the release. Field draws a distinction between general-purpose generation and deep workflow software, pointing to coding tool Cursor as an example of how dedicated, user-obsessed applications can outmaneuver frontier labs. Professional software operates like a physical instrument requiring ergonomic precision, edge-case management, and nuanced interaction mechanics that cannot be replicated by wrapping foundation models in generic interfaces. Tooling for Dynamic Parameterization and Human Agency (30:41) Rather than relying entirely on one-shot generative prompts, Figma is developing systems such as Weave—a node-based environment that lets creators route outputs from various AI models into subsequent inputs, shaders, and procedural pipelines. Field characterizes this approach as crafting clay, enabling users to step up levels of abstraction and set behavioral rules rather than manually drawing static states. Field also distinguishes verifiable domains like mathematics and cybersecurity—where automated models scale exponentially through formal proofs—from non-verifiable creative fields that lack objective scoring functions. Addressing fears that generative technology diminishes human capability, Field frames AI adoption as a choice: individuals can either surrender their critical thinking to algorithmic feeds or use automated baselines to conquer the blank page while preserving deliberate curiosity and craft. While automated models compress the technical difficulty of building digital products, Field stresses that human taste and biological responsiveness to nuance will remain the primary drivers of meaningful innovation. As computational tools begin resolving historic mysteries across formal domains, maintaining active creative agency and intentional points of view will define the boundary between genuine invention and derivative mimicry.

Figma 联合创始人兼首席执行官 Dylan Field 认为，随着人工智能编码模型使软件工程逐渐大宗商品化，设计、品牌辨识度与独特观点已成为产品竞争的核心壁垒；AI 降低了创作门槛，但决定作品上限的依然是人类的审美与主动意志。 代码大宗商品化与设计的全新壁垒 (0:00) Dylan Field 指出，现代软件工程的发展正经历根本性转变。随着大模型编程能力的飞速提升，单纯的代码编写能力正在被大宗商品化，无法再作为科技企业的核心护城河。相比之下，软件价值的差异化正转向更高维度的创意层面，包括品牌定位、大胆的视角选择以及与用户产生的情感共鸣。他由此提出核心观点：“设计是新的代码”。 应用商店的现状印证了这一转变：软件供给呈现指数级增长，但普通消费者的实际应用消费需求却基本持平。Dylan Field 强调，代码已不再是阻碍产品诞生的绝对技术瓶颈。面对充斥着同质化软件的市场，能够在噪音中脱颖而出并持续留存用户的关键，在于设计团队是否敢于打破共识、投入深层精力和审美意图，从而为产品注入不可替代的人文关怀与实际问题解决能力。 反垄断审查与独立发展的韧性验证 (3:16) 回顾 Adobe 拟以 200 亿美元收购 Figma 却最终因监管告吹的历程，Dylan Field 表示，美国、欧盟以及英国竞争与市场管理局（CMA）历时 16 个月的反垄断审查带来了巨大的不确定性。他批评监管机构往往立足于狭隘的历史市场定义向后看，而非以技术专家的前瞻视角审视行业演变，并以亚马逊收购 iRobot 案最终导致后者破产为例，警告漫长的审查会严重损害企业生存能力。在交易终止后，Figma 提供了无条件且包含数月薪资的离职方案，最终仅极少数员工领取遣散费离职，组织迅速调整并重回独立发展轨道。 面对社交媒体上宣称 Anthropic 推出 Claude Design 将引发“SaaS 末日”并摧毁 Figma 的舆论，Dylan Field 明确反驳称，外界极易将基础模型的单点技术突破与成熟的工作流产品混为一谈。通用大模型生成的视觉内容虽然广泛，但专业设计属于对人体工学和微小精度要求极高的领域。在 Claude Design 发布后的季度中，Figma 实现了 48% 的营收同比逆势增长。Dylan Field 认为，正如代码编辑工具 Cursor 在基础模型实验室夹击下依然崛起一样，深度专注垂直工作流和专业用户体验的软件企业仍具有坚固的竞争壁垒。 降低底线与拉升上限的创作辩证 (18:52) 针对生成式 AI 对人类创造力构成的冲击，Dylan Field 修正了自己早期的判断。他原先认为 AI 会提升创意的上限，但如今观察到 AI 真正改变的是底线——它极大降低了代码、设计、视频等数字内容的生产门槛，让任何人都能跨过技术执行的障碍快速建立初稿。然而，由于模型本质上倾向于输出统计分布中心上的平庸结果，只有当人类创作者选择注入深度构思、探索未曾出现的非共识思考时，作品的上限才会被真正抬高。 为了应对生成式工具泛滥导致的真实感稀释与低质内容泛滥，Figma 建立了内部规范，要求所有使用 AI 协助起草的文档必须在顶部标注“Written with AI”，以此维系团队互信。在产品设计维度上，Figma 倡导在画布中将代码视为一种原生材质，并研发了基于节点的系统 Weave，允许创作者将不同模型的输出链接起来进行程序化雕琢。Dylan Field 认为，AI 擅长打破“空白页面”的初期停滞，但创作者必须掌握精细调节的控制权，而非被动接受黑盒生成的既定结果。 非验证领域的审美与人类主体性 (32:27) Dylan Field 区分了形式化技术与创意领域的演进逻辑。在数学、网络安全等具备客观真理标准的可验证领域，模型能力正通过强化学习和形式证明实现智力密度的指数级跃升；而在设计、艺术等非可验证领域，模型表现往往在最初的新鲜感退去后迅速遭遇审美瓶颈。人类生物系统具有对新奇感的本能渴望，并且能够通过单次社交反馈即时调整损失函数，这种对情境的敏锐感知和解决具体用户问题的能力，在结构上不同于需要海量试错循环的机器学习架构。 面对自动化浪潮，Dylan Field 认为人类始终面临根本性的存在主义选择：是走向被动消费、将思考完全外包给算法，还是主动选择付出努力去攀登创造的高峰。他援引 Holly Herndon 将自身人声训练为开源模型等前沿艺术探索，主张将技术转化为扩展人类可能性的工具，坚决反对在智能时代放弃批判性思考与个人好奇心。 生成式技术虽大幅压缩了数字产品的生产壁垒并迅速攻克形式化问题，但在缺乏客观真理标准的创意与设计领域，机器输出仍受限于统计均值；如何在技术奇点消解未解之谜的伤感与生产力跃升之间保持平衡，并借助精细化工作流工具坚守人类自身的主体性，依然是摆在创作者和软件企业面前的长期命题。

[12:09] Dylan Field: I think it's easy to mistake model progress for product progress, right? [22:44] If you're not doing something that's never been done before, you're not creating something. You're mimicking something. You're copying something. [0:00] Dylan Field: My point of view, the thing you'd expect me to say too is that design is the new code. [30:41] Dylan Field: The whole idea is like it's not bad to generate with AI, but you need a way to craft that clay.

[0:00] Dylan Field, welcome to the All-In [0:02] Interview. [0:02] >> Thank you. Thanks for having me. [0:04] >> We are here in beautiful Napa, Yanville, [0:06] California. [0:07] >> Yeah, you're three drinks in. [0:08] >> We're a couple drinks in and we decided [0:11] to go on camera. [0:13] >> Yeah, why not? [0:13] >> Why not? Well, what are you doing here? [0:20] >> As your wealth grows, have you noticed [0:22] how much more complex things get? [0:24] Between investments, tax strategy, and [0:25] estate planning, managing it all is a [0:27] full-time job you didn't ask for. [0:29] Creative Planning can help. Their [0:31] coordinated team of investment managers, [0:33] CPAs, and affiliated estate attorneys [0:35] can do much of the heavy lifting so you [0:37] don't have to. Creative Planning, where [0:39] wealth works together. Go to [0:41] creativeplanning.com/allin. [0:46] >> I was uh for this event that you're [0:49] going to be at tomorrow as well. [0:50] >> I'm at an event tomorrow. uh same event [0:52] and we were basically talking about [0:55] design, product development. Uh I stuck [0:57] in a little bit of security. Thought [0:58] that'd be nice. Uh but overall a lot [1:02] about just like how product development [1:04] is evolving. And my point of view, the [1:07] thing you'd expect me to say too is that [1:09] design is the new code. Uh and what's [1:13] happening right now is as coding models [1:15] get better, I'm not saying that you [1:17] don't need engineers. We definitely need [1:20] engineers at Figma. We love our [1:21] engineers. But also the nature of [1:24] programming is changing and as it [1:26] changes the value is rising in terms of [1:29] differentiation to the top, [1:31] >> right? [1:31] >> That's design, that's brand, it's [1:33] marketing point of view. It's being [1:35] bold. Uh and it's like the emotional [1:37] reaction you're able to elicit. [1:39] >> Uh and I think that a lot of people are [1:42] missing