OpenAI Design Chief Says Automated Code Shifts Power to Human Taste

OpenAI设计主管称自动化代码将重心转向人类品味 随着编程智能体使软件开发走向商品化，传统以工程为主的团队配比正在瓦解，转而青睐利用第一性原理品味和模块化原语来构建更简便界面的设计师。 针对产品设计师在面对AI驱动的快速自动化时日益加剧的焦虑，Ian Silber认为生成式工具使代码普及化，并将人类的工艺、品味和系统思维提升为核心竞争壁垒。他阐述了OpenAI如何通过模块化、可组合的原语来设计产品，在4小时的实验周期与对ChatGPT输入框等长效界面的极致微雕之间取得平衡。展望未来，Silber概述了对话界面从简单的文本提示词向主动式、多模态工作流的转变，后者可在后台无缝处理用户意图。在整个访谈中，他鼓励设计师克服冒名顶替综合征，摒弃刻板的流程教条，专注于构建能够解决真实用户问题的可运行原型。 科技行业的团队配比正在发生根本性逆转：尽管AI编程智能体让工程研发速度提升了10到100倍，但代码生成的商品化正推动初创团队的人员配置模型从每10至15名工程师配1名设计师，转向2名设计师配1名工程师等全新比例。 (13:41) OpenAI通过保持默认消费级界面的极简风格，并将高级能力部署到Codex、桌面端应用和ChatGPT Work等专用界面中，以此管理ChatGPT逾10亿月活用户面临的“能力过剩”问题。 (39:21) 虽然AI已经是一个合格的初级产品设计师，但人类设计师仍保留着三大持久优势：深刻的定性用户理解、在没有训练数据的情况下创造全新交互范式的能力，以及具有鲜明倾向性的观点。 (22:24) ChatGPT的长期界面愿景正在超越对话交互本身，迈向主动式工作流以及极度简化的通用输入框，该输入框可在后台动态调度响应速度、推理深度与模型选择。 (46:07)

Ian Silber, Head of Product Design at OpenAI, argues that despite widespread anxiety among designers facing rapid technological change, human craft and taste remain primary differentiators as AI interfaces evolve beyond text boxes into adaptive, proactive workspaces. The Velocity Gap and Industry Anxiety (2:17) According to Lenny Rachitsky's industry survey, product designers and UX researchers report the highest levels of unhappiness, exhaustion, and career anxiety in tech. Silber observes that this unease stems from ambiguity around evolving expectations and an uneven productivity shift across disciplines. While software engineers have experienced an estimated 10x to 100x acceleration using coding agents on deterministic, binary programming tasks, product design cannot simply be prompted into completion. Silber emphasizes that design remains a fluid, non-linear feedback loop requiring practitioners to explore multiple dead ends, solicit human feedback, and align cross-functional teams. While AI significantly speeds up initial ideation, it does not bypass the organizational and qualitative alignment necessary to produce high-craft software. Consequently, designers who attempt to measure their output against engineering code volume feel misplaced friction. Why Human Taste and Specialized Roles Persist (22:24) Silber rejects the prediction that AI will fully collapse product managers, designers, and software engineers into interchangeable generalists. While startup environments benefit from fluid generalists, scaling organizations still require specialized anchors: product managers define problems and coordinate resources, engineers guarantee system stability, and designers protect craft and systems-level cohesion. Silber notes that individuals who can execute all three disciplines at an elite level remain exceptionally rare. Although Silber considers AI an impressive out-of-the-box design tool, he asserts that human practitioners remain irreplaceable across three core dimensions: diagnosing latent emotional and qualitative user friction, inventing novel interaction paradigms that lack historical training data (comparable to early multi-touch or camera-first interfaces), and imparting a distinct, intentional point of view. When code generation becomes commoditized, products will compete primarily on taste, empathy, and composable architectural clarity. Dual-Track Execution for Moving Foundations (32:08) Because underlying model latency, reasoning power, and capabilities advance on a monthly cadence, hardcoded interface designs can become obsolete overnight. To navigate this volatility, Silber encourages designers to 'just do less' by extending composable primitives—similar to Notion's early modular philosophy—rather than creating bespoke features from scratch. OpenAI approaches product delivery through a split methodology based on surface stability. On exploratory features and developer surfaces like Codex, teams often ship functional prototypes to production within hours or days to gather direct public and internal feedback. Conversely, durable, high-volume consumer touchpoints—such as the core ChatGPT composer—undergo exhaustive scrutiny, where teams test dozens of variations and discard 99 before settling on a single hardened interaction pattern. Evolving Interfaces Beyond the Chat Box (41:57) Silber addresses the critique that conversational AI is merely a 'glorified terminal,' explaining that while natural conversation effectively accommodates varied user intelligence levels, text transcripts are not the terminal interface. The goal is creating adaptive software that detects user intent and automatically morphs its layout into dedicated direct-manipulation workspaces, such as modular writing blocks or visual canvas tools, rather than requiring complex prompt engineering or follow-up clarifications. Looking forward, Silber outlines a roadmap centered on proactive assistance and voice integration. Rather than waiting idly for prompts, future systems will ingest background context—such as calendar schedules and communication channels in workspace environments—to anticipate user needs. The long-term design objective is an aggressive simplification of controls, routing complex agentic tasks and model selection in the background so users interact with a single, frictionless entry point. Silber stresses that the technology sector remains in the earliest innings of the AI transition, leaving room for practitioners to invent workflows rather than adhere to obsolete orthodoxies. However, whether mainstream product organizations will successfully nurture craft-driven design or continue pressuring teams toward pure engineering output remains an open industry tension.

[0:00] Ian Silber: nobody should feel behind right now. [9:46] Ian Silber: I was trying some coding agents and it worked, but it just kept kind of running into these hiccups. And then like a month later, Codex drops and I try it again and it's like, 'Oh my gosh, it just works.' And that, I think, changes a lot of the design process because now all of a sudden, if you do have an idea, you can like plop it in there and just try it. [13:41] Ian Silber: I've always found that these like single-siloed role definitions don't really make sense. It's just that at big companies you have to have like a ladder to put people on so you can generally kind of like organize. But in reality, when you look at a team, you see that there are people that have various strengths and people that play different roles... Those lines have always been blurry, but they're becoming blurrier for sure. [22:24] Ian Silber: I think it already is an incredible product designer. [22:24] Ian Silber: Somebody made this, somebody had a point of view about this. [32:08] Ian Silber: Just do less. [32:08] Ian Silber: We try 100 things, we throw out 99. We finally ship one. [51:40] Kevin Weil: the model we have today is the worst the model will ever be. [53:41] Ian Silber: At the end of the day, we're not here to build tools like processes. We're not here to like build or token max or something like that. We're here to make great software that people love or great products or whatever it is you do. [57:43] Ian Silber: I felt like I was failing every day [60:44] Ian Silber: the most magical experience in the world [60:44] Ian Silber: have fun with it and like just try stuff [68:38] Ian Silber: Do the simple thing first

[0:00] I've been doing this sentiment survey of [0:02] the tech workforce. Surprisingly, [0:04] designers are just the most unhappy on [0:07] every dimension. [0:08] >> We're unclear right now what is expected [0:10] of a designer. Someone who might be [0:12] classically [music] trained in a certain [0:14] way of working now feeling like, "Oh my [0:16] gosh, I need to really change [0:17] everything." [0:17] >> You believe this the best time in [0:19] history to be a designer. [0:20] >> We're all extremely early in this [0:22] process. There's just a huge opportunity [0:24] for designers. Nobody should feel behind [0:27] right now. I think that if you literally [0:29] started today, you're going to have a [0:31] leg up on pretty much most people. [0:32] There's just so much opportunity [music] [0:34] to change the way you work to make great [0:36] things. [0:36] >> If you look at engineering, it's just a [0:39] completely new job versus a few years [0:41] ago. I'm curious how that translates to [0:42] product design. [0:43] >> Engineers have 10 or sometimes like 100x [0:45] their productivity, but our design team [0:48] hasn't because the design process still [0:50] takes time. It's still sometimes you [0:52] have to try a bunch of things and throw [0:54] them out. Still [music] very messy and [0:56] fluid. We had a podcast with head of [0:57] design cloud code Jenny when she talked [0:59] about [music] there's been this [1:00] traditional design process prototype [1:02] mock test research iterate and now [1:04] there's just no time for that. [1:05] >> There's certain features for chatbt [1:08] where we really obsess and we really [1:10] spend a lot of time we try 100 things we [1:12] throw out 99 we finally ship one and [1:14] other things we really embrace building [1:16] in public we take big swings and then we [1:18] learn quickly. Do [1:19] >> you think AI will ever be an extremely [1:21] good product designer? I think it [1:23] already is an incredible product [1:24] designer. [1:26] >> Today my guest is Ian Silber, head of [1:28] product design at OpenAI. Ian's been in [1:30] this role for three years now, which is [1:33] a lifetime in AI time. Prior to OpenAI, [1:36] he was at Artifact, the news app from [1:38] the founders of Instagram that I [1:39] absolutely loved. Before that, he spent [1:41] 8 years at Instagram. The role of design [1:44] is in such an interesting place right [1:46] now and Ian has such a unique and [1:48] insightful and fascinating take on the [1:51] future of product design and what the AI [1:53] tools that we use everyday will look [1:55] like in the future. A huge thank you to [1:57] Cleo and Joel Loenstein for suggesting [1:59] topics for this conversation. Before we [2:01] get into it, don't forget to check out [2:02] lennisproass.com [2:04] for a year free of the hottest and most [2:06] beautifully crafted AI products in the [2:08] world available exclusively to Lenny's [2:10] newsletter subscribers. With that, I [2:13] bring you Ian Silbert. [2:17] Ian, thank you so much for being here. [2:20] >> I'm so excited to be here. Thanks for [2:22] having me. [2:22] >> I'm even more excited. Something that [2:24] I'm very excited to get your take on. I [2:27] don't know if you've watched me doing [2:29] this, but I've been doing this uh [2:30] sentiment survey of the tech workforce [2:33] over the past couple years. I think it's [2:34] the biggest survey of its kind where we [2:36] basically survey people in tech about [2:38] how they're feeling about their careers, [2:41] their work, their roles, especially with [2:43] the rise of AI this last year. And kind [2:47] of surprisingly, designers and user [2:50] researchers are just the most unhappy on [2:54] every dimension, like every actual [2:56] dimension. They're the most overwhelmed, [2:59] the most anxious, the most the least [3:01] optimistic, the most worried, uh the [3:03] most tired, also the least likely to [3:06] recommend their role to people entering [3:09] the workforce. [3:11] I'm curious if you're seeing this [3:13] yourself in the design community, and if [3:16] you are, what do you think is going on? [3:18] >> Uh, first of all, it's an incredible [3:19] survey, and it's such great, I think, [3:21] information to have as an industry. Um, [3:24] and I think it's something we need to [3:25] unpack, you know, and I see I I see a [3:28] mix uh for sure. I think like I can [3:30] speak most to design. Um, and I mean I [3:35] talk to a lot of designers that are very [3:38] optimistic and very excited. Um, and so [3:42] I see in my world I see luckily I I'm or [3:46] I I I see a lot of that. But I can [3:49] totally understand why people might be [3:52] feeling very uncertain, right? Because [3:54] there's so much discussion about, well, [3:57] is the role changing? Should my [3:59] expectations be different? Um, and you [4:03] know, well, do I need to [4:06] do I need to do X, Y, and Z? Do I need [4:08] to be shipping code every day? And you [4:11] know someone who might be uh sort of [4:13] classically trained in a certain way of [4:14] working now feeling like oh my gosh I [4:16] need to like really change everything [4:18] and I think it's to me right now it [4:20] feels like the the uncertainty of of [4:23] that I can totally understand right [4:24] because as an engineer I think people [4:27] have they've already sort of gone [4:28] through that shift like we see that [4:30] engineers are working totally different [4:32] and we also you know one thing I do hear [4:35] and I this came up a lot in my own team [4:37] when when we did we do similar surveys [4:39] internally [4:40] One of the big things that came up was [4:42] you know we see that uh uh you know [4:46] engineers have 10 or sometimes like 100x [4:49] their productivity [4:51] uh but our design team hasn't right [4:54] because the design process still takes [4:56] time. It's still sometimes you have to [4:58] try a bunch of things and throw them out [4:59] right versus maybe working with a coding [5:01] agent you ask it to do something it's [5:03] pretty binary now at this point it [5:05] usually sort of gets it pretty well. Um, [5:08] and but the design process is really [5:10] still very messy and fluid and you think [5:14] you have a great idea and you try it and [5:15] it sucks and then you try it again and [5:17] it sucks and then you you give it to you [5:20] know users or internal you get internal [5:22] feedback and it's wrong and you still [5:23] have to go through that arduous and kind [5:26] of messy but really fun process. Um, and [5:30] that hasn't really comp that it's [5:32] compressed in that you can get much more [5:34] ideas out fast, but you still need that [5:36] that whole sort of feedback loop. [5:38] There's the alignment of getting many [5:40] people uh, you know, especially in a [5:42] bigger company like that still has a lot [5:45] of overhead. So the point is that I [5:48] think that it's we're unclear right now [5:50] like what is expected of a designer and [5:53] that lack of clarity I can see totally [5:55] lead to you know um some some fears uh [5:59] or some unease I guess. But on the other [6:02] hand I've seen that I mean our designers [6:04] what's been what's great I think about [6:06] at OpenAI is like we're such an [6:08] experimental company. We are driven by [6:13] you know research and we treat our [6:16] company as a research lab and we totally [6:18] embrace that and that's a new thing that [6:19] I really kind of learned and and have [6:21] really come to love about OpenAI is that [6:24] started as a research lab that DNA is [6:27] like deeply embedded in everything we do [6:30] and as designers I think that you know [6:32] you come here we look for people that [6:34] are not don't have a background in AI [6:36] but are like really curious about the [6:38] tools and how it can [6:40] And we found that like just by [6:44] sort of supporting that excitement to [6:46] explore that people feel less pressure. [6:49] Um it's not like you need to change the [6:51] way you work but it's more about [6:53] encouraging exploration. Um, and when [6:56] we've done that, when I go and talk to [6:59] um, both very senior designers and early [7:01] career designers at our company, they're [7:03] so excited because they I I literally [7:05] was just asking some people before this [7:07] cuz I wanted to get some different takes [7:08] and they, [snorts] you know, I think the [7:10] feeling is that like they have so much [7:13] more ability to be creative, to express [7:17] themselves, to try bad ideas and throw [7:20] them out. you can just like really move [7:22] through lots of different ideas and it [7:23] becomes this incredible way to create [7:26] and the things that I'm seeing from my [7:27] team the the bar is raising um the the I [7:32] think the thinking is getting stronger [7:34] the more that they're using um these [7:36] tools. So, I think for the people that [7:38] that that are starting to embrace this, [7:39] they're starting to feel really excited [7:40] and realizing that they have just so [7:42] much agency as as designers and like um [7:47] you know, so much more ability to have [7:49] like really high impact uh as as a as a [7:52] designer. [7:53] >> This episode is brought to you by our [7:54] season's presenting sponsor, WorkOS. [7:57] What do OpenAI, Anthropic, Cursor, [8:00] Replet, Sierra, Clay, and hundreds of [8:02] other winning companies all have in [8:04] common? They are [music] all powered by [8:06] work OS. If you're building a product [8:08] for the enterprise, you've felt the pain [8:10] of integrating single signon, skim, [8:12] arback, audit logs, and other features [8:15] required by large companies. Work OS [8:17] turns those deal blockers into drop-in [8:19] APIs with a modern developer platform [8:21] built specifically for B2B SAS. [8:24] Literally, every startup that I'm an [8:26] investor in that starts to expand up [8:28] market ends up working with Work OS. And [8:30] that's because they are the best. [8:32] Whether you are a seedstage startup [8:33] trying to land your first enterprise [8:35] customer or a unicorn expanding [8:37] globally, work OS is the fastest path to [8:39] becoming enterprise ready and unblocking [8:42] growth. It's essentially Stripe for [8:44] enterprise features. Visit workos.com to [8:46] get started or just hit up their Slack [8:48] where they have actual engineers waiting [8:50] to answer your questions. Workos allows [8:52] you to build faster with delightful [8:54] APIs, comprehensive docs, and a smooth [8:57] developer experience. Go to work os.com [8:59] to make your app enterprise ready today. [9:02] What I'd love to do in this conversation [9:03] is help folks that are in that bucket of [9:05] well this is sucks and things aren't [9:07] going well because as you said and this [9:09] is exactly what we found that survey is [9:11] about half of people are having the best [9:14] time of their career having the most fun [9:17] the most optimistic about where things [9:18] are going and then the half are not [9:21] having a really hard time. And exactly [9:23] what you said, the people that are [9:25] having a great time report like the main [9:27] the most the strongest correlation is [9:29] they feel amplified by AI. It's actually [9:32] making them better. And that's that the [9:34] way you described engineers, they're [9:35] they're seeing that now. They're like, [9:36] "Holy I can do so much more." And [9:38] so just to kind of reinforce the point [9:40] you're making, think attributes of [9:43] people that are doing well um feel [9:46] amplified, very curious, have time to [9:48] explore, feel like they can be more [9:49] creative. Is there any more thing more [9:51] along those lines of just like patterns [9:53] of people that are doing really well and [9:54] having a great time that might be [9:56] helpful for people to hear that are not [9:58] having a good time? [9:59] >> What I see in in the the folks that I [10:03] think are really embracing this, right, [10:04] they are using it every step of the [10:06] process to kind of help them, right? And [10:08] it and I think it helps, you know, we [10:10] have we have a community of course of [10:12] designers that are all trying this [10:13] together. So if there's any way to kind [10:15] of uh find that and that that's [10:17] something that uh we should think about [10:18] more as a broader community how we kind [10:20] of help each other but like I think that [10:23] learning kind of different ways of [10:24] working right and so it might be okay I [10:27] have this idea let me just actually [10:28] start by trying what if I just threw [10:30] this right into codeex or cloud or [10:33] whatever tool you use like throw it into [10:36] uh into an agent to help you think about [10:37] it to help you prototype it just try a [10:40] different way of working um it doesn't [10:42] necessarily mean that that has to be the [10:43] exact right way. But um I don't know. I [10:46] think we found that just people that are [10:48] are embracing this that are adaptable. I [10:50] think adaptability is like another term [10:52] that I think we all hear a lot now, but [10:54] it's like so true. It's like knowing [10:57] that this stuff changes very quickly. [11:00] What didn't work yesterday might work [11:02] today. I mean literally like within [11:06] you know [11:08] six months this entire industry has [11:10] shifted. I remember I was actually out [11:12] on on paternity leave uh towards the end [11:15] of last year and I was working on some [11:18] side projects. I was trying some [11:21] uh coding agents and it just like it [11:23] worked but it just kept kind of running [11:24] into these hiccups [11:26] and and then like a month later like [11:29] codeex drops and I try it again and it's [11:30] like oh my gosh it just works. Um and [11:33] that I think changes a lot of the design [11:35] process because now all of a sudden if [11:36] you do have an idea you can like plop it [11:38] in there and just try it. Um, and so [11:40] anyways, I think it's it's just really [11:42] about like not being scared. I think [11:44] another thing to think about is that [11:48] we're so early. We're like so in early [11:50] innings of this entire [11:53] um evolution, right? It feels like [11:55] things have changed a lot and it feels [11:58] like maybe we're pretty far into this [11:59] shift. But the good news for anybody who [12:02] is trying to design or get into design [12:05] whether again like if you're early [12:07] career or late career or you never even [12:11] like actually design anything but you've [12:12] been thinking about it is like we're all [12:15] figuring this out as we go. Nobody has [12:18] like some perfect answer um about the [12:22] how the design process should work. Um [12:25] and it's whatever you think is working [12:27] today will probably change again. And so [12:29] I think what's encouraging about this is [12:32] that nobody should feel like behind [12:35] right now. I think that if you literally [12:37] started today, you're going to have a [12:39] leg up on pretty much most people. Um [12:42] and uh just really thinking about like [12:44] how you can just make great stuff. Um [12:46] and uh there's just so much opportunity [12:49] to change the way you work work or to to [12:52] just to make great things. and [snorts] [12:54] what we think is like really cutting [12:56] edge today will look so archaic in you [12:59] know a year let alone a month right and [13:03] that's something I' I've learned you [13:05] know I've been at open air for three [13:06] years and I think about the stuff that [13:08] they you know we were working on when we [13:11] when I first joined um I mean I remember [13:15] getting a demo of of like our our early [13:18] voice mode that we hadn't shipped or [13:20] computer use that like you know barely [13:23] worked worked uh and now it's like [13:25] become it's it literally has become part [13:27] of the process of how we design. We use [13:29] both of those tools to to build [13:31] software. And so anyways, the point is [13:34] it's like changing very quickly. We're [13:36] all extremely