Tommy Geoco - The state of the design industry right now

00:00:00AI fluency is going to eventually justbecome like an ambient thing. I don'tthink we're going to like highlight[music] that language forever. But Ithink right now it is an importantlanguage cuz a lot of companies arelike, "Yeah, we want this designer whocan come inand think. [music] We want someone whoknows what to do with the question whatif in a way that doesn't feel like it'sfrom 2012, 2015, [music]you know, 2018." That question seems tobe being answered by like, "What are youshowing that you're sharing?" Likesocial media is the new discovery engineright now. Like that is the way to>> [music]>> for better or worse, that is the way toprove that you are someone who isexperimenting Welcome to Dive Club. Myname is Reid and this is where designersnever stop learning. This week's episodeis with Tommy Geoco who just finishedvisiting some of today's leading designteams like Vercel, Perplexity, Ramp andwe're going to go deep into everythingthat he's seeing around AI workflows andwhere design is headed next.>> [music]>> And to start, I wanted to dig into hisrecent state of prototyping survey tofigure out what details stood out themost. The number one thing that stoodout to me was that designers who arecurrently vibe coding are more satisfiedwith their workflows. I hear a lot of noand we both hear a lot of noise, man.And

00:01:15Idid not expect that. I was very curiousto see cuz I was like, "I know this ishappening a lot. How much and are peoplehappy with it?" Cuz I assumed I mean wehear the stories likemeta incentivizing token layers>> leader boards, yeah. You know, and thislike just like being crammed down thethroats of employees. People seem to digit in these in this survey and I thoughtthat was really unique. I also thoughtand I wish I had parsed this a littledifferent, but it's the people who arecurrentlydoing the majority like so if a majorityof their workflow is spent on vibecoding activities, it's designengineers, that makes sense. Leadprincipals, okay. But then it's likenon-designer roles which might bestudents and researchers, then it'smanagers,and then it's like your general juniormid-level IC. And that part wasfascinating that managers are doing morethan junior and mid-level ICs. Eitherthings are trickling down and people areexperimenting and then they're going topass it, you know, they're going to passlearnings down, which is kind of whatwe've seen on location, but it alsomight mean that like some managers orteams haven't yet made room for the restof the team.>> I'm looking at a bunch of junior and middesigners that uh are getting cut out ofthe process as my bad data says.>> Yeah. Yeah, and and and that to me it'slike,

00:02:30you know, I want I'm trying togive a little benefit of the doubt, butultimately it's like, yeah, we need tomake more room at at companies. I don'tknow if it's trending in that way, butI'm starting to see more and more ofthat at certain types of companiesanyway. I think it's interesting tothink about the types of vibe coding,too, because the experience of somebodywho is building all of these crazyprocesses or systems or files to get atool like Figma make or even B 0 to lookand feel like their design system versussomebody who's just every day they openup the production code base and juststart exploring and shipping. That is apretty big chasm, too. Where I get a lotof people coming to me that arefrustrated because they're trying tobend these tools to their will, and thenthere's another class of designer whoreally is I mean, what's a designengineer and what's a designer if allyou do is ship designs in code thatultimately make it into the productitself? That's line. Like we don't evenhave the right language to talk aboutthese different types of roles now. Ifeel like there's a a group of designerswho aren't aware just how common I guessI wasn't either it is for designers toroll their own tooling that they'reusing in their workflow and then eitherlike tossing it to the side or likestarting to share it internally withother team members. 59% of designershave built their own like tool for theirworkflow. And