that. We also talked a lot about [1:44] like AI and just kind of like [1:46] >> Okay, so you did a thing on stage today. [1:48] You talk with Nikesh. [1:49] >> Yeah. Sorry, I'm just recapping an [1:49] interview here. [1:50] >> Yeah. No, it's great. Like, it's good. [1:52] Like, we can fast forward, but I want to [1:53] back up. You're the CEO of Figma, [1:55] publicly traded. Maybe you can just give [1:58] the quick elevator pitch on Figma. And [2:00] when did you start the business? [2:02] >> Yeah, we started Figma back in August [2:04] 2012. And the quick pitch is this is [2:07] where we want to make it so that the [2:09] best best teams in the world can design [2:12] and create and build the world's best [2:14] software. Uh we're also used for all [2:16] sorts of stuff. So, marketing, uh, [2:19] advertising, [2:20] slides, whiteboarding, diagramming, [2:24] uh, you know, you name it. Like, if it's [2:26] visual at all, someone's probably doing [2:28] in Figma, even if it's not meant for [2:30] that. [2:30] >> People call it a design software. Is [2:33] that a fair characterization or does it [2:34] not capture the full spectrum of what [2:37] the service is? [2:38] >> No, I I think that design is really our [2:41] center and our core. Uh, and like when [2:43] I'm in a product review and talking with [2:45] the team, we talk about designers, we [2:47] talk about what they want. Uh, design is [2:50] also a term that's like so interesting [2:52] because there's like a 100 definitions [2:54] of design, [2:55] >> right? [2:56] >> I I think about a lot of a lot about the [2:59] definition of like it's uh maybe [3:01] creativity meets problem solving. Um, [3:03] but there's like so many others you can [3:05] find. And then also design can refer to [3:08] everything from interior to industrial [3:10] to product design to brand design and so [3:12] on. And so people think about a lot of [3:15] things when they think about design. [3:16] >> Just to get caught up on the history of [3:17] the business. So you started in 2012. [3:19] You raised some venture money. The [3:20] business takes off. You're selling [3:22] enterprise software. You're doing really [3:25] well and then Adobe offers to buy the [3:27] business for $20 billion. [3:30] >> Uh yeah, correct. Quite the fast [3:32] forward. [3:32] >> What year was that? [3:33] >> Uh it was a few years back now. I'm [3:35] skipping through literally building the [3:37] business like [3:38] >> yeah you're like uh you know 10 11 years [3:40] in but the um no I mean we uh have [3:43] always respected Adobe still do and the [3:46] generational incredible company that [3:48] like started everything from printing to [3:51] making Photoshop what it is and like the [3:54] entire graphic design revolution. [3:56] >> By the way I was on Adobe Photoshop [3:58] since 1990. [4:01] I did graphic design layers actually you [4:03] and I haven't talked pre layers. Yes. [4:05] So, I did graphic design pre- layers as [4:08] a kid on computers and I did it as like [4:11] a service for like people in the [4:12] neighborhood and small businesses and [4:13] stuff. [4:14] >> So, cool. How did you get access to it? [4:16] >> I was like I at my school and then I had [4:18] well I was on a Macintosh. I got the [4:19] first Mac original 1984 [4:22] and um I was always fascinated with [4:25] graphic design. It was like something I [4:27] love doing as a literally a hobby. you [4:30] and I actually never talked about this [4:31] and then I used like Bryce 3D and all [4:33] these early 3D tools and then when Kai's [4:34] power tools came out [4:35] >> Kai's power tools is really cool. [4:36] >> It reinvented Adobe Photoshop and for me [4:39] it was like this incredible like [4:41] expansion of creative potential and what [4:43] you could do with Photoshop. So I've [4:45] been a big Photoshop fan since the [4:47] earliest days and a big user as a [4:50] personal hobby and like I made money on [4:52] the side doing graphic design and stuff [4:54] as a kid. [4:55] >> Yeah. [4:55] >> Yeah. I mean, Photoshop of course is [4:57] like a product that is very mature and [5:00] maybe beyond mature too in that it has [5:03] so many different parts of it that have [5:05] been added over the years. Uh which I [5:07] think is the nature of a lot of [5:08] software. Uh it gets more bloated over [5:11] time if you're not careful. But there's [5:13] just so much that's possible with it and [5:16] uh you know Adobe as a company has been [5:19] very inventive. So I've always had [5:21] respect since I was a kid too. [5:22] >> Totally. And it felt like it was a time [5:24] when, you know, we could join and not be [5:26] totally absorbed. Uh if we were smaller, [5:30] I felt like maybe we would just kind of [5:31] dissipate. And yeah, we we signed the [5:35] agreement. We said yes. And then uh [5:37] various world governments decided that [5:39] they didn't like it, [5:41] >> right? [5:42] >> Like antitrust got involved and they [5:44] blocked the deal. [5:45] >> Yeah. Yeah, I mean it's it antitrust is [5:47] kind of like a broad way to refer to it, [5:49] but you have the United States, European [5:52] Union, CMA and you know it's very [5:55] different vibe and a couple years [5:56] >> ago Adobe would be taking out a big [5:59] competitor. [6:00] >> They said a lot of things but the I [6:03] think most important part for us was [6:05] that we just kept putting our foot in [6:06] the gas. I mean, yeah, I thought it [6:08] would be like sort of we would just go [6:11] right through and as soon as a month or [6:14] two in, it was clear that was not the [6:15] case. [6:16] >> It kind of became, okay, like we got to [6:17] like really power through as an [6:19] organization here [6:20] >> and ship and build like the next stuff [6:23] >> because if we don't power through and [6:24] keep our foot on the gas, you know, [6:26] what's on the other side? We don't know [6:28] what's on the other side. [6:29] >> Uh, we got to make sure everything's [6:30] fine no matter what. [6:31] >> So, this is 2023. [6:34] How many employees did you have? Uh [6:37] >> I want to say it was about 1100, 1200. I [6:40] have to double check. [6:41] >> Okay. So when I heard this story, I was [6:43] fascinated by and I think I saw you in [6:45] 2024, March of 24. I