early in this process and [13:38] nobody really has like the it figured [13:40] out. [13:41] >> I love that you've also when we were [13:43] chatting ahead of this year, you you [13:44] actually told me you believe this is the [13:46] best time in history to be a designer. [13:48] >> Yeah, I mean I I truly believe this. Um [13:50] I can't even imagine you know I so when [13:53] I uh okay history lesson when I was [13:55] getting into design I mean I was like uh [13:59] in my teenage years right and I was like [14:03] uh you know to learn things it was like [14:06] you know I uh I was kind of like [14:08] learning design and code and you could [14:11] like view source on a HTML web page and [14:14] kind of see how things were built and [14:16] that was like kind of how you had to [14:18] learn and you know um you could I don't [14:22] know it's like Dreamweaver and you know [14:25] like there were these really to tools [14:28] that actually were really cool. Um but [14:30] to compare those to today if I were just [14:33] getting into design the fact that you [14:34] can like learn so quickly you can kind [14:36] of like have an idea to working version [14:40] of it in like [14:42] like a matter of minutes. Um [14:46] it's just you can be so expressive and [14:48] and you can um I think anybody can be a [14:52] designer right um and I think that as [14:56] more and more software is made more and [14:59] more products are made I think that the [15:04] I do think that the human element of [15:06] design is going to really be [15:11] start to become more and more important. [15:13] It's that's always been the case that I [15:14] think that, you know, companies with [15:16] great design, you know, can uh of course [15:19] stand out. I think it's going to be more [15:21] and more important, right? I think that [15:23] products that um fully understand their [15:26] users and that make something delightful [15:29] when anybody can make anything, you [15:31] know, that's going to become one of the [15:33] most important kind of uh like sort of [15:36] dimensions of a product and that's how [15:37] people are going to evaluate it. And so [15:39] I think designers uh stand like that is [15:42] that is our core job, right? Um you know [15:44] and and then on top of that [15:48] I think that designers are just much [15:49] more empowered. Uh so you know at a um [15:54] you know I was talking to some friends [15:55] that are thinking about starting a [15:57] company and talking about the uh sort of [16:01] ratio of design to engineer and how like [16:04] at a startup maybe that's totally [16:05] different. It might it may have been [16:07] like you're going to have one designer [16:08] and uh I don't know 15 engineers or [16:11] something like that. You know that maybe [16:13] a startup I've seen that sort of ratio. [16:15] Uh but they were they were talking about [16:17] well maybe we'll have like two designers [16:19] and one engineer. One really good [16:21] engineer who can make sure that like [16:23] we're really building something that's [16:25] rock solid like from from an engineering [16:27] standpoint. But then we want two really [16:29] creative people that are going to be [16:31] thinking about how we kind of build this [16:33] product. And so thinking about that, I [16:36] think that there's just like a huge [16:38] opportunity for designers. I think that [16:40] internally we see that more and more [16:43] we're we like designers we are are [16:46] totally completely empowered to to take [16:49] an idea to take a early research concept [16:52] and try to figure out how we can [16:54] actually like package that up into [16:55] something that people want to use or [16:56] that um that solves some sort of uh [16:59] unmet need for our customers. and um [17:02] that is a designer's job. And so I think [17:04] that maybe more and more people start to [17:06] think of themselves as designers. Um but [17:10] either way, I think it's like the best I [17:11] I really do think it's the best time to [17:12] be a designer. [17:13] >> So along those lines, when Mark Andre [17:16] was on the podcast, he had this [17:17] hilarious way of describing the state of [17:19] the of the product or right now. Uh he's [17:22] like basically there's this three-way [17:23] standoff. PMs, engineers, designers, [17:26] they're all they all think the other [17:27] roles are going to go away and they're [17:28] the ones that are going to win in the [17:30] end. And so there's less like Spider-Man [17:32] meme where they're all pointing at each [17:33] other. And I'm curious because I I can [17:36] see a case for each role becoming the [17:38] dominant role. And I completely agree [17:41] with you that with all of the tools [17:43] being shipped, like there's like a [17:45] thousandx explosion and products being [17:47] shipped, there's just like endless [17:49] streams of things to pay attention to. [17:51] You think design would be such an [17:53] opportunity to differentiate and stand [17:55] out. I haven't seen that yet in just [17:58] like an acceleration of design hires and [18:00] explosion as you said like two designers [18:02] to one engineer. Do you think do you [18:04] think that's coming? Why do you think [18:06] that hasn't happened yet? Do you think [18:07] it's as you said just early and people [18:09] will realize oh wow we need a lot more [18:10] designers. What do you think is going [18:12] on? [18:12] >> We might start to see that happen more. [18:15] Um or if anything we might start to see [18:18] engineers wanting to become a little bit [18:20] more design focused. I also think that [18:23] designers are getting more productive. [18:26] Um it's not necessarily the same direct [18:30] output of like engineers 10xing. Um but [18:33] we are seeing that that designers um are [18:36] are able to do a lot more. But no, I [18:39] mean I think that that that is something [18:40] we're sort of always thinking about and [18:42] and evolving. Um the other thing I [18:45] wanted to talk about too is like the [18:46] difference between PM and engine and [18:48] design and are they all blurring [18:49] together? And my point of view is that [18:52] they they definitely are and they [18:54] there's always been overlap as far as [18:56] I've seen it, right? I've seen designers [18:59] that are like kind of really good [19:02] product, you know, manager type [19:04] archetype, but can design and I've seen [19:06] designers that are like really good sort [19:09] of polish engineer type, you know, like [19:12] so that I think I've always found that [19:14] these like single siloed um role [19:18] definitions don't really make sense. [19:20] It's just that at big companies you have [19:22] to have like a ladder to put people on [19:24] so you can generally kind of like [19:26] organize. [19:28] But in reality, you know, when you look [19:31] at a team, you see that there are people [19:33] that are have various strengths and [19:36] people that play different roles. And so [19:38] I've I've always kind of seen that. Um I [19:41] think there's those lines have always [19:42] been blurry, but they're becoming [19:43] blurriier for sure. Um, but the other [19:46] thing is I do still think that and and [19:47] maybe this is thinking about it from [19:49] like a slightly larger company [19:50] perspective, but I think you do still [19:53] need people that have certain [19:54] responsibilities, right? Because if you [19:57] have one person that's expected to [20:00] design, to rally a team around [20:02] something, you know, to align a whole [20:05] company, uh to do all the work that like [20:07] a PM does, for example, and also be [20:09] designing it and also be engineering it, [20:11] it's just maybe not you you might it [20:13] might be hard to find one person that's [20:15] good at all of those different skills. [20:17] And so I still find a lot of value, for [20:19] example, in having like an excellent PM [20:21] that can help rally multiple teams [20:24] together that can help um sharpen kind [20:27] of the perspective or the problem space [20:30] that a team is focused on um that can [20:33] help with like resourcing, all of those [20:35] things, right? And so I do think there [20:37] will still be value in these um and [20:39] people focused on different parts of [20:41] building a project or a product. Um that [20:45] doesn't mean that um maybe there's more [20:47] fluidity that people can move between [20:48] them but having somebody who is sort of [20:50] generally responsible for these [20:51] different things or an engineer who's [20:53] like really making sure that the output [20:55] is totally sound and you know is [20:58] actually like um lading up to something [21:00] that is going to work for the broader [21:02] system. [21:03] >> I completely agree. So just to make sure [21:05] I understand what you're saying is just [21:07] there's talk about all the roles [21:08] blurring. is like maybe we'll all just [21:10] be builders and we do everything and [21:12] you're saying is like no that's not [21:14] going to happen the like the ven di I'm [21:16] picturing ven diagrams of product design [21:17] engineering like converging but you're [21:20] saying they're going to continue to stay [21:21] their own roles there's going to be a PM [21:23] design engineer but each can do more of [21:25] the other roles [21:26] >> I think that's right I I do think that [21:28] like for and again this is not a blanket [21:30] statement right because I think if [21:32] you're at a startup that's like very [21:33] different I would I would if I were [21:35] starting a company right now I [snorts] [21:37] would hire like the best generalist [21:39] people I could possibly think of, [21:41] someone who I think could kind of [21:42] fluidly move between that. But maybe [21:43] again that's maybe that's always kind of [21:45] been true. Um, but at at a bigger [21:49] company, I do think that I've seen like [21:50] designers um that are, you know, maybe [21:55] incredible at designing, but could [21:57] really use a product manager who can [21:58] help them kind of with with like hone um [22:02] both their thinking and how and and then [22:04] how they work and the operations of it [22:06] all. So the point is I do just think [22:08] that there are there are different [22:11] it's it's almost like in t taking the [22:13] people out of it I think that there are [22:15] different sort of hats that you need to [22:17] wear to build great software and I think [22:20] that um those are continue to be [22:23] important functions I think [22:24] >> so this begs the question do you think [22:26] AI will ever be an extremely good [22:29] product designer and just to like it's [22:31] interesting to ask you this question [22:32] because on the one hand you're like oh [22:35] yeah open like ChachiP is going to [22:36] become incredible at everything. On the [22:38] other hand, it's hard for a head of [22:40] design to say, "Oh, no, design's going [22:42] to be done by AI." I'm curious to get [22:44] your your take and you know, it like [22:45] sounds crazy, but like engineers never [22:47] thought it would be able to write code [22:48] the way it is now. Um, so just yeah, [22:51] thoughts on just how good AI will get. [22:53] >> Maybe one way to put it is I think it [22:54] already is an incredible um product [22:57] designer. [laughter] [22:59] Um, it it really is actually already [23:02] great. The cool thing about it though is [23:04] it's accessible to every single person. [23:06] Um, and it's a tool that anybody can [23:08] use. I do think it's really good. Now, [23:10] that doesn't necessarily mean that it's [23:11] like the best visual