00:03:45and sometimes when I say,"Oh, all these designers are creatinginternal tools," someone will respondand say, "Yeah, it's all AI slop." AndI'm like, I think we're misunderstandingwhat we're saying here. Is that you hadsome sort of visual asset, you neededsome effect applied to it, and you wrotea tool to help you, and maybe you threwit away after. Or in the case like whenI went over to Vercel, they had this abrand designer, I can't remember hisname off the top. Really talented guy.Had never coded before, and now was vibecoding a tool their marketing team wouldlike put out blog posts. And they werelike, "Why do Why does the design teamneed to create like the OG blog postcards for every page? Like that's not agood use of it." So he built a tool thatjust allowed them to insert any sort ofimages, and it just like already had allof the branding and the sizing baked in.And they just roll these things outquickly. And I'm like, "And that justbecame a tool, an internal tool. That'scool." And so cuz it was reallyinteresting that they started referringto him as a brand engineer. And I'mlike, "What is a brand engineer?">> Yeah. And then then he showed me that,I'm like, "Okay, that kind of qualifiesit actually. That's really cool.">> I'm seeing way more people trying tohire brand engineers. I never evenconsidered that as a role recently. Thatis a thing. Brand engineers are on therise. I think every blank engineer isgoing to be on the rise.>> Sales engineer, marketing engineer. Youknow, I I feel like everyone

00:05:01iscompressing to, for better or worse,builders of some sort. I talked to WillKing, design engineer at Snowflake,yesterday. He's like, "I get it. I lovesystems thinking. I like that it's allgoing towards this. You know, we'rewe're building factories. We'reindustrializing this even further.But there is like a loss of the likefree-form creative workflow, theexploration part that some people arejust going to have to like reckon with.And it's like, "Where does that go? Doesit just get eliminated, or does itsprout up in a different area? Like doeseveryone have to be a builder?"And I just thought it was an interestingquestion. I love the building and designsystems and just like systems thinkingside a lot.But like I recognize that part is that'ssomething to reckon with.How much How much time do you spend onthat stuff? It's more that the nature ofit has changed, right? And I actually Ithink about this a lot. Part of me isalmost slightly self-conscious about it,but I do the vast vast majority of my,you know, messy explorations with AInow. Like I feel like I have made thejump to the {quote} creative directorwhere I'm just working with AI to showme a certain thing 50 different ways,and then I'm pulling the pieces that Ilike and then combining them again, andfinally I get to somewhere where I'mlike, "Yep, that's good." And then Itake that from like

00:06:16paper, run itthrough Claude code, and now it existson localhost, and then I will sweat thedetails and and actually do the likeprecision designing in code. Which isThat's crazy, man. Like that's a veryvery different workflow than I've doneat any point in my career. And sometimesI wonder, am I losing some of thecritical thinking associated with doingthings manually and hitting command Dand making tweaks myself? And yet I'mable to go through, you know, a spectrumof explorations that are so much largerthan I could do in a day, and I can doit in minutes. And I always see thingsthat I didn't plan on seeing, but that'sdifferent, right? Like it's just adifferent thing that I'm bringing to thetable. It's different muscles that I'mgrowing. It's maybe different musclesthat I'm atrophying, too, you know? Idon't know. I'm kind of TBD on I thinkwhat the impact of that is for me evenas a professional yet. Real quickmessage and then we can jump back intoit. People are getting answers fromtools like ChatGPT and Perplexity a lotmore than they're clicking on Googlelately. So, if your site [music] isn'tAI readable, then you're losingvisibility that you can't see in youranalytics. So, Framer has just launcheda free AEO tool, which shows you howthese AI systems read and understandyour website.

00:07:32So, in 10 seconds it canscan your site, score it on AEOreadiness, surface what's hurting yourvisibility. And I ran it on the DiveClub website, and I only got a 50 out of100, but thankfully Framer gives me theexact steps that I need to take to fixit. So, definitely take a minute to useFramer's new AEO scanner on yourwebsite. You can head to framer.com/aeo[music] to get started.Okay, so get this. Jitter's latestrelease lets you build your own creativetools inside of Jitter to make customanimations and effects. Just describe[music] anything you want and thenJitter will give you custom parametersthat you can use to tune every singledetail, whether it's 3D control,dithering, or even an old VHS filter.And then once you get the effect whereyou want it, you can save it, share itwith your team, and easily reuse itacross projects. And it really is assimple as just typing out your ideas,which is perfect for someone like methat doesn't have this rich backgroundin animation. And sometimes I even liketo have Claude to help me with theprompts. This feature has blown the roofoff of my creativity and I cannot waitto see what you make. Just head todive.club/jitterto get started. Now, on to the episode.I started