don't know if you [6:47] remember I mentioned this to you, but [6:49] for me the story of telling,00 employees [6:53] of a startup, we have one of the most [6:56] incredible exits in history, $20 billion [6:59] exit of this company and it's not going [7:02] to happen. What was the all hands like [7:04] when you get in front of everyone and [7:06] you tell everyone, "Guys, this isn't [7:08] going to happen." And how do you get [7:10] everyone to realign on building a [7:13] business and continuing forward when it [7:16] feels like the floor for some people may [7:18] have felt like the floor fell out from [7:19] under them? Well, I think you have to [7:21] remember the length of the regulatory [7:24] review, which definitely needs to be [7:27] fixed somehow because it's, you know, if [7:29] you're in a situation where it's [7:32] actually anti-competitive, which I [7:34] don't, I still don't believe the Adobe [7:35] thing was for what it's worth. A lot of [7:37] people, you know, on the internet [7:38] probably would say it is, uh, like, or [7:41] at least they did back then. But I [7:43] thought their arguments were poor and [7:45] they were fighting past battles, not [7:48] looking at the future and what this all [7:50] would become. Uh, but the um, and I say [7:53] that not because I'm not like I'm stoked [7:55] to be independent, too. I'm stoked to be [7:58] running Figma. That was the plan for [7:59] Adobe was just run Figma. Uh, either [8:02] way, uh, running Figma, I'm happy. [8:05] Independently, we do move fast. I'm gl [8:07] I'm grateful for that. But uh you go [8:09] back to your question, it's like we had [8:11] 16 months of regulatory review. 16 [8:13] months of regulatory review for any [8:15] company is punishing and uh in an actual [8:19] anti-competitive situation that could be [8:22] like very destructive for one of the [8:24] parties, [8:24] >> right? [8:25] >> Or both. Um and I think Roomba is a [8:27] great example actually. [8:28] >> Roomba uh was going to be acquired. [8:31] >> Small deal. Uh yeah, it wasn't like a [8:33] crazy deal, but they wanted to pun [8:34] punish Amazon and so [8:37] >> they did not let it go through and [8:39] because of that Roomba went bankrupt and [8:42] the entire market at the time basically [8:43] went to China. Uh so from the United [8:46] States standpoint, it made no sense, [8:47] >> right? [8:48] >> Anyway, I go on a tangent, but I do [8:49] think it's an important point. And [8:50] there's some policy stuff here that's [8:52] like when you go for the ideals [8:54] idealism, it doesn't always practically [8:56] like help America. [8:57] >> But it's also an important point which [8:59] is almost all markets now are more [9:02] expansive than they have been [9:04] historically. [9:05] And nearly anything that looks like, oh, [9:08] it's the market. That's not actually the [9:10] market. The market is [9:12] >> AI. turns I mean the market is this [9:15] massive market and this deal is 1% and [9:19] 2% of the market [9:20] >> like you define the market looking back [9:21] at the history which is like the only [9:24] logical way you do it as a regulator or [9:26] do you do it as a technologist looking [9:28] forward uh and that could be very [9:31] different in terms of how you think [9:32] about markets sizes but um in any case [9:35] the all hands uh it was actually the [9:39] first day of a winter break uh which is [9:42] pretty funny. So, we had told everybody, [9:44] okay, uh you know, enjoy the holidays. [9:49] We actually have way less traffic during [9:51] the holidays because we're a work [9:52] product. And that was a Friday. And [9:56] then, um realized sort of over the [9:58] weekend, okay, this is going to come to [10:00] a conclusion. and on Monday morning uh [10:04] sent a message to everybody saying hey [10:06] all hands at noon and uh dial in if you [10:09] like and here's what's going on and [10:10] here's like the full sort of uh details [10:13] and so first of all so much of the [10:16] company showed up that was awesome just [10:18] to see the turnout you know first day of [10:20] break really and secondly I'd say just [10:24] like the mood was relief because it [10:27] wasn't about relief of not Adobe it was [10:29] relief of not having ambiguity to have [10:31] ambiguity booty that long is really [10:33] intense and people have to kind of have [10:35] two worlds in their head. They're not [10:36] exactly sure what the future holds. So [10:39] that was the immediate mood. It was just [10:41] kind of like lighter. And then from [10:43] there, uh yeah, I think that the [10:45] question you asked is right, how do you [10:47] make sure everyone's all in? [10:49] And [10:51] the way that we did it was we said, [10:52] "Look, I I just kind of put a video out [10:55] into the company Slack and emailed [10:59] everybody and said, "Hey, if you're [11:03] tired after working really hard, if [11:05] you're someone who joined during this [11:07] period thinking you're going to go work [11:08] at Adobe and actually the startup thing [11:11] is not for you, like please know, we're [11:13] going to be working very hard, hustling [11:15] hard, and no shame like we're still on [11:19] good terms. If you want to leave, here [11:22] is a few months pay. Uh and you're [11:26] welcome to just let us know. We'll make [11:28] it super easy. And from there, uh you [11:32] know, 6 months before you can reapply. [11:34] That's the only condition. Uh after 6 [11:37] months, you can reapply and we'll uh [11:39] rein you. And it was a pretty small [11:43] percentage of people that took us up on [11:45] that, but overall it was healthy. And [11:48] some people actually ended up doing [11:50] total career pivots like in wild ways. [11:54] Uh went on different life paths. Like I [11:57] I think it was healthy for everyone. You [11:59] continue to build the business. It [12:01] grows. You go public [12:04] and I just want to fast forward to [12:06] earlier this year. You guys announced a [12:09] partnership with Anthropic Claude [12:11] Design. [12:12] >> We not quite. We we had been working [12:14] with Enthropic as a partner and [12:16] Enthropic uh separately announced cloud [12:18] design. Yes. [12:19] >> Oh, sorry. Yeah, that's right. But you [12:21] had a connector with them