designer or best [23:13] at, you know, uh, information hierarchy [23:17] or or typography or even, you know, user [23:21] experience and interaction design. It's [23:23] not necessarily like the best at any of [23:25] those. It's impossible to predict where [23:28] things are going. And like I said [23:30] earlier, like you know, how much has [23:32] changed in three years? Of course, [23:33] they're going to continue to get better. [23:35] Um, but the way we think about it and [23:38] and I and I really do believe this is [23:40] like we're building tools to enable [23:42] people to be more creative, to be um to [23:44] to do more, to build more. And so it's [23:47] it is already a great designer, but you [23:51] it's at your fingertips to kind of [23:52] figure out how to use it and how to get [23:54] the best out of it. Where do you think [23:55] human brains will continue to be most [23:58] valuable as AI gets better at product [24:00] design specifically? What are kind of [24:02] like the remaining areas for us to be [24:04] useful? [24:05] >> Truly understanding [snorts] [24:07] what [24:10] people need. um you know I mean the [24:15] human uh feedback loop of watching [24:18] people use something and and [24:19] understanding that um the you know [24:24] inventing something new um I think uh [24:28] and then of course there's just the pure [24:30] like [24:33] human kind of take right like oh [24:36] somebody made this somebody had a point [24:37] of view about this um I just think that [24:40] I do think that will stand out more and [24:42] more and I think that that's like an [24:44] important thing for a designer to like [24:45] really hone is like what is my point of [24:47] view? What who am I building this for [24:48] and why? Um and I don't know I think [24:51] like as as as AI becomes more and more [24:55] like better and better for sure there [24:56] will be an opportunity to always kind of [24:59] stand out and push how can you push that [25:00] even further. [25:01] >> Yeah, that touches on uh a conversation [25:04] I had with um the PM of codeex. I asked [25:06] him just why is AI not good at design? [25:09] It's actually gotten better since that [25:10] conversation, which was just like a [25:11] month or two ago. And and he had a [25:14] number of reasons for it. One is just [25:16] like like part of great product design [25:18] is doing something new. And so if it's [25:20] trained on all the most interesting [25:22] websites today, it's just like okay, but [25:24] that's not what that's not interesting [25:26] anymore. And the other point he made is [25:28] like AI researchers aren't innately [25:30] great at design and so they may not [25:32] understand exactly what it takes to [25:34] train a model to be really good at this. [25:35] Uh thoughts? What do you think? What do [25:37] you think it'll take to get AI to be [25:38] much much better at product design? [25:40] >> I mean, the new thing is interesting, [25:42] right? Because I think that [25:45] everything we're doing is trying to [25:47] design something that hasn't existed [25:49] before, right? Um, you know, if you look [25:52] at if you look at great great products [25:54] that have come out in the last, you [25:56] know, 10 20 years or whatever, right? [25:59] It's like all of a sudden, you know, [26:01] take the iPhone, we're designing for [26:02] multi-touch. And that opened up like [26:04] this whole new interaction paradigm. And [26:07] there's no training data for that, [26:08] right? You have to you're kind of really [26:10] trying to understand this new thing, [26:11] right? And then you had something like [26:13] Instagram that was like really embracing [26:16] um the sort of like the fact that [26:19] everybody had mobile phones. And then [26:21] Snapchat, I think, is another good [26:22] example of great design and that it was [26:25] kind of a maybe people thought strange [26:27] sort of like interaction design, but it [26:31] completely thought like flipped on its [26:33] head the way people might want to [26:35] communicate with each other. And to me [26:36] that that is like great design. It's [26:38] like truly understanding like I you know [26:43] the people that like people have this [26:45] problem or there's this unmet need and [26:47] and I'm going to build it with this in [26:49] this unique way. So yeah, I do think [26:51] that there's that sort of like fresh new [26:53] kind of way. And then now with AI, it's [26:54] like we're designing like how do you [26:56] talk to something that uh is how do you [26:59] talk to AI with your voice, you know, or [27:01] how do you use an agent to like carry [27:04] out tasks for you? It's like yeah, there [27:06] isn't a lot of precedent for this. And [27:08] so I think that, you know, humans are [27:10] able to kind of do a really good job [27:13] kind of evaluating, you know, what good [27:16] is and um, you know, or at least kind of [27:19] like trying a bunch of things and and [27:20] learning along the way with people. And [27:22] so um I don't know, but again, like all [27:25] this stuff changes really quickly. It's [27:27] hard to say, but uh I think um I think [27:30] there's a lot of reasons why uh having [27:32] that human in the loop is going to be so [27:33] important. And kind of following that [27:35] thread of the human in the loop when [27:37] you're hiring designers these days, [27:39] what's kind of trending up in what you [27:42] look for, what you think is more and [27:43] more important in the designers that [27:45] will be successful today, especially at [27:46] OpenAI, what's trending down with [27:48] anything? [27:50] >> This might be a luxury of having a [27:52] slightly larger team, but if you're, you [27:54] know, at our scale, um, you know, but [27:58] even at at a startup or a smaller [27:59] company, I think this still be true. I [28:01] really believe in hiring a well-rounded [28:03] team that has many different strengths. [28:05] So, I think it's important to have kind [28:08] of like your handful of or or the right [28:11] ratio of sort of generalists that can [28:13] kind of do everything, but uh you know, [28:16] we also hire like incredible visual [28:18] designers or we have like like a really [28:22] amazing sort of brand team that helps us [28:24] with kind of like the sort of maybe [28:27] Yeah. expression of our brand both out [28:29] and in a in the product um or great [28:33] prototypers or um you know um and so I [28:37] think that rounding that out or or great [28:40] sort of product thinkers that are a [28:41] little bit more strategic or people that [28:43] are like really obsess over like the [28:46] very cutting edge of where these things [28:48] are going and I haven't found that one [28:51] designer necessarily checks every single [28:52] one of those boxes. I think people just [28:55] naturally are, you know, drawn to [28:58] certain things and I think embracing [29:00] that when you build a team is actually [29:02] like um if you if you have the luxury of [29:04] doing that is like really valuable. [29:06] Okay. So that's like my a broad [29:08] statement. But then as far as what we [29:10] look for, we do definitely, you know, I [29:13] always tell candidates, we do not expect [29:15] you to have a background in AI because [29:17] again, this is evolving so quickly, but [29:19] we definitely expect high curiosity [29:22] about it and and and a high aptitude to [29:26] jumping in and learning how how you know [29:29] how you're going to use this in your [29:31] process, how it works. and not just how [29:33] you use it in your process, but how [29:35] especially for us like how you're going [29:36] to create tools to enable more people to [29:39] use it. Um, and so it's more about the [29:43] curiosity I think um is something that [29:46] like we really push on uh or look for. [29:50] And then um definitely prototyping um [29:53] prototyping is something I've always [29:55] valued in my career. Um, but I just [29:58] think that, you know, it's become more [29:59] and more accessible and so that's [30:02] something that, uh, I think really does [30:04] stand out usually when we talk to to [30:07] candidates. And then I do think that, [30:10] yeah, just like having some having a [30:12] point of view on on where where these [30:14] things are going or or just like any of [30:16] the tools that people have worked on and [30:18] really thinking that strategic thinking. [30:20] I guess last last one on this is like [30:23] systems thinking um is becoming more and [30:26] more important and [30:29] um you know especially as you're I like [30:32] in our world we're it's interesting [30:33] because we're trying to build [30:36] you know we're building the super [30:38] intelligence that has uh that has all [30:41] these different capabilities and we need [30:44] to think more about the underlying [30:47] primitives that we can build that can [30:49] kind of build on top of each other, [30:51] right? Um, you know, this is something [30:53] that I think like notion did really well [30:55] early on was like we're going to build [30:57] these building blocks, right? And then [30:58] like everything kind of is composable. I [31:01] think that's becoming really important [31:02] too as you think about like well we [31:05] don't want to build all these sort of [31:06] one-off siloed experiences. We want to [31:08] be able to build and ship systems that [31:12] can build on top of each other so that [31:14] you know when someone goes and uses our [31:16] product it feels like one cohesive [31:18] simple thing and the model can reason [31:20] about how it can pull all these things [31:21] together. So that like idea of like [31:23] really understanding the system and [31:24] thinking about that is becoming even [31:26] more important especially because it's [31:28] kind of complex of all the things that [31:30] it could possibly do. This is the fifth [31:32] podcast in a row I've done where systems [31:34] thinking has come up as a skill that [31:37] people are more and more looking for and [31:40] it's it's like so funny. I need a bingo [31:42] card for systems thinking but it makes [31:44] so much sense and I think it especially [31:46] makes sense from the designer [31:47] perspective. If we come back to just why [31:50] my take on why designers are kind of [31:52] frustrated and and stressed right now is [31:53] just there's less time to do design the [31:56] traditional design process. things are [31:57] just moving like engineers are shipping [31:59] 5,000 PRs a day and so to your point [32:02] this idea of building blocks and design [32:04] system langu and and just a platform for [32:06] people to build on feels incredibly [32:08] important. [32:08] >> Totally. It's very hard but I um I I [32:12] yeah it's like it's funny we've had [32:15] these conversations where I'm like just [32:17] do less [laughter] [32:19] like my encouragement has been do less [32:21] there like don't design it if you don't [32:23] have to. Maybe there's already an [32:24] existing system or component or [32:26] something that we can build on top of or [32:28] maybe we don't even need this feature. [32:30] We actually just need to kind of uh uh [32:33] like you know extend an an existing [32:36] feature. Um I think that's an important [32:39] thing and and part of that is like [32:42] thinking outside of the bounds of what [32:44] you're working on, right? and thinking [32:46] across everything I think is becoming [32:48] much more important and the speed for [32:50] sure I think is is a reason why that's [32:52] so important right now [32:54] >> just to follow that thread because that [32:55] feels like a pretty big