00:08:47[music] something 2 weeks agowhere I'm I'm forcing myself to have 1day of no AI.No way.>> Where like I I know tasking because I'vebeen building these like open clawsystems, it's really nice to get thecontext inside of that of like my intentand there's like inputs everywhere. Whatam I thinking? What's What am I workingon? What did I vote was worth myattention on Twitter this conversation.And And it's so I'm trying to figure outall these ways to get like reallymeaningful Tommy's worldview contextinto the system. But then I'm like,okay, it's really nice cuz my outputsare really rich from this thing and thenI start to become a crutch. So, no AIallows me to really think through thingsagain and then also I leave the day andI have like more context now that waslike truly authored by me and I can feedit back. So, it it kind of like twopurposes and I'm seeing how youdefinitely feel it. Like you when youdon't have it, the productivity feelsreally different. I feel stunted atfirst, for sure. Yeah, I think thebiggest difference for me is I don't donear as much thinking from a blankcanvas. My thinking would be enoughthinking to give AI a pretty good set ofguardrails for it to explore in thedirections that I want it to go.

00:10:03Butlike sometimes if I'm going to design anew layout or component or section of awebsite, I will just run a prompt to getAI to go do a bunch of things on thecanvas and I won't think criticallyabout that design until I see something.I'll see 20 things and then I'll belike, oh, that's interesting. Oh, thatmaybe we should do this instead. Butlike that's when my my critical thinkingstarts at the point where I'm alreadylooking at something rather than blankcanvas, maybe I'm typing words, drawingarrows, doing rough little sketches.Like that doesn't currently have a rolein my process, which is, you know, it'sit's just different. [clears throat]It's different. You have this likespectrum of probabilistic todeterministic design and I really likeworking over here. Like here's aquestion I have for you because it's I'mreally interested. We see we see so muchnoise online and what's been really niceis to just ask people like, what's youractual workflow like right now? And youshared something recently that was youhad made like these buttons, but it waslike the AI had made these buttons, butit was like in the perfect design systemof InFlight. And so what was the processfor you of like what inputs werecollected to the point where like youcompletely were out of the process andit just like spun those up? It actuallyties back to a single Claude sessionthat I had maybe a few months ago, where

00:11:18like most Claude sessions, I started offsaying, I have no idea what to do here.Like I feel like I should be usingmarkdown files and passing contextthrough my codebase more effectively. Ifeel like you should be reading rulesets and gathering context moreeffectively given a prompt and I wasn'tdoing anything and I was like, what do Ido? And I actually went out and I founda bunch ofuh like Dan Shipper articles and a bunchof every stuff and I and tweets that Ihad bookmarked and I just fed a bunch ofstuff in here. I was like, consume this,help me think about a plan. I've no ideawhat to do. And it created this chart ofnested uh you know, context, you know,Claude MD files throughout the code baseat different levels that would teach ithow to achieve certain things and one ofthose is a set of uh visual rules in howI like components to be built. And to behonest, I've never even opened the file.I don't actually know what's in there. Ijust regularly contribute to it. Like ifI get an output that I don't like, I'mlike, yo, never do that again. This iswhat I want instead. And then it justgoesand it does something and it goessomewhere>> [laughter]>> and I never look at it. But I've donethat enough now where Claude will know,okay, like if I'm going to build asubtle interactive element, Fredprobably wants me to use this tertiarystyle with