that you [12:22] announced earlier in the year. [12:23] >> Yeah, we have an MCP. We are partnered [12:25] with many companies. But yeah, [12:26] >> then they announced cloud design. [12:29] >> Was that surprising to you when they did [12:31] that? And how worried were you when they [12:33] came out with claude design relative to [12:36] what you guys were offering? Cuz the [12:38] market looked at it and said, "Oh my [12:39] god, Figma's dead. Claude is going to [12:42] crush Figma." I mean, the nature of [12:44] online discourse right now is uh we're [12:47] so back it's so over and it's just a a [12:50] cycle, right? I mean, how many times [12:51] have you been all over [12:53] >> uh in the last like month? [12:55] >> Yes. [12:56] >> Probably, you know, a bunch of times I [12:58] don't even know about. [12:58] >> Yes. [12:59] >> And [13:00] >> everyone [13:01] >> I think I'm was kind of used to that at [13:02] that point [13:03] >> cuz you know you when when you kind of [13:06] die uh so many times and then you're [13:09] actually like not dead, [13:10] >> right? you kind of realized at some [13:12] point maybe maybe this is like not the [13:14] truth. Social media is actually like not [13:16] reality. But um yeah, I I uh was overall [13:22] surprised at the way that kind of [13:25] communication happened. Uh I think that [13:29] you know look everybody [13:32] wants to do good usually. I mean there's [13:36] some exceptions but for the most part [13:38] that's that's what I assume for people. [13:40] They want to win. They want to do good. [13:42] And I think that organizations scaling [13:44] really fast. Anthropic like are there [13:47] any examples of organizations scaling [13:49] that fast ever? [13:49] >> Never [13:50] >> can get very confused and um mixed up [13:54] internally. And due to that weird stuff [13:58] can happen and I think that it's just [14:01] the case that yeah, we're on the other [14:03] side of kind of like a weird bout of uh [14:06] communication there. [14:08] >> Did it affect your business? Oh, no. Uh, [14:10] like not that not any degree that was [14:13] like, [14:15] you know, significant [14:16] >> cuz I read your quarterly report after [14:18] that announcement. You guys grew 48% [14:20] year-over-year and the business [14:22] basically accelerated its growth. [14:25] >> So, I'm looking at that and I'm, you [14:27] know, this was when everyone had the SAS [14:29] apocalypse [14:30] >> thesis in general and I was a big [14:32] subscriber to SAS apocalypse, but I do [14:34] think that there's a bunch of ways to [14:36] distinguish SAS companies. One of which [14:38] I think is founderled and I think that [14:40] there's user experience that matters a [14:42] lot embedded user experience and the [14:44] ability to kind of build from there [14:47] especially if you know how to use AI to [14:49] accelerate and reinvent your business as [14:51] a founder particularly but your business [14:53] accelerated. [14:54] >> Yep. Um, no, it's been continuing to do [14:58] well. And I think that [15:02] look, the hard part about model [15:05] companies and product is like what is [15:07] the product? Is it the model? Is it the [15:10] UI on top? Is it some hybrid of both? If [15:14] it's a hybrid, how is it weighted? And [15:16] uh, in any case, [15:18] >> you're talking about like an anthropic, [15:20] an open AI. [15:21] >> Yeah. Yeah, I mean Gemini like whether [15:22] it's a harness for coding or it's a [15:25] product like claw design which you [15:26] mentioned. [15:27] >> I think it's easy to mistake model [15:31] progress for product progress, [15:32] >> right? [15:33] >> The reality is a tool like Figma, you [15:35] know, has taken a very long time to [15:36] build uh and a very long time to make [15:39] sure that we get the workflows right for [15:40] our users. And we still have work to do. [15:43] Our users will tell you that. Uh and I [15:45] and we're always trying to make it [15:46] better. So I think that like [15:50] it's it's sort of um not surprising that [15:54] it wasn't an instant hit or something [15:57] when there's actually quite a lot of [15:58] work on the just core user journey [16:01] workflow side that you have to do to [16:02] meet the needs of professionals. Now [16:05] there uh they've made cloud design into [16:07] a product that is is broad. So is trying [16:10] to do a lot of things like you'll see [16:11] like posters and stuff as well like [16:13] graphic design scenarios I believe when [16:16] you go to their homepage and and like [16:17] maybe they'll do well in some areas and [16:19] not other areas. Like [16:20] >> I'm not bullish or bearish on the [16:22] product. I think it's just like hard to [16:24] make good design tools for [16:25] professionals. [16:25] >> Yeah. There's a lot of things to design [16:28] >> and a lot of categories of design. [16:30] >> The feel of it is like an instrument. [16:32] You have to get the feel right. You have [16:33] to get the mechanics right. [16:34] >> Yeah. And we've stressed every detail of [16:36] that and have had a lot of feedback over [16:38] the years about what people like and [16:40] don't like. It depends on where they're [16:41] coming from. It depends on uh their task [16:44] and sort of like what they're trying to [16:46] overall do. And there's so many [16:49] scenarios. But overall uh you know I [16:53] think that um there are more modes for [16:57] software than are implied by the SAS [17:00] apocalypse. And it's not like software [17:02] as a category is binary, good or bad. Uh [17:07] there are plenty of other ways that you [17:10] can um make sure that you're able to [17:12] substain yourself long term and have a [17:14] growing business long term as a software [17:17] company or app layer company. Uh and [17:20] part of it is yes, if you're as [17:22] specialized as possible on a target [17:25] audience and you know your audience, [17:26] you're building for them, that is [17:28] something that matters, [17:29] >> right? Uh, and I think a great example [17:32] of this is Cursor. Um, Cursor like by [17:36] all means [17:38] should not have grown the way they did. [17:41] You've got all these frontier labs that [17:43] are chasing them with their own models [17:44] before Cursor had a model of any