change in the [32:56] design world there used to be this we [32:58] had a podcast with u she was head of [33:00] design club code Jenny Wen and she [33:02] talked about this and I think it blew a [33:04] lot of people's minds just there's been [33:06] this traditional design process [33:07] prototype mock test research iterate [33:11] okay cool ready for engineering build it [33:13] iterate ship it and now there's just no [33:15] time for that. And so the role in her [33:18] eyes has just moved to big picture [33:20] planning and just like steering people [33:22] in a good direction. Does that resonate? [33:25] How much is that has changed in in the [33:27] world of a designer? [33:28] >> I think that 100% we we do work that [33:31] way. It it's it depends. I think for us [33:34] it depends on like what sort of what [33:35] stage you're at in the in the uh sort of [33:38] process or um you know uh how uh maybe [33:43] how far along the product is. There are [33:45] certain features for chatbt where we [33:48] really obsess and we really spend a lot [33:50] of time. We try a million things and we [33:52] throw out you know uh we try 100 things [33:55] we throw out 99. We finally ship one. um [33:59] you know uh the like uh the or like the [34:04] composer is like changing every single [34:05] day and we try many things we throw them [34:07] out but or like and you could take that [34:09] for many different things within chbt [34:12] um and then there are other things like [34:14] codecs where um you know I don't know I [34:18] mean the super app is a good example [34:19] where we definitely are just like trying [34:22] stuff really quick I think that we [34:23] really embrace building in public [34:26] [laughter] which is a different way of [34:28] working. But I think that like when you [34:30] get comfortable with it is actually [34:31] really gratifying or satisfying because [34:34] you can try stuff and get instant [34:36] feedback both internally and externally. [34:39] Um but yeah, I mean it is uh moving [34:43] really fast and I think it just depends [34:45] right like some like like I said certain [34:47] things we really sweat the details and [34:49] we try um you know and and we do test we [34:53] AB test we'll do user research um and we [34:56] kind of iterate and evolve a lot and [34:58] then other things we uh we take big [35:00] swings and then we learn quickly um but [35:03] yeah I don't know I think you have to [35:05] like I think the biggest thing is like [35:08] embracing that the technology might [35:11] change in like a a month or you know and [35:15] maybe a certain interaction or or [35:19] setting or or whatever that you needed [35:21] doesn't need to exist or now a new one [35:23] does need to exist or a new capability [35:25] is there or the latency profile has [35:28] changed such that you can like rethink [35:30] this or whatever all these capabilities [35:33] are changing or the the intelligence has [35:35] changed or and so you have to be just [35:38] really quick on your feet and and maybe [35:40] not be so precious about it until it's [35:43] at the point where okay this is kind of [35:44] hardened like we don't think that this [35:46] is necessarily going to change tomorrow [35:48] because of the technology so let's make [35:50] this the best it can possibly be you [35:52] know like we're we're thinking a lot [35:54] about how we should evolve sort of [35:57] really think about chatbt becoming like [35:59] deeply embedded in a per like your [36:02] personal life and how how should the how [36:05] should the app feel that's something [36:07] that we're not going necessarily design [36:09] and ship overnight and just like put it [36:11] out there. We're spending we're really [36:12] thinking about that and we're trying a [36:14] bunch of different ideas and we're [36:16] putting it up on you know uh you know a [36:19] whiteboard or you know we're doing some [36:22] design in Figma. We're prototyping it. [36:24] We're probably going to dog food a bunch [36:25] of different ideas [36:27] long before we ever ship something. Um [36:30] that but that's a bigger project, right? [36:32] And then there's other things where [36:33] we're like idea to ship in like, you [36:36] know, 4 hours. [36:38] >> This all makes sense. Essentially, what [36:39] I'm hearing is you pick your battles and [36:41] you focus the design effort on the [36:44] things that will be durable and not [36:46] change. [36:47] >> Exactly. Yeah. Yeah. And it's hard to [36:49] know what that is. [laughter] [36:51] >> Yeah. [36:52] >> Yeah. This episode is brought to you by [36:55] Mercury. Radically different banking [36:57] loved by over 300,000 entrepreneurs. And [37:00] now with command. I've been a customer [37:02] of Mercury's for over six years. I have [37:04] never once thought about leaving. [37:06] Mercury is basically what happens when [37:08] banking is built by product people, not [37:11] by bankers. They make it so easy, dare I [37:14] say fun, to send invoices, move money [37:17] around, set up virtual cards for folks [37:19] on my team. Does your bank have an API, [37:21] [music] [37:22] a terminal native CLI, or an AI ready [37:24] MCP server? I don't think so. And just [37:28] recently, they launched Command, a [37:30] conversational interface built directly [37:32] into Mercury, [music] which acts as your [37:34] financial operator. I've been using [37:35] Command to transfer money around, to [37:37] figure out what categories I've been [37:39] spending the most money in, analyze my [37:41] cash flows, [music] and just today, I [37:43] used it to find out how much I've made [37:44] from a specific sponsor over the past [37:46] year. I just [music] asked, "How much [37:48] have I made from X over the past year?" [37:50] 10 seconds later, I have an answer. It [37:52] is so freaking cool. Visit mercury.com [37:55] to learn more and apply online in [37:57] minutes. Mercury is a fintech company, [37:59] not an FDIC insured bank. Banking [38:01] services provided through Choice [38:02] Financial Group and column NA members [38:05] FDIC. So, I asked uh a bunch of people [38:08] that know you well what to talk to you [38:09] about. And uh Joel uh Loenstein, who [38:13] happens to be the head of design at [38:14] Anthropic, is a good friend of yours, [38:16] which I love. and he said what he [38:18] imagines the hardest part of your job is [38:20] is balancing the needs of the existing [38:24] billion monthly active users of chat GPT [38:27] with the need to innovate and push and [38:29] iterate and invent. And so I'm just [38:32] curious how you just think about that, [38:33] how you find that balance knowing you [38:35] have to cater to both sides. [38:37] >> That's a great question and it's a [38:39] really fun part of the way we work. Um, [38:44] I think that we have we have vastly [38:47] different audiences, like really really [38:49] different. I've never worked in this [38:51] kind of quite quite a spread. Um, you [38:55] know, like at Instagram, uh, I I I spent [38:59] eight years working at Instagram, we had [39:00] a billion user or billions of users, [39:02] maybe. I don't actually know what [39:03] they're at now, but it was a lot of [39:04] people. Um, but the product was pretty [39:07] simple and it was sort of, you know, a [39:09] couple ways you could use it. We did [39:10] think about like creators and, you know, [39:12] larger accounts that had special needs, [39:14] but this is like vastly different where [39:16] you have someone who's, you know, using [39:19] this to help them come up with the [39:21] recipe to eat tonight. Um, or, you know, [39:25] brainstorm their next haircut or [39:27] something like that. And then there's uh [39:29] somebody who's like transforming the way [39:31] they're doing farming for their you know [39:35] um their like you know helping to [39:38] automate their farm in Japan or [39:39] something right and uh it's just like [39:43] the broadest spectrum ever. Um generally [39:47] what we've seen is that um [39:52] we we believe in this idea of this sort [39:53] of capability overhang where the the [39:56] product um you know like vast majority [39:59] of the people are getting a sliver of [40:01] the the true value that they could out [40:03] of the bottle. Um that's not necessarily [40:05] a bad thing. Many you know they're still [40:06] getting tons of value. Um but we found [40:10] that like we you know make sure to focus [40:14] on creating a extremely simple, [40:16] streamlined experience for for many many [40:20] people. Um, and then we put like uh all [40:24] of the, you know, cutting edge stuff. We [40:25] also make make sure to give that and and [40:27] make sure that's like really accessible [40:29] for the people that care. The challenge [40:31] of course is like not necessarily [40:33] overindexing and and sort of like [40:35] putting that putting uh something that [40:38] um people might not understand in front [40:40] of a billion users. So, but generally [40:42] what we've seen is we can experiment [40:44] early with the the people that really [40:46] care and that's where we have like our [40:47] desktop app and we have codecs and now [40:50] chat work is kind of stepping into a [40:52] much broader audience which we're really [40:54] excited about and eventually what we [40:56] want what we what we know and we think [40:58] about this a lot in design is like it [41:00] will tr like things will truly become [41:03] like for billions of users when they [41:06] don't have to like think about a switch [41:07] or a mode or anything like that. And so [41:09] we know that like eventually it has to [41:10] kind of get distilled into that main [41:12] experience. Um but it is like a really [41:15] fun thing to balance and um we have to [41:18] think about like people that are using [41:20] this for free every day to people that [41:22] are paying us hundreds of dollars and [41:24] lots of different needs. It's just a [41:26] matter of sort of balancing and as a [41:28] designer it does add a lot of complexity [41:30] for sure to think to you really have to [41:33] have empathy for every sort of different [41:35] type of user. Um, but this it it makes [41:38] it fun for sure. [41:39] >> That is such a mind-blowing point you're [41:40] making about the spectrum of usage of [41:44] chat GBT and CEX. I I didn't think about [41:46] it that way. Like is there has there [41:47] been a product that has such a wide [41:49] spectrum from just like what's this rash [41:52] on my finger to building Salesforce, you [41:56] know? [41:56] >> Yeah. And it and it makes it really hard [41:58] to you know because I a lot of uh it's [42:00] so funny like one of the big challenges [42:02] we're always thinking about is this like [42:03] blank box problem, right? And people [42:06] talk about this all the time. It's like [42:08] is chat really the final frontier? It's [42:11] a terminal bas it's you know glorified [42:13] terminal is like some critique that that [42:16] I think these chatbots get. And then [42:18] it's like okay so how do you design [42:21] something that can kind of shapeshift [42:24] into anything? um you know it's like [42:27] maybe someone is coming here for [42:30] image generation for fun or you know [42:33] they're trying to generate a PowerPoint [42:35] or you know so having this adaptable [42:38] interface is like a really interesting [42:40] uh design challenge and um it's like [42:43] sort of uncharted territory because it [42:44] can really like shape shift into many [42:47] different tools um and it's this like um [42:51] incredible at so many different things [42:53] and trying to figure out how to present [42:55] that. And