00:12:33this hover overlay like blendthat adds a little bit transparency ontop of the gradient. And I just get thatout of the box now and I get all mylittle inner shadows out of the box now.I never get things with borders anymore.I only get things with like using innershadows to simulate borders. And it justtookone pretty dedicated session. Like Iworked with Claude for probably 2 hourson this set of files and just refiningit and feeding context into the system.[clears throat] And then from there,it's just taking a beat to tell it thatsomething sucked and explaining why andthen saying, never let this happenagain. And I just have done that like100 times. I really think that sort ofblack box, cuz I do the exact samething. Like I have not looked at much ofmy code in a long time and my openClaude setup I is very much like I knowit's matching the intent based on theoutputs. I'm not going to read amarkdown file. Who has time for that?>> [laughter]>> But I mean, it's it's like a really fastsatisfying loop. And so when I hearpeople say GPT 5.5 is worse than Opus4.7 and I'm like, what does that mean?You know, like what are you comparing?Are you just saying like you one shotteda thing andthat probabilistic output that you gotfrom one verse the other on that givenday was better or worse? Cuz that's nota very good way to evaluate it. Like howdoes it drive?

00:13:48You know, like likethere's this sense of like you have todrive them for a little while. When youdrive it with intent and you're justlike screwing into like that's wrongfix, that's wrong fix and you do that afew iterations. I think then you have agood feel for like how it drives.So I've never understood this like oneshot mentality of like I compared threeof the same prompts. It's like you justshot like a bunch of probabilisticstuff. I hate those tweets, man. I hatethose tweets. It's like that's not evendesign. I get how that would be reallyappealing for somebody who's not adesigner, but the fact that so much ofthe discourse from professionaldesigners exists around which models putthings out of the box in a more prettyway or which models have less purplegradients. Like what are we doing here?AI gives me a level of precision thatI've never had in my career and yet wealways are quick to say like well AIslaps. Like what what? This isn't AIslap. I'm making micro decisions that Ididn't have a grid for a year ago andit's only because of the abilities thatAI gives me. It's really hard to talkabout right now. You have to really talkto somebody about like how are you usingit? How are you evaluating it? Oh,you've gone down a rabbit hole that likehasn't like that's a dead end. Okay,let's pull you out of the rabbit holefirst and like make sure we understandlike here's how you drive. All right,now that we understand like what drivingmeans, now let's talk about like some ofthe other things that like are tertiary

00:15:04to it that can inform. Because there isstarting to emerge I think a right wayto talk about I know designers hatetalking about like using tools, but likethat's the that's the moment we're in.How do you use these things? How do youinstall them? And so for me it's likehow do I talk about agents right now?It's this like headless design concept,I think. I think everything that youjust described on how you created thatbutton, these documents, this intent.Like, I don't think there's the futurewhere we're going from tool to tool andreinventing that over and over. All thehours spent just like screwing in ourintent the correct way is going tobecome just a briefcase you carry fromone tool to the other. You know, likelike Ramp's glass. They they realizedthey needed it for their company, sothey just built it internally.They shipped it this way, you know,someone who'swho doesn't have this like baselineeducation on how any of these work. It'salready on their computer and it justliterally onboards them. Cool, you'renow connected to Workday. You'reconnected to whatever data sources youneed in a secure environment. Now justlike you you have you have yourbriefcase of context. Use it whereveryou need to. I think it's going to bethe same in in so many contexts likedesign. That's going to be true fordesign. I feel like actually InFlightwould be a a great use case to this ifyou guys decide to go that way. I I wasyou're you're in my iMessageconversation

00:16:19right now.>> [laughter]>> I I think I think also uhI think like households eventually willhave this, right? Like I was talking toyou I think at one point about this ideaI had. Like, you know, you call aproduct like Hearth or something. It'sjust the family's context layer. It'slike your Life360 for anyone who doesn'tknow like which tracks your children ontheir phone cuz as parents you want tomake sure they are where they say theyare. Or it tracks how fast they'redriving cuz if you have a kid who'sdriving, you definitely want to makesure they're staying on the speed limit.You might have that data. You might havethe family calendar. You might have likeall these things. And so now a familymember doesn't have to vibe code theirown agentic setup, but they have a toolthey have this context layer that likewhen when all these tools becomeconsumer friendly, they can just likeconnect it to it and it has context. Ithink that's going to be in everyvertical. I've got to imagine. Yeah, Ithink about that a lot cuz like I havevisual language context right now. Andthen there's context from the codearchitecture or some there's somebusiness logic in there. I also havesome running brainstorm context fromClaude. Like Claude understands whatInflight is, but it's not it's notalways amazing to, you know, hook in andpull in get everything or maybe it's alittle bit out of date or it's liketalking about this idea that I thoughtwas cool six months ago that I don'treally think about that much more. I'mreally interested in the idea of likeyou tee up Inflight,