kind. [17:47] But yet, Curser really obsessed over the [17:50] user. They saw around corners. They [17:52] strategized. They made sure they're [17:53] doing the right thing for the customer. [17:55] And they've had an incredible success. [17:57] >> Yeah. And I I think it's amazing and I [18:00] think it's an example that will be [18:01] repeated uh across many different types [18:04] of companies. [18:05] >> SAS is not dead. [18:07] >> Well, I I think that it's not true that [18:09] all SAS companies are fine. Yeah. [18:11] >> And it's also not true that all SAS [18:12] companies are dead. [18:13] >> Fair enough. [18:21] >> Protecting nature is often labeled [18:22] anti-growth. The Conservation Fund [18:24] disagrees. They believe protecting land, [18:27] wildlife, and outdoor access goes [18:29] handinhand with economic opportunity. [18:31] For 40 years, they've moved with market [18:33] speed to protect at risk land while [18:35] supporting local communities. With 9 [18:37] million acres protected across all 50 [18:39] states, the Conservation Fund is proving [18:42] that they deliver lasting outcomes that [18:44] benefit communities and nature. Learn [18:46] more at conservationfund.org. [18:50] I'm going [18:52] >> in a world of AI, [18:54] agentic AI, I have this capacity to [18:59] prompt [19:01] as a human and say, I want to make [19:04] something that looks like, feels like, [19:06] and does X. [19:09] And this magical technology takes off [19:12] and does it. And I end up with a [19:14] product. [19:16] Sometimes the product is great. [19:18] Sometimes the product sucks, but it's [19:21] improving the ability for these tools to [19:23] do that. Historically, I would have [19:26] needed a product manager, a designer, a [19:29] bunch of engineers to do the actual [19:31] building of the product, whether it's a [19:33] software product or a hardware product. [19:36] Increasingly though, it seems like AI is [19:39] doing more and more of those tools. Can [19:42] you just give me your view on the future [19:44] of human creativity in this era? This [19:48] future era that we're walting into, the [19:49] singularity we're falling into. Where [19:52] does human creativity play a role? And [19:54] where do the gaps still need to be [19:57] filled for taking human creativity to [20:00] the realization moment? It's the case, I [20:04] think, that uh [20:07] we're lowering the floor with AI, making [20:10] it so that all of the production, [20:14] whether it's code or it's uh motion, [20:17] video, design, whatever you name it, uh [20:20] if it's digital, you're lowering the [20:22] floor to make it happen. Um [20:26] everyone can be part of that and it's [20:28] more democratized. [20:31] Ironically, and I I used to believe uh [20:33] even a few years ago, I said it our [20:35] annual user conference config. I said [20:38] the ceiling will be raised by AI. You'll [20:41] be able to do greater things than me [20:43] before. That was wrong. Uh the ceiling [20:46] has stayed the same. AI is not raising [20:49] the ceiling. Humans are raising the [20:50] ceiling though. uh the creatives, the [20:53] designers out there, when they choose to [20:56] focus, invest their time, and to be bold [20:59] and try to do something that no one else [21:01] has ever done before, that's when we see [21:03] the ceiling raised. And at that point, [21:06] uh it's a question of do you care about [21:10] being in the center distribution? Do you [21:13] care about like a mean or average [21:15] result? Uh is that sufficient for your [21:18] task? or do you need something that [21:20] shows that you care that you took the [21:22] time to actually like go make this [21:25] thing? Do you need to go and be beyond [21:29] uh what is the minimum bar? And in the [21:33] world of software design and product I [21:36] mean we're seeing exponentially more [21:38] software than ever before. It's exactly [21:40] what you'd predict. The exponential [21:42] continues. Uh you look at app store uh [21:45] numbers. the app store numbers are [21:47] exactly that same exponential and we're [21:50] seeing that the usage and utilization of [21:53] apps flat. So people are using just the [21:56] amount same amount of apps but the [21:58] amount of apps that you can get it's [22:01] like bigger than ever and growing. And I [22:04] think it just speaks to okay there's a [22:07] need to differentiate yourself if you're [22:10] building any kind of software. Well how [22:12] do you do that? It's not code because [22:14] anyone can create code now. That's not [22:16] the hard part anymore. So my view is [22:19] design is the new code. It's the layer [22:21] above. You have to make it so that your [22:23] design is differentiated and you have to [22:25] have a point of view and a good brand in [22:27] marketing. And the point of view I think [22:29] is the part that is probably missed the [22:31] most. [22:31] >> That's the human part. [22:32] >> It's not just the human part. It's also [22:34] the part that humans are uncomfortable [22:36] having. Uh, I think that folks are [22:41] scared to say something that's not [22:43] consensus. [22:44] >> It's so true. I mean, this is what the [22:45] greatest artists, the greatest business [22:47] leaders, the greatest politicians have [22:50] always demonstrated is the willingness [22:52] to to be brave to say and do the things [22:54] that no one else would say or do. And [22:57] that's what changes the world. [22:59] >> And I would also add comedians to the [23:00] mix, too. [23:01] >> And comedians, [23:02] >> comedians are underappreciated. I had [23:03] Tony Hinchcliffe fight our thing last [23:04] year because he's like one of the few [23:06] guys that is willing to say and do [23:07] things nowaday in the cancel culture era [23:09] >> but comedians often like speak truth in [23:11] a way that people otherwise couldn't [23:12] hear [23:13] >> willing to say things [23:14] >> and uh it's not about political [23:15] correctness. It's about do you have a [23:18] message that can rise above the noise [23:21] and do so in a sustained way? Because [23:22] look at social media right now. Like [23:25] every unique point of view, everything [23:27] interesting, it gets diffused into like [23:28] the ether of social media instantly. And [23:31] so unless you've got uh something that [23:34] allows you to kind of keep generating [23:36] content and a message that is really [23:42] uh evergreen because it's so outside of [23:45] the over 10 window, you're not going to [23:47] be able to actually break through and [23:50] have any attention. And the attention [23:52] economy is ruthless. [23:53] >> Is this also what human creativity is? [23:57] If you're not doing something that's [24:00] never been done before, you're not [24:01] creating something. You're mimicking [24:02] something. You're copying something. [24:04] You're xeroxing something. But to do [24:07] something new is to create. And newness [24:11] requires doing things that have never [24:13] been done before [24:14] >> by definition. [24:15] >> By definition, [24:16] >> and uh therefore [24:18] you are operating outside of this [24:21] distribution of training. It's not even [24:24] about operating outside the mean of the [24:26] distribution which is where you start if [24:28] you're just prompting without any like [24:30] any kind of advanced scaffolding around [24:32] it. Uh but you're actually doing [24:35] something that is breaking through and [24:37] it's unique and I do think that will be [24:39] rewarded by the market. Everyone's going [24:41] to see it. I think it's even rewarded by [24:43] the attention market of organizations. [24:45] I'm already hearing from so many like uh [24:48] I got another deck made by AI today. [24:51] >> Totally. And I don't even know if the [24:53] person who made it, I like don't know if [24:54] they read it. Do they even know what's [24:56] in their deck? And at Figma, we actually [24:59] made it so that uh it was just a simple [25:01] thing, but we said, "Hey, if you are uh [25:04] using AI in your writing of any kind, [25:07] just put at the top written with AI." [25:12] Very simple, very easy to follow. Uh and [25:16] you know, people made jokes about [25:17] internally, but people do it. And it [25:19] makes it so that we can all trust that [25:21] when we see something, we know that it's [25:24] been cared for enough that it's like [25:25] worth your time to process. [25:28] And I uh I think that we're going to see [25:31] some bifurcation here around trust. Uh [25:34] and one way to get trust is to do [25:37] something with your hands. The other [25:39] point that's important is if you have a [25:41] vision or if you have a point of view in [25:43] your head of how something should be [25:45] designed or look aesthetically or from [25:47] an experiential standpoint, [25:50] you're I I think it's great to start [25:52] with building with AI. It gives you [25:55] something to get past like the blank [25:57] page, blank canvas problem. You can then [25:59] react to it and say, "Here are all the [26:01] things I don't like. I'm gonna I'm gonna [26:03] go fix these things. I'm going to change [26:04] them up." because it's really helpful to [26:07] have a back and forth. [26:10] But if you don't if you have your own [26:11] vision and you're trusting AI to [26:13] implement it, you're doing it through [26:15] the lens of AI and like AI has kind of [26:19] like its own distribution [26:21] >> the model fit by it. Yeah. [26:22] >> Yeah. And so you're not going to get [26:23] exactly what you want. Now if you're [26:25] able to do both and that's what we're [26:27] trying to enable in Figma make code a [26:29] material on the canvas make it so that [26:32] you can go start by building you can [26:34] build to think if you like uh but also [26:38] when you want to think and work with [26:39] your hands you can do that too [26:41] >> right [26:41] >> and you go back and forth rapidly [26:43] >> right I remember the early days of [26:45] Photoshop just having been a user avid [26:48] user from the early days [26:50] people freaked out when layers happened [26:52] and then they freaked out when all the [26:53] filters came out and when Kai's power [26:55] tools came out, everyone was like, [26:56] "What's the role of the photographer [26:59] anymore?" It's like you're destroying [27:01] the art of photography, the creativity [27:04] of photography. [27:06] But it unleashed a new era of creativity [27:08] because now these new artists emerged [27:11] that could use these filters and use [27:13] these layering tools and capabilities to [27:16] reimagine what's possible on a canvas. [27:19] And it was basically a new toolkit. [27:20] doesn't mean that the old creatives [27:23] mapped perfectly to the new creative [27:25] system and tools, but there was an [27:28] explosion in creativity and human [27:30] potential that arose from this and we [27:33] pushed boundaries in all sorts of ways. [27:35] Feels like we might be entering that era [27:38] again. [27:38] >> I'm a big believer in that and I think [27:41] that part of it is what I described [27:43] about market dynamics. In order to get [27:45] attention, order to get engagement, in [27:46] order to get people to care, you have to [27:48] do something new. uh you have to [27:50] actually be daring and try things that [27:52] are bold. The resistance to change is [27:54] real and there's a great thing you can [27:56] Google actually for your point. Look up [27:59] the mailing list for when layers was [28:01] introduced and it's like people going [28:05] this is the worst thing ever. Uh and you [28:08] can find it if you Google it's like on a [28:10] a mailing list that's publicly available [28:13] and you can see all their reactions. [28:14] Totally. Um, but obviously it became the [28:17] future and I think that uh there's also [28:21] still photographers in the world who are [28:23] doing incredible work. But the bigger [28:25] disruption, if we're going to be real, [28:26] it was the camera phone. Uh, that was [28:28] what made it so that people stopped [28:30] buying cameras that were dedicated to [28:33] the task of photography because the best [28:35] camera is the one that's on you and [28:38] started treating photography in a [28:39] different way with Instagram and their [28:41] camera phone. Uh so it transformed for a [28:44] lot of people and I can't tell you if [28:46] it's better or worse as a result. Uh but [28:49] there was change. There's one view on [28:51] human creativity [28:53] that human creativity is basically [28:56] connecting dots in the unconscious that [29:00] you've had, seen or experienced all of [29:02] these things in your life or in your set [29:04] of inputs, sensory inputs and from those [29:07] inputs you connect dots and then you [29:09] extrapolate a forward view of things [29:12] that hasn't perhaps been seen in the [29:14] world before. Do you agree with that [29:17] view on creativity and is it distinct [29:19] from actually what AI does? And again, I [29:21] want to go back to the uniqueness of the [29:23] human in an AI era. [29:25] >> Well, I I think that um [29:30] one thing that happens with humans is [29:34] they're always seeking novelty. Like our [29:37] reward function is different than like [29:39] an AI. Also, our loss function. Like if [29:41] I was to open this bottle of water and [29:44] splashed all over you and you got mad at [29:45] me, like you bet I'm not gonna do that [29:47] again. But an AI like you know with RL [29:50] you gota like really tell it a bunch of [29:52] times uh uh that no you don't do that. [29:56] Um [29:56] >> robots stop kicking me. [29:57] >> Yeah. Well and maybe that'll change with [29:59] architectures over time but like overall [30:02] the brain is different uh than the [30:04] current systems that we've made with AI. [30:07] And I I think that uh it's unknown. We [30:11] don't know much about neuroscience [30:13] uh overall. I'd love to get to the point [30:16] where like we all know more about [30:17] neuroscience, but something is [30:19] fundamentally different there when it [30:21] comes to creativity. And I also think [30:24] design is different than creativity. [30:25] It's not just creativity. It's also [30:27] problem solving like I mentioned [30:28] earlier. And the combination of the two [30:32] is I I just find it fascinating because [30:35] you have a user, you're trying to solve [30:37] their needs, you're trying to do so in a [30:39] way that stands out as creative and it [30:41] connects with them. And also like you [30:45] have to connect to wider systems, the [30:47] marketing brand, the culture, the [30:50] business problems uh and so on. and and [30:53] like holding all that in your head at [30:55] different levels of abstraction while [30:57] trying to like deliver a screen or a [30:59] flow uh that like from a searchbased [31:03] standpoint can be a comatorial explosion [31:06] is just pretty cool. It's pretty [31:07] fascinating. I'm actually personally [31:09] really excited about the idea of [31:10] reinventing [31:12] existing art form and media where AI can [31:16] enable more dynamic generation where the [31:18] creator steps up a level, sets a bunch [31:21] of parameters, but doesn't actually [31:23] necessarily render the final product, [31:25] but the final product sort of like in a [31:26] video game [31:27] >> is rendered on the fly with the user in [31:30] front of it. But the creator, the [31:32] creative team behind a video game, for [31:34] example, is building all of the object. [31:37] They're building the story line. They're [31:38] building the narrative flow. And the [31:40] realization of that is much more [31:42] dynamic, but I feel like it's going to [31:44] find its way. It's going to leech its [31:46] way into many more totally different [31:47] media. [31:47] >> And we we have products around that too. [31:49] Like in Figma Design, you can have now [31:51] generative tools and shaders where you [31:53] can do very much that or we have a [31:55] product called Weave. And weave is [31:58] basically a nodebased system. You can [32:01] connect model outputs into uh other [32:04] inputs and such. And the whole idea is [32:07] like it's not bad to generate with AI, [32:10] but you need a way to craft that clay. [32:12] You need a way to mold it. And uh with [32:15] weave, people do like the most amazing [32:18] video and photo work. Uh and overall [32:22] just like the the outcomes that are [32:24] achievable with it are incredible. Uh so [32:27] yeah I think that if you've got the [32:29] intention if you actually care to do the [32:33] work um you can always get to a better [32:36] result and people you know new model [32:39] drops there's like this AI psychosis [32:41] moment that people are in for a few days [32:43] and it's like I think also aided by algo [32:46] feeds [32:46] >> totally [32:47] >> the best results rise to the top [32:48] everyone's like this model is everything [32:50] >> and then a few days later it's it's okay [32:52] okay I know the [32:53] >> kind of like main tropes of the model. [32:56] Um, and you adapt fast. Like the novelty [33:01] is gone. It no longer is a thing that's [33:04] that's kind of like rewarding you or [33:06] firing off your dopamine system or [33:07] whatever. And instead, you're like, [33:09] "Okay, well, yeah, it always does that [33:11] thing. How do I move past it?" And that [33:14] seems to be the continued loop we're in. [33:16] Uh, which is sort of fascinating because [33:19] models keep getting so good when it [33:21] comes to these formal domains, math, [33:23] cyber security. uh like the IQ points [33:26] just stack and stack and stack and then [33:29] for creativity and for design it's very [33:33] incremental progress overall. Uh maybe [33:36] that will change at some point, but I [33:39] think the fact that these are [33:40] non-verifiable domains has something to [33:41] do with it. [33:42] >> And it's probably the fact that maybe [33:44] that is the human role in the AI [33:49] toolkit, which is conceive, create, [33:52] design, and let the AI do more of the [33:54] work, but it gives you the ability to do [33:56] much more work. Any art or media that [33:58] you're particularly excited about or [34:00] interested in that's kind of emerging [34:03] right now? You know, I I find it [34:05] fascinating uh when [34:08] people do sort of like new AI [34:11] experiments. For example, uh put this in [34:15] the media category. Um there was a [34:18] designer artist type that made a web [34:22] page the other day. Uh and it was like a [34:25] letter to Figma actually. So it caught [34:27] my attention because I was, you know, [34:29] getting the Figma hit on the algo feed. [34:32] And every time you moved away from it uh [34:35] and came back, it got slightly more [34:38] degraded and it was basically a comment [34:42] on attention. I thought that was really [34:44] cool. [34:45] >> Oh, cool. [34:45] >> Um [34:46] >> that was a piece of art. [34:47] >> I thought so. It was web page. [34:48] >> Yeah. [34:49] >> But like I thought that the technique [34:51] and the way that they went about it was [34:54] really neat and like not something that [34:56] many others have done. I've seen a few [34:58] other things like that, but overall I [35:00] thought it was a very cool way to do it. [35:02] Um, and it was also just a very [35:04] well-ritten letter. Uh, I like the [35:06] letter, too. But I Yeah, I see stuff [35:09] like this. Um, and you know, back in the [35:12] day, I remember um seeing people do like [35:16] NFT art with AI when NFTs were a thing, [35:19] you know, so this like 2019 or something [35:21] like that. But the way they did it [35:23] >> sometimes was uh really remarkable. [35:26] loving Holly Hernand's work around uh [35:30] taking her voice and making it so that [35:32] her voice was something that people [35:33] could use in almost open source way uh [35:37] that was quite incredible and ahead of [35:39] its time. Um and I think this like sort [35:43] of play between uh what is me, what is [35:48] the collective and what is it that is [35:52] unique and ownable versus what is not. [35:56] Uh is is quite fascinating. I've always [35:58] been fascinated by that intersection. [36:00] >> Yeah, that's cool. Just as we wrap up, a [36:03] lot of the commentary on AI that's [36:05] negative is that it takes away human [36:07] agency, human potential gets capped out. [36:10] Can you just talk a little bit about [36:12] your view on the the potential of every [36:15] human to be a creative, to be a [36:17] designer, and how AI might perhaps [36:19] magnify every every person, not just the [36:22] quote creatives where they're in some [36:24] elite class of artists in some special [36:26] building, but the ability for every [36:28] human to create and the ability for [36:30] every human to design and produce things [36:32] in the world and where you sit with that [36:35] right now? [36:35] >> I I think it's even broader. I think [36:37] it's it's um as a human in general, [36:41] preAI, post AI, you have a choice. Do [36:44] you want to consumer create? Do you want [36:47] to take the easy path or do you want to [36:50] like climb the mountain? And what I'm [36:52] seeing happen right now a lot is people [36:54] are letting AI think for them. Uh less [36:58] at Figma, but like more just like in my [37:01] interactions with the world. Uh I just [37:04] see a lot of this at the moment. And [37:07] that's one path you can take. But the [37:09] other one is [37:11] actually you use AI to further your [37:13] understanding of the world to learn to [37:15] do things you never would have dared to [37:16] do before because they were would have [37:19] taken so much time to learn or it's so [37:20] hard. Uh and to really expand like the [37:24] potential ways that you can like live as [37:26] a human. Uh that to me the latter is the [37:29] interesting part. [37:30] >> Yeah. [37:30] >> The former like why would I want [37:32] something else to think for me? like my [37:35] entire life has been about like figuring [37:37] out how to think and um I don't know I [37:40] just hope that people use AI to create [37:45] good and to like go be a force for it [37:49] rather than just kind of like joining [37:51] the hive mind. [37:52] >> Everyone can be a maker. [37:53] >> Everyone can be a maker. Everyone can go [37:55] do bold things. Uh it's about willpower [37:58] and it's about curiosity. [38:01] And if you're curious and you're willing [38:04] to do the work, uh I believe [38:06] >> be brave and speak your truth. [38:08] >> Yeah. Exactly. And then I think that you [38:10] can go do incredible things. Uh we will [38:13] probably see a vastly different setup in [38:17] the future where things that we can't [38:20] imagine are possible and things we take [38:22] for granted are no longer in sort of the [38:26] realm of what humans do every day. uh [38:28] how exactly that will look. I don't [38:30] think either of us can tell, but it is [38:33] the case that both were on the precipice [38:35] of great change. And nothing hit home [38:37] for me more than the math drop last [38:40] night uh from OpenAI of 700 plus papers [38:44] on all sorts of very famous problems. Uh [38:49] that that was a show of something. [38:52] Uh I had a lot of emotions. I don't I [38:54] don't know how to process I thought it [38:55] was the greatest day of discovery in [38:57] human history. [38:58] >> Oh, I mean factually, yes. [39:00] >> And also like something about it was [39:02] like [39:04] brilliant and exciting and something [39:06] about it was kind of sad. [39:07] >> Yeah. [39:08] >> And I I don't know how to articulate the [39:11] sadness, but it felt like we lost [39:12] something and gained something at the [39:13] same time. [39:14] >> I don't know. I lift my head up and I [39:15] look at the next frontier and I think [39:17] that the frontiers are brighter and [39:19] wider and more accessible than they've [39:21] ever been. So, I'm pretty excited. [39:22] >> Oh, I'm excited, too. And I think that [39:24] the world will look amazing and uh I'm [39:27] an optimist. I think there's gonna be a [39:29] lot of challenges and we'll figure them [39:30] out. [39:31] >> Yeah. [39:31] >> Uh how exactly and what the challenges [39:34] will be in their full form, we have yet [39:36] to see. But yeah, I do think that like [39:40] math is uh something that somehow has a [39:43] quality of mystery to it like an [39:46] unsolved problem. Uh is it you know is [39:50] this conjecture true or false? How? why [39:54] uh grappling with that mystery is [39:56] something that is special and uh I I'm [39:59] both excited for the progress and like [40:01] all the things it'll make possible and [40:03] also like I don't think it's wrong to [40:05] like mourn the loss of a little mystery. [40:06] >> Right. Fair enough. Dylan, this has been [40:08] a great chat. Should we go have another [40:10] drink? [40:10] >> Sounds good. [40:11] >> Let's do it. [40:11] >> Thanks, man. [40:12] >> Thanks. [40:28] I'm going all in.