then thinking about, you know, [42:58] we think a lot about like a novice user, [43:01] someone who's sort of brand new to this [43:02] experience that we want to help them [43:05] understand how this can add value uh to [43:08] their lives or to their sort of [43:10] professional uh work or what whatever [43:13] whatever capacity they're using this in. [43:15] And then we also want to like get out of [43:17] the way of people that, you know, use [43:19] this every single day and they use this [43:21] as like a real workhorse tool. And so [43:24] designing for that spectrum is something [43:26] we have to think about a lot. Um but [43:28] yeah, it's all a world of spectrums I [43:31] guess. [43:31] >> There's a comment that Kevin Wheel, your [43:33] former chief product officer, made on [43:35] the podcast that has really stuck with [43:36] me, which is uh to your point, people [43:38] always diss this idea of just like it's [43:40] just like a chat box. Why hasn't it ever [43:41] changed from that? Even though it has, [43:43] but still it's kind of this chat [43:45] chatbot. and he made this really good [43:47] point that what is an interface that is [43:49] very good at uh spanning a wide spectrum [43:51] of intelligence. [43:53] Uh it's actually just talking. If you [43:55] think about humans, we talk to people of [43:57] low IQ to very high IQ. We can all just [43:59] talk to each other and it's actually a [44:02] really good interface for uh packaging [44:05] something that is increasingly smarter [44:06] and smarter and smarter. [44:08] >> Yeah. And and I think we're seeing [44:10] what's what's been really fun design [44:12] space for us is you know there's two [44:16] things we think about. One is like how [44:17] do we how can we help you express your [44:20] intent better, right? Like uh we don't [44:23] want you have to think about being a [44:25] perfect prompt engineer, but like how do [44:27] we even one teach you that oh this is [44:29] something that the model can help me [44:30] with and then how do we make sure that [44:31] we get we we sort of best understand [44:34] what you want. And so like, you know, we [44:36] recently shipped some features where if [44:38] you ask sort of an open-ended question, [44:40] it can follow up with, you know, some [44:42] follow-up questions that you can like [44:43] just tap through um an answer versus [44:46] having to type everything out, right? Um [44:48] or another good example is, you know, so [44:51] then then once you actually ask it to do [44:53] something, there's the chat transcript [44:55] doesn't need to be purely text, right? [44:57] We've seen these now evolve so far from [44:59] what it was with when chat just [45:02] launched. you know, it's no longer just [45:03] a pure chatbot. It actually has many [45:06] features within that, right? So, a good [45:08] example is what we call writing blocks. [45:09] You ask it to um you know, write you a [45:13] letter or an email uh and it now [45:17] contains that um that email in sort of a [45:20] block that you can directly manipulate [45:21] and you can type with it and you can [45:22] brainstorm and you can edit little bits [45:24] and parts of it. You can copy just that [45:26] and that has made that interaction much [45:29] better. And we're seeing the same thing [45:30] with like image generation and creating [45:32] special tools for that kind of mode. And [45:34] I think that's where the UI is going to [45:36] start to really continue to evolve is [45:39] how do we make if you're if you come at [45:41] this and you're a designer, maybe chatbt [45:44] should actually really look differently [45:45] uh look different or have different [45:47] affordances. [45:49] um or you're you know a a data scientist [45:53] or you know you're using this for to run [45:55] your home you know like your your [45:57] personal life maybe it should really [46:00] kind of work or look differently so like [46:02] understanding the context and giving you [46:04] different affordances based on that is [46:06] something that you know we're thinking a [46:07] lot about [46:08] >> in terms of the vision and future of [46:10] where all this goes I know this is [46:12] always evolving but is there kind of a [46:14] sense of what the future of chat gpt [46:16] codeex is I know there's this idea of a [46:18] super app. You talked about voice, this [46:20] personal assistant idea. What's like a [46:22] way you think about where this goes long [46:24] term? [46:25] >> I think that it will start to become [46:30] more and more adaptive to, [46:34] you know, all of the context of your [46:37] life. Um, I think proactivity is a big [46:42] untapped kind of part of the experience [46:44] that we're really excited to think [46:45] about. Um, you know, we're starting to [46:48] see this with chatd work now where it [46:51] actually has pretty good suggestions [46:53] based on, you know, your calendar or [46:55] your slack or these different sort of [46:57] inputs and um external kind of context. [47:02] I think that's going to become more and [47:04] more important um or kind of shift the [47:07] way you know you use it. I think voice [47:09] as you know we we we just had a step [47:11] change in the model that just feels much [47:14] more natural and again like you know I [47:17] can only imagine how that's going to [47:18] feel in 2 3 years as it just gets better [47:21] and better. I think people I I do love [47:24] interacting with chatbt especially voice [47:26] in like just I want to just kind of like [47:28] really quickly say something and and [47:31] then I think like its ability to kind of [47:33] go sort of like not just from asking [47:36] questions but actually doing things for [47:37] you. Um that's all going to change and [47:41] hopefully all of this gets simpler as we [47:43] do it too right you know when we think [47:45] of the super app I think eventually we [47:48] really think of the idea of especially [47:50] for our broad consumer base we do think [47:53] that there should be sort of this one [47:55] universal input that you can ask that [47:58] knows when it needs to answer like [47:59] really quickly or it should be chatty or [48:02] whether it should go and like go off and [48:04] do work for you. And so that's a big [48:06] like excitement that we can get to is [48:08] how can we continue to simplify the [48:10] experience such that you shouldn't have [48:12] to think about how this works. You [48:14] shouldn't have to think about what model [48:15] you're on or what sort of speed you're [48:18] using. You know that like that's like [48:20] goes back to the balancing the different [48:22] kind of you know personas and user needs [48:24] like right of course we want to expose [48:27] that for the right people but then we [48:29] want to make sure that we can get a [48:30] default experience that works great for [48:32] for most people. Um, but yeah, I think [48:34] it's going to be really proactive. I do [48:36] think that we're going to like get much [48:38] uh richer sort of input through voice [48:41] and like richer visual outputs that are [48:44] more interactive. I think that you'll [48:47] start to build more durable kind of [48:50] workflows with attribut that you can [48:52] kind of use over and over. Right now, [48:54] you know, it a lot of times it feels [48:55] like you're coming and starting from [48:56] scratch each time. and we want to really [48:59] make this kind of the tool that you can [49:00] use to to help you uh with whatever sort [49:03] of task you have. [49:04] >> This all sounds wonderful. I can't wait. [49:06] [laughter] [49:07] >> Uh just kind of zooming out a little bit [49:09] and then I want to talk a little bit to [49:11] designers again, but you're in this uh [49:13] epic role right now. I imagine there's a [49:16] lot of surprises as you entered OpenAI [49:18] and took on this role. What's something [49:20] you wish you'd known? What's something [49:21] you wish you could whisper into Ian of [49:23] of day one and to give you some advice? [49:26] I mean so many things. [49:29] I think you know the first one is like [49:32] recognizing that we're really early and [49:34] instead like I actually was pretty [49:36] intimidated coming to OpenAI if I'm [49:38] being honest. Um, you know, uh, I [49:41] remember like, okay, interviewing. Oh, [49:43] no, I got to like, you know, G GPT4 had [49:47] kind of recently come out and what GPT, [49:50] like I, you know, I was like, what is [49:52] all this stuff? Like, I I was using it [49:54] and I was using the the stuff, but I I [49:56] didn't feel like I had a true [49:57] understanding of of it. And to be [50:00] honest, I, you know, there's I still [50:02] have so much to learn. So that idea of [50:05] like you are early, we're we're figuring [50:07] this out like that I think I would I [50:10] would want to hear. And the other thing [50:12] like that I found uh you know and for me [50:15] this is more of like a personal thing. [50:17] It's less about AI or anything but I [50:19] stepped into a really different role. um [50:22] you know most of my career I I've [50:24] managed uh I definitely managed before [50:26] but never like at a scale that's growing [50:28] so fast and and then at a company that [50:31] is like so different you know that it [50:35] the role was just every day was [50:37] something new uh and for a long time I [50:40] felt like I'm kind of like bad at this [50:42] or you know I don't know what I'm doing [50:46] um and then I went and like talked to [50:48] friends that had led like really large [50:50] design teams dreams or um or or were in [50:53] a similar transition as me and they all [50:56] felt exactly the same or the problems [50:58] that I had were not unique. Um and so [51:02] you know this is I guess yeah more [51:03] personal advice that it's like I think [51:05] at any stage of your career there's that [51:07] feeling of like imposttor syndrome and [51:09] like what am I doing? I don't I don't [51:12] know how to do this. Finding people you [51:13] can talk to that are going through the [51:15] similar things. you'll find that there's [51:17] usually a lot of shared kind of uh [51:19] challenges and struggles and having that [51:20] like sort of whether it's a mentor or [51:22] just like a a buddy that is like [51:24] impartial and out of it has was really [51:26] helpful to me. Um and you know finding [51:29] that community I think has been has been [51:31] helpful. Um so yeah so like that and [51:33] then understanding that this is all so [51:35] new and evolving really quickly. [51:37] >> That is extremely helpful to hear. Elena [51:41] Vera had this really great post. I don't [51:42] know if you know her. She's how to grow [51:44] to lovable right now. uh she had this [51:46] great post called stop the AI confidence [51:48] theater where she just kind of like all [51:51] these people just keep posting all these [51:52] things they've done with AI and all the [51:53] ways it's changes their life and it's [51:55] just they're killing it figured all out [51:56] and really it's like very few people [51:58] like very few of these things are [51:59] actually that great and working and and [52:02] it's just like stop stop pretending like [52:03] this is actually you figured it all out. [52:05] >> I think it's much better to be humble [52:07] about it and be like wow yeah I don't [52:09] know this is like changing quickly and [52:10] like I'm just going to embrace it and [52:12] see where it goes. Yeah, and that [52:13] connects to what you said was a core [52:15] component of people that are doing well [52:17] right now, people that are curious and [52:19] just kind of poking around and trying to [52:21] trying these things and try them again [52:22] because maybe it didn't work now. It [52:23] didn't work before, now it works now. [52:25] >> 100%. Yeah. It's like being [52:29] kind of like Yeah. Every every model, [52:31] every new every new feature, it's like [52:33] the people that are poking around at [52:35] those edges, I I found are just having [52:37] such a blast doing it, you know. And [52:39] this connects to something else Kevin [52:40] Wheel, former CPO at OpenAI, said on the [52:42] podcast, which has always stuck with me, [52:44] which is the model we have today is the [52:46] worst the model will ever be. [52:48] >> Exactly. I mean, that that goes back to [52:50] like when, you know, [52:54] we we Yeah. I mean, a new model comes [52:56] out what, every week? I don't know, [52:59] every day. Like, [53:01] >> uh, if you count all the labs and and [53:03] everything, it's like it's changing so [53:05] quickly and and everyone gets better. Um [53:09] and I don't know it's like yeah I I [53:12] always try to remind myself it's like [53:15] you know you you know it must be like [53:17] what like I don't know building or [53:19] designing and like early days of [53:21] computers when like componentry was like [53:23] really evolving really quickly and there [53:25] were these drastic step changes and um [53:29] you know like you would hit these [53:30] limitations and all of a sudden those [53:32] limitations were gone. You know, I think [53:33] it's the same, right? It's like these [53:35] things just change and they're what was [53:38] totally broken yesterday now actually [53:40] just kind of works. [53:41] >> So maybe to start to close out the [53:42] conversation, say a designer is [53:46] listening to this and they're in that [53:47] bucket of overwhelmed, anxious, [53:49] struggling, stressed, uh feeling [53:52] uncertain about their future. What's one [53:54] suggestion you have for them to help [53:56] them move out of that state and into a [53:58] better place? just recognizing that [54:02] it's it's a it's a totally rational [54:05] feeling [54:07] and it's a great time to just try to [54:09] explore and see what if I tried working [54:12] what if I tried you know think [54:14] rethinking about the the tooling or what [54:16] if I just embraced like oh maybe I [54:18] should prototype more let me try you [54:20] know um try using these tools but again [54:24] like recognizing that I maybe it helps [54:27] to hear that I think that we're all [54:29] still so early and it's and it's still [54:31] changing and and just kind of embracing [54:32] that like just have fun with it and and [54:35] see where it goes. And and I think maybe [54:38] another thing I would say is really try [54:40] to focus on the output the outcome, [54:43] right? At the end of the day, we're not [54:45] here to build [54:48] tools like processes. We're not here to [54:50] like build uh or token max or something [54:54] like that. we're here to make great [54:57] software that people love or great [54:58] products or whatever it is you do. Um, [55:01] but that's the thing that really matters [55:04] and if you focus on that, you know, I [55:07] think that you'll hopefully find that, [55:09] you know, these tools can help you do [55:11] that, but you still have a huge role in [55:13] that as as the person that's kind of [55:15] steering and directing it. [55:17] >> So, along those lines, I'm going to take [55:18] us to a couple recurring corners of the [55:20] podcast. The first corner is AI corner. [55:23] The question is just what's one way that [55:25] you have found AI to be really useful in [55:28] your job that maybe might inspire [55:30] someone to use say uh JGBT or Codex [55:34] >> as like a designer designer. I mean [55:36] nowadays I spend less time in the [55:37] pixels. Um but I really do use it to [55:41] kind of help me [55:43] think through broader ideas. Um or may [55:47] like, you know, I'll have an idea at [55:49] night or something and then I'll just [55:51] like quickly, you know, type it in and [55:53] have it like prototype something and [55:55] it'll give me like an idea that I can go [55:56] and talk to the team about and it help [55:58] visualize something. And so there that's [56:00] like a part of the design process. It's [56:02] like early ideas or we're riffing [56:05] through something and and I can I found [56:07] that chatbt whether it's CHP work or [56:10] Codex can like really kind of help me [56:12] visualize that. That's been something [56:13] that's been really fun. And then like [56:16] there's there's more of the operation [56:17] side, you know, it's like I have it now [56:20] like checking and making sure that like [56:22] important Slack uh messages that I need [56:24] to follow up with are summarized and [56:26] giving me context before um sort of like [56:30] uh you know important meetings or for [56:32] recruiting or for all like every pro [56:34] every part of my process I I I use this [56:36] to kind of augment u uh the work that I [56:39] do. [56:40] >> It makes sense. on the prototyping [56:41] piece, just to give someone a tip on how [56:42] to do this. Are you opening up on your [56:44] phone and firing off just like a codeex [56:46] thread? What's like the mechanic of [56:47] doing this? At [56:48] >> CHP work has been incredible because um [56:52] it now runs fully in the cloud, right? [56:54] And so the idea is you can kind of just [56:57] be anywhere and sort of like talk to [57:00] chat and have it like really spend the [57:05] time and and come back with like a [57:07] really great artifact. So, I found that [57:09] it's been really fun to just be like, [57:11] "Hey, you know, we're working on I don't [57:13] know whatever um this thing is like [57:16] whether it's helped me think through it [57:18] or um you know, yeah, sometimes I will [57:21] just have like some ideas that I would [57:22] love to see like quickly visualize [57:24] before I even like take it further and [57:26] it will just like kind of help me like [57:28] play with it and try it and stuff." And [57:30] so that's been really fun. The fact that [57:32] it's so accessible, right? um like I [57:34] might not even be near a computer or I [57:36] can you know um I don't know just kind [57:40] of like have that that flexibility I [57:42] think is really fun. [57:43] >> Awesome. Okay, I'm going to take us to [57:45] another corner uh fail corner. [57:47] >> Yeah. [57:48] >> So the reason I do this a lot of people [57:51] awesome awesome people like yourself [57:52] come on the podcast and all people hear [57:54] is all the amazing things they've done [57:56] all the ways they've been really [57:57] successful and they don't hear all the [57:59] times things didn't work out. And I feel [58:02] like that's very important for people to [58:04] hear. It's not always up until the right [58:05] and success. What's a story of a time [58:08] you failed in your career and and what [58:11] did you learn from that? [58:12] >> I mean there are so many times um [58:16] I think that [58:19] I mean even at OpenAI it's like I I [58:21] think that I found that I've had to [58:24] learn every step of the way, right? Um, [58:26] you know, when I first uh joined, I [58:29] remember, you know, not really even like [58:32] fully understanding like how I wanted to [58:34] build a design team and, you know, um, [58:38] not understanding how to hire or [58:39] whatever. Um, I mean, just I I sort of [58:43] felt like I was kind of failing every [58:45] day, but just kind of like quickly [58:48] trying to adjust and learn. Um, I think [58:52] that that uh I I don't know. I mean, [58:55] that's like maybe kind of a copout [58:57] answer, but like really just feeling [58:58] like every day I'm kind of like learning [59:01] on the go. [laughter] I I feel that my [59:04] whole career has sort of been that way [59:05] and that I have never really kind of [59:07] like planned like any any one step and [59:09] I've in and sort of like quickly [59:11] learning from your mistakes. I've [59:13] shipped plenty of things that did not [59:16] work out. Um and uh I've also shipped [59:19] things that you know you know have [59:22] worked out but like the most many times [59:24] I I've shipped or you know you try [59:26] something and it's kind of a failure and [59:29] uh you know at Instagram we had flops [59:32] and then we you know we'd like quickly [59:33] move on. I mean, I worked on IGTV which [59:35] was like a huge flop and then we shipped [59:37] reals uh and reals was like we and and I [59:41] mean that's a good example cuz like we [59:43] had we we just like kind of baked in [59:45] some wrong assumptions and we held on to [59:47] a few constraints but then we learned [59:49] from that quickly and we were like okay [59:51] actually let's change that and then we [59:54] sort of redid it and then it became a [59:55] really successful product. Um and uh you [59:59] know I I've seen that many cases many [60:02] cases of like and that's helped I think [60:04] that's helped harden me to the sort of [60:08] critique and feedback of things because [60:10] you know sometimes you get it wrong [60:12] sometimes you try something and [60:14] sometimes it is just about having to [60:15] like put it out in the world and work [60:17] through it and um and get the feedback [60:20] and and iterate and I think that that's [60:22] like the mo more importantly is not sort [60:25] of what you shipped one time but how you [60:27] react and how you um you know how you [60:29] kind of like fix forward and how you [60:31] continue to iterate and listen to listen [60:33] to your users. [60:34] >> This is a very common theme in this [60:35] corner is just people's careers are like [60:37] it's mostly failure along the way. [60:39] [laughter] [60:39] >> Yeah, definitely. [60:40] >> And so I think that's helpful for people [60:42] to hear. [60:43] >> Yeah. [60:44] >> Uh okay, before we get to a very [60:46] exciting lightning round, is there [60:47] anything else you wanted to share that [60:48] we haven't covered? If I could distill [60:51] it, it's like I hope that this is just [60:53] like a really fun time to build build [60:56] stuff. And I hope that we, you know, as [61:00] designers can be really adaptive and [61:04] focus on making really fun, making [61:07] really great stuff. And I and I and that [61:09] I think that designers are just like [61:12] hope, you know, I think I think great [61:14] companies are are seeking designers and [61:17] seeking great great design. And I found [61:19] that to be really true. There probably [61:21] going to be a bunch of companies that [61:22] will be left behind because they haven't [61:23] shifted, but I think that that there's [61:26] plenty of room for amazing, great [61:28] design, and that's going to be more true [61:30] than ever. [61:30] >> I completely agree. I'm very bullish on [61:32] design. [61:33] >> Yes. [61:34] >> Well, Ian, with that, we've reached our [61:36] very exciting lightning round. I've got [61:37] five questions for you. Are you ready? [61:39] >> Okay. Yes. [61:40] >> All right. Here we go. What are two or [61:43] three books that you find yourself [61:44] recommending most to other people? [61:46] >> I'm not like, you know, it's funny. I'm [61:48] not like a big big book recommener. Um [61:52] I think like when people ask me uh I do [61:54] get asked about like books for design [61:56] and I'm not I haven't found that many [61:57] that really have stuck with me like [62:00] purely design. Although I will say I [62:02] have I do love the design of everyday [62:04] things. I think that's like such a [62:06] classic. [62:07] >> That's the one that's the one that you [62:09] know it's not like yeah it's not a [62:11] unique recommendation. Um, but it it is [62:15] just kind of a classic of thinking about [62:17] how you how we think how I think about [62:19] design. Yeah, [62:19] >> you have to read it if you're in tech or [62:21] not in tech. [62:22] >> Yeah. [62:24] >> Awesome. Uh, anything else there or [62:26] should we keep moving? [62:27] >> We can keep moving. Yeah, [62:28] >> let's do it. Favorite recent movie or TV [62:31] show that you have really enjoyed? [62:33] >> Okay, so I have one that actually is not [62:35] a movie or TV show, but it's a I I'm in [62:38] New York. I just saw a Broadway show [62:41] >> um called Maybe Happy Ending. [62:44] >> Yes. I've seen it. Love it so much. Yes. [62:47] >> Oh, you saw it? Okay. Yeah. It was so [62:50] unique, right? Um and of course it kind [62:53] of hit home a little bit because it's [62:55] about these like super intelligent [62:57] humanoid uh robots and like you know um [63:01] but um but the the production was [63:04] incredible. the set uh design, [63:08] everything about it I thought was like [63:09] so unique. Um which was um I thought [63:13] Yeah. So if you have the chance to see [63:15] it, I thought it was really really good. [63:17] Um I also just saw the invite which was [63:20] I thought and [laughter] have you seen [63:22] it? [63:23] >> Yes. Also love it. So simple but so [63:26] good. [63:26] >> It that's what I loved about it is it's [63:28] like so simple. It's like it actually is [63:31] a like a play, you know, and it's just [63:34] four people in one kind of beautiful [63:37] apartment and the set design is amazing. [63:40] Um, but like the writing, act, [63:42] everything about it was like incredible [63:44] and it's just kind of a wild ride [63:47] [laughter] [63:47] and I went in knowing nothing about it. [63:50] I thought it was incredible. [63:52] >> Excellent taste, sir. I love both those [63:54] so much. Uh, and the invite just came [63:56] out. It's in theaters right now. [63:57] >> Yeah. Next question. Favorite product [64:00] you've recently discovered that you [64:01] really love. I I feel kind of like a [64:03] typical Silicon Valley person [64:05] recommending this one or something. But [64:08] um I did finally switch to an EV after [64:12] having uh many cars for a long time and [64:15] I got a Rivian and um I don't know they [64:19] just I do respect the design a lot of [64:22] just kind of how they've really made it [64:24] just [64:26] all of the they like I think if you know [64:28] probably Tesla's similar but like [64:30] there's a lot of assumptions that how [64:33] get operating a car should work have [64:35] gone away. You just walk up to it, it [64:37] unlocks, you know? It's like you just [64:39] like you don't have to like turn on and [64:42] off the car. And it's cool to see like [64:44] that evolving. Of course, I love, you [64:46] know, at Whimos are probably if if I [64:48] could like own a Whimo right now that [64:51] that is the most magical experience in [64:53] the world. Um, and uh I I'm waiting on [64:59] baited breath for that to to be broadly [65:01] accessible for everybody. Do you have a [65:03] favorite life motto you find yourself [65:05] coming back to in work or in life? [65:07] >> Not necessarily other than I think it [65:09] like kind of trying to sum up like how I [65:11] felt about all of this is like have fun [65:14] with it and like just try stuff. Um and [65:18] you know I think don't feel like I've [65:22] never felt like you have to fit into a [65:23] specific box. Um, and know that it's [65:27] perfectly all right to be very fluid [65:29] about the way you work or the way you [65:31] think about things and um, I don't know, [65:33] be flexible and adaptable. But yeah, [65:36] it's not quite a life motto. More of a [65:38] vibe of just trying to like have fun [65:40] with it and and build great stuff. [65:42] >> I I I should shift this question to [65:44] what's just like the vibe. What's [65:45] [laughter] the vibe that's been helpful [65:46] to you? Because I love that. It's very a [65:48] Sam Alman in terms of just you can just [65:50] build stuff. [65:50] >> Exactly. Yeah. [65:52] >> Okay. Last question. Something I love [65:54] about your career history. If you look [65:56] at your LinkedIn, you were a designer at [65:57] Groupon. I love that you are at Groupon [65:59] and now you're heading design for [66:01] essentially super intelligence. [66:02] [laughter] Uh what's something you [66:04] learned at Groupon that is is being is [66:06] maybe helpful these days building [66:08] >> super intelligent AI? [66:10] >> I love that. Groupon was a really fun [66:13] company. Um, [66:16] Andrew Mason was like a really [66:22] incredible CEO and just that he brought [66:24] such character [66:27] uh to the product, you know, it was like [66:30] kind of a brandled product. Um, and um, [66:35] you know, like [66:38] we we had a there was like a a comedy [66:40] team. Uh there was like a writer team at [66:43] at at but they were like all comedians [66:46] basically and the product really did not [66:49] take itself seriously at least in its [66:50] early early couple years. It was like it [66:53] was it didn't take itself seriously and [66:55] I think that the brand really stuck out, [66:56] right? You weren't signing up for [66:58] Groupon because NEC I mean of course the [67:01] daily deal was great. You were getting [67:02] like 50% off something but people loved [67:06] that email because it was like really [67:08] well written and it was funny and it was [67:10] and it had character and I thought that [67:13] like the early on the company did a [67:15] really good job with with brand and sort [67:17] of embracing uh that side of things. Um [67:21] and and I think the other thing if I [67:23] were to carry that forward is I found [67:25] that every company that I've worked for [67:28] has such unique DNA and the way you have [67:31] to work kind of maybe changes a little [67:32] bit based on that right. So I you know [67:35] Groupon was run one way you know then I [67:37] went to Instagram which was like this [67:39] very design and and creativity focused [67:42] and Kevin's you know Kevin Cyrum was you [67:44] know really like an incredible designer [67:47] and he he sort of saw himself or like [67:51] worked in that way with us which was [67:53] really cool and it just created a [67:54] different way of working and then you [67:56] have like Sam Alman who's like a [67:59] completely different way of operating [68:01] and he's like uh you know I [68:04] you know, and then and open being a [68:06] research lab. And so learning that [68:08] companies have these really different [68:10] DNAs and understanding how to sort of [68:12] embrace them and kind of leverage that [68:14] to your uh like or help kind of like [68:16] empower that or like I guess lean into [68:18] that um has been something that I've [68:21] learned a lot um from moving around and [68:23] getting the opportunity to work at very [68:25] very different companies. [68:26] >> And also just to add to that, it's also [68:28] important to realize you're not going to [68:29] change that. you're not gonna like come [68:31] in and change that culture because [68:32] that's just that's very hard especially [68:34] if it's very founder oriented. [68:36] >> I think it's about finding like oh wow [68:38] like at OpenAI you know yeah sure like [68:42] we spent more time at Instagram we [68:44] definitely moved slower and I would say [68:45] that we um we changed the product less [68:50] and that actually worked really well for [68:52] most stages uh of of Instagram. [68:57] uh but I don't think that that would [68:58] work in in the current stage of where [68:59] we're at with open AI and that way of [69:01] working. Of course, I bring a lot to [69:03] that and I think like you know at at um [69:07] at Instagram we have this motto like do [69:09] the simple thing first and we always use [69:11] that and we really embrace that. I've [69:13] tried to carry that through to the way I [69:15] think about building product at at [69:17] OpenAI. Um but that's not the only way I [69:21] think of working. You know, sometimes [69:22] it's actually not necessarily the right [69:24] way to think about. Sometimes we [69:26] actually do throw some things out that [69:28] we're still experimenting with. And so [69:29] just being really I found that having to [69:31] yeah be flexible and yeah, not [69:33] necessarily trying to change it, but try [69:34] trying to kind of like help round it [69:36] out. Maybe. [69:36] >> Ian, what a what an incredible uh place [69:39] you have in the world right now. Just [69:41] being on the inside of such cool [69:44] technology, such cutting edge stuff [69:46] that's happening. Uh, and I feel very [69:49] fortunate that you found time to chat [69:50] with us and share what you're seeing, [69:52] where things are heading. Two final [69:54] questions. Where do you want to point [69:56] people? If you want to plug anything and [69:58] how can listeners be useful to you? [70:00] >> Where do I want to point people? I mean, [70:01] I think that um please, you know, I [70:05] guess use, you know, have fun with with [70:07] the stuff we're making with JBT and [70:09] Codex and fire away feedback and [70:12] thoughts. I'm on Twitter. Uh I love to [70:15] talk with designers uh at all sort of [70:18] stages. Um or not just designers, [70:21] engineers, PMs, whatever. Um, that's [70:25] probably where I'm most uh sort of [70:26] reachable and accessible. Um, and you [70:30] know, how can you? Yeah, I think to be [70:33] useful. Again, it's like I love when [70:35] when people do show off the way they are [70:37] kind of working and how that's kind of [70:38] shifting and that's like a great way for [70:40] us to learn how we can we can build [70:42] better products both like internally at [70:45] OpenAI but just as a as a design [70:47] community like maybe showcasing a little [70:49] bit more of our process because it is [70:52] becoming a little bit more opaque like [70:53] well how is it working just um I think [70:57] sharing that is is like such a gift to [70:59] the community so doing more of that [71:01] >> also point people to your website which [71:03] I have here now. And I think my face is [71:05] probably changing colors because the [71:06] site [laughter] has all these crazy [71:08] colors. [71:09] >> So, if you go to Iansber.com, you could [71:11] see this uh insane website. That's [71:13] [laughter] your personal website and [71:14] there's a [71:15] >> Yeah, that's a deep dive into all of my [71:18] past work. You can see there. Yeah. [71:19] [laughter] [71:20] >> And the trick is you Yeah. You'll have [71:22] to discover the Easter egg where there's [71:23] a game in involved. [71:25] >> That's right. [71:25] >> Uh [71:26] >> see if you can set the high score. [71:28] >> Yeah. This high scores are out of [71:29] control. I don't know what's happening. [71:30] [laughter] Someone's found a cheat code. [71:32] Uh, Ian, thank you so much for being [71:34] here. Thank you so much for doing this. [71:36] >> Buddy, this was awesome. Super fun. [71:38] >> Same. Bye everyone. Thank you so much [71:42] for listening. If you found this [71:43] valuable, you can subscribe to the show [71:45] on Apple Podcasts, Spotify, or your [71:47] favorite podcast app. Also, please [71:49] consider giving us a rating or leaving a [71:51] review as that really helps other [71:53] listeners find the podcast. You can find [71:55] all past episodes or learn more about [71:57] the show at lennispodcast.com. [72:00] See you in the next episode.