00:17:34but gosh, wheneverI'm riffing with Claude, I should beable to give it all of the feedback I'vereceived as context. I it should knowthat my co-founder Kyle always hates onthe lack of negative space in mydesigns. He always thinks it's toodense, you know? Like that should exist,you know, that should exist there or itshould if I'm designing a certainsurface area, everything's going to begoing through Claude or, you know,Codex, your tool of choice, it shouldn'tmatter, right? Like I love yourbriefcase analogy, but like that modelshould be able to seamlessly tap intoall of the decisions that were madeabout that area of the product a yearago. Right. And it should be no copyingand pasting from you. It should just beable to get everything, you know? Likethe design brain, you know, like the thebrain YC tweet kind of went viralyesterday, actually. I have, you know,mixed opinions on the viability of thatat a large scale, but when you'relooking at a single vertical, maybe it'sa family, maybe it's design context,that's going to be a huge part of how wework. Yeah, I think the general purposetool might have a place eventually. Idon't think that's the right answerright now. Maybe I won't just likeoutright say it's not the right answer,but it's not interesting to me.Um and and one of the things I askedDiego at Ramp was like, why did you guysbuild Glass? Like why didn't you justuse one of the general purpose toolsthat are

00:18:49out there that like Co-work orsomething, you know? And he's like,because Co-work is good at generalpurpose, Glass is good at Ramp. And it'slike that that context I think is justreally important piece that a lot ofpeople haven't experienced what thatmeans. And so it's like I have my openClaude and it's really good at Tommy.Like it's very it's not like, oh, ittalks in Tommy's voice. It's like it'sreally good at everything Tommy.And when you have that, it's it's sodifferent than just using ChatGPT orlike Claude code for something. It'sreally, like you said, like if if itcould be like, "Hey, we did it this way,but you know, Kyle's not going to lovethis, and you had a customer call lastweek that, you know, they called outsomething kind of interesting. Mightwant to flag that that's an edge case,you know?" That's pretty wild. The otherthing, too, is, you know, we talk aboutthis, and immediately I think people areso binary about this. It's either ahandcrafted human offered or fullyautomated. think that that has to be thecase, and I don't like when things arefully automated. There's people who talkabout like, "Oh, I got my agents, and Igot these processes to run all nightlong." That's my nightmare. Like, don'tdo that. You know, like I could see thatbeing interesting if you have like areally specific, well-outlined microtask that needs like to be iterated on.But I want the friction. Like, there isa preserved level of friction that I'mfinding is kind of important to me, sothat I

00:20:04make sure, like, let mefact-check intent. Let me like re-steer.You know, you you drove like I fellasleep, you drove in the wrong lane.Let's like get you back over here. Iwant to make sure I have the opportunityto do that more. And those lanes, likeit's a really fun process. Like, I atleast for me personally, I don't feellike I'm losing myself or like replacingmyself. I feel like I'm very muchsteering and driving. And then the otherpiece with that briefcase analogyis while it's there, on demand, you havecustomer insights potentially. And youdon't have to like go in and find them.But who's to say you don't that's notjust like you just open the briefcase.Go if you want to, like go fact-checksome stuff. Get in there and get yourhands dirty. It just There's just aprogressive disclosure problem, right?And I And I feel like there's thisbinary thinking that likeAI means fully automated. And it's like,"No, in fact, I think some of the betterworkflows are the ones that aren't fullyautomated." Yeah, agreed. Are you usinglike What are you the What's your stack?Go ahead.>> [laughter]>> That's not fair. I was going to ask youLet's We'll trade really quick. We'lltrade really quick. But I'm going to askthe question slightly differently. And Iwant you to pretend that a, you know,legacy org that hasn't adopted AI hasbeen watching your YouTube videos andthey're like, all right, this person'ssuper plugged in, which you are. Theybring you in, you're like a series

00:21:19Aconsultant effectively and they're like,hey Tommy, we want to save money ontooling, so we are getting annualcontracts. What are the annual contractsthat we should commit to as a designorg? Oh. Oh, man.Yeah.That's hard. That is a hard questionbecause the one luxury I have is I don'thave the security issues, the securitykind of compliance problems that I thinka lot of these enterprise companieshave. Let's imagine that, you know, SOC2 even isn't a requirement. Which Ithink a lot of the company, I think alot of these tools, if they don't haveSOC 2, they've got to be like right onthe horizon, right? So like let'simagine it's like mid-level securityrequirements. You really can lean intowhat you think what tools empower an organd shape an org in a way that you wouldbe happy with. Right. Well, for thework, for design work, this is what I'mseeing and what I would recommend. Firstfrom a high level, like if you thinklike sort of primitives, you havecanvas, I canvases I think are alwaysgoing to have a place in this workflow.What's your canvas tool? I think youneed some sort of code intention tool. Ilike Codex a lot. I really like theCursor Codex new UI pattern for agents.You need

00:22:34to have like an agentic codingtool inside of this process and then Ithink you need your briefcase, you needyour context, you need the thing thatingests everything and it's like a backend to to the the two types of toolsthat you're using. And then you need thesurface area where the work lives. Andthat maybe that's like Linear, right?Maybe you're like an Atlassian personand they have an agentic tool there, Ican't remember what it's called. So it'slike you need you need the work surfacearea where where the work, the the oftruth of work lives. For us, when avideo is done, the source of truth isnot like the script we wrote in Notion.The source of truth is like where thatthing gets uploaded in our back end,which might be like frame.io, right?Like when if it's in frame.io, we knowit's done. That's like a place where ifit gets uploaded there, it could becomea lot of different artifacts. It couldbecome a task to upload on YouTube. Itcould become a description we need towrite. So, you you need somewhere wherelike you guys know like the work ishappening, the intention of the work,and then where the work is done. So, Ithink likea linear a tool like that. And then youneed the interface. You need like Slack.I really hope Slack is paying attentionto this becausetalking with agents through Slack, ithas a lot of areas where I think thatprocess could be better. Talking withagents is very different cuz I'll scrollthrough like seven different chatinstances with my one

00:23:50agent. And it'sjust it's it's kind of messy, and youcan like lose context easy. So, thereneeds to be a better tool for that. Butright now, I think Slack is still prettygood. The way I've seen people using itin Slack is good. So, so you need likean interface surface area. You needwhere the work lives surface area. Youneed a canvas for more deterministicwork. And you need your agent or yourlike interface for codingfor more probabilistic work. You know,when I think about a canvas, there'sthree that are on my radar right now.This will get me in trouble. I stillthink Figma is in the game. A lot ofeven a lot of these like agentic teamsthat are like super forward-thinking,like they're using Figma a lot still.Like this is still a big part of the theequation.>> Totally. And as long as Figma cancontinues to innovate with these theagentic side considered,I would consider them umin.Paper is Paper didn't show up in ourlike top 30 tools on this survey, whichit was wild to me because everyconversation I'm having, I'm hearing,"Oh, and we're we're playing with Paperright now." Like I'm hearing Paper popup more than any of the other new toolsright now in the canvas side.And then uh pencil.dev is reallyinteresting. And I think Tom I thinkTom's a sneaky little winner. I thinkhe's going to come out with somethinglike he's going to continue to make thatproduct really good. The other thing isis is it's not just when you

00:25:05when youthink about stacks like these two,there needs to be like a shift in howyou think about work. And that's thehard part. It's like you know, I thinkwhen we started installing like assemblylines and stuff, people came in earlyand just like tried to install newequipment on top of of old factoryworkflows and it didn't work, right?Until Ford came out and said, "No, wegot to like rearrange the entirefactory." A version of that has tohappen right now. Metalab has a teamcalled Team Zero. It's like threepeople. They're tasked with like, "Hey,you pick the project, the client,assuming the client's cool with it.You're going to just do like a fullyautomated like you're not a fullyautomated, but you're going to exploreAI in a way that like the rest of thecompany's not and you're going to reportback to us how it goes. They have a theyhave their own set of criteria for howthey determine success on that." And Ithink that's a really smart way for acompany to mitigate the risk of hopingthat they reinvent the whole factoryfloor. This way this team can come backand just say like here's what's working,here's what's not and they can kind ofslow roll into it. You know, versus thenyou have the Rams who are just like,"No, we're just like everybody do it."And those are like the two differenttypes of ways that I'm seeing factoryfloor restructuring. Okay, let's usethis as a launching point because you'venow went on the ground with some of themost prolific design teams

00:26:20in the world.Perplexity, Vercel, Ram, Metalab. So,talk to me a little bit about theunifying threads or signals that arestanding out to you about how the bestdesign teams are operating today andwhat they're exploring. I think thenumber one is that the company is makingroom for the experimentation and it'sexpensive. It's expensive in token cost.It's expensive in cuz you go down rabbitholes that like are throwaway work alot. Leaders are saying, "No, I'm goingto unblock it. We're going to throwspend at it. We're going to give peopletime for this, you know?" And it'scoming in different flavors, but if theones who are really doing it well, likeat Vercel,they do a Friday hackathon presentationevery Friday. And it started as like afun thing for a very specific group. Andthen other people from cross-functionaldepartments were like, "Hey, can we getin on that?" And people just like createthings. They They're They're presentinglike startup killers at every single oneof these things. And those things justget shelved. Like it's wild to me. Andwe were there just for 1 week. Like justa little company getaway that they did.And everybody The ideas were like thesales team, they would go out and, youknow, like a typical hackathon, they'drecruit like a designer or an engineer.And if they didn't have one, they'd justdo it anyway. And it was like people arefighting to get in line to present onthat Friday.

00:27:36It's so company-wide. I'm trying to dothat. It's hard, right? Like I I feelfor everybody that's like, "Man, when doI get time to do that, you know?" Youget both ends. I've done experimentswhere I'm like, "Dude, I can't believehow well that worked. I can't believehow efficient that was." And then I'vedone other ones where I'm like, "I justspent an entire night trying to buildlike a custom icon generation platformon top of the Quiver API. And I justdidn't really quite get it to work theway that I wanted to." And I'm like, "Iwonder how much money I just lit on fireand 6 hours of my life, you know?" Thethree reasons that people are sayingfrom our survey that they're notadopting the tooling are pretty tied. Itwas like a multiple-choice question, andit was that the outputs they didn't feellike were good enough yet. It was tooexpensive in some cases, and that therewas no time. Like people are frankly tooemployed. And if their company's notmaking room for them, people are alreadyspending like their weekend probablytrying to get stuff out the door just sothey can show up Monday and look like itgot done last week. So on top of that,you know, I I feel for that a lot. Andthat's why I put the onus on like theleadership to be like, "Hey, you have tomake time." And if if if it's reallyexpensive, another way I've seen thisdone is they'll just do like thehackathon once a month or a quarter andlike give like a week to people. You cando kind of like Carmack week methodwhere it's like, "Hey, go heavy on worktime on an idea and then like let'spresent it and and like that

00:28:51could belike a really cool lessons learnedopportunity." The one thing I I'll callout to you about that is if like you domake room for people to go heavily insome sort of hackathon, like justsqueeze their brain in the base, you gotto attack on like that day of restafterwards. Cuz it's cognitively a lot.Do you feel that? Do you get cognitivelywiped out?>> Yes. I don't know. I think I'm on my ownweird creativity roller coaster andsometimes I don't know what the dominantfactors are.>> having a Carmack year. You're just likein it.>> Yeah. I can go so hard for like four orfive days where I'm just flying, youknow, everything I'm putting out feelsgood. I'm being incredibly productiveand then, you know, it'll be a Tuesdayafternoon and I'm like, "I'm just goingto go walk aimlessly because I don'tknow what to do with my life, right?"There is another piece of this that saysall the things that I've just talkedabout and have seen skew more helpful topeople like me who are changing lanes alot. Whereas like when you get to likethe specialist focusing on likeindividual tasking, you can get a lotout of it, but like at some point likethere's a a faster ceiling maybe. I'mnot sure. Maybe and and it might be morelike agentic like kind of systemrelated, but I'm starting to wonder thattoo. I love this stuff because I switchlanes daily, you know?>> I tweeted that

00:30:06months ago, but theability toeffectively multitask, like being ableto paralyze work at a really high levelis one of the core skills of operatingin this like AI world. Like I alwayshave multiple different types of taskgoing simultaneously or I'm working ondifferent work trees simultaneously,different linear issues simultaneously.Maybe I'm designing and coding at theexact same time and you're just bouncingin between. Like my tooling inside islike I use Conductor for everything andthey have this little choo-choo soundthat you can turn on when an agent isdone working. Like I just move from onechoo-choo to the next choo-choo[laughter] and it's like this littlegame of whack-a-mole and the choo-choosare doing all different kinds of things,man. They're doing all different kindsof things and and that's like my entirelife is just like choo-choo option spacebar talking to my computer. Choo-choooption space bar talking to my computerand I just do that all day. I want to Iwant to say something too because likewe're geeking out on I don't get to talkto a ton of people, I feel like, whohave the perspective that we have. We'vegot this like 10,000 foot view where wetalk to I don't think people willrealize. I was trying to count yesterdayhow many conversations I've had withdesigners and design leaders this yearand it's it's already in like likealmost 200, I think. It's been wild likedaily

00:31:22I'm having multiple conversations.The thing that I will I'll tell peopleis that, you know, when I talk to you Ilike geeking out about this because wecan compare a lot of notes that I thinkI can't compare with a lot of folks.But I also will say this is still reallyearly. It is accelerating prettyquickly, but if you hear us yappingabout this and you're like this isreally overwhelming, this makes me havea whole bunch of FOMO. You like you arefine to wait. I really think you canwait at the moment for some of this,especially all this agentic stuff,because someone's going to come out withthe tool that is eventually going tomake me want to off-board of my openclaw setup because it's just so welldone and it works really well and it andand most importantly the change cost isreally cheap. That's probably going tobe the great time to get in on thisstuff. You're not going to have to worryabout all the like tightening the screwsthe way I have to do right now, youknow, I think that there will come atime where and at that point like it'llbe reallycognitively easy to get into it is mythinking.One thing that Dive Club has madeabundantly clear to me over the lastyear is that the practice of design ischanging and the old process [music] ofgetting feedback just doesn't quite cutit in today's world. That's why I'mexcited to announce that InFlight isofficially in open [music] beta. It'sthe feedback tool that I've alwayswanted and it's built for a world thatmoves at the speed of AI. So I can share

00:32:37my prototypes, give context in videowalkthroughs, and Inflight makes it easyto get the exact feedback that I need tomove forward, whether it's voting ondirections or maybe even getting thegreen light to ship a new idea. And allof this is available in a single linkthat I can drop into Slack or maybe evenshare with power [music] users to testout a new prototype. I use Inflightevery day and it's totally transformedthe way that I share work. So, I'mexcited for you to try the product andif you ever want to jam about it, justemail me at rid@inflight.co. [music]Let's zoom out from there then and tapinto that kind of vantage point andperspective given the fact that we dohave a lot of conversations with peoplein the industry and leaders, teams, thatkind of a thing. So, maybe going back tosome of the on-site team visits thatyou've done. I'm curious, especiallygiven you know what you're talking aboutwith the open claw stuff, like I knowthat a lot of people listening right nowfeel this little tinge of pressure andmaybe this worry that they're fallingbehind and how do you adopt? Whichdirection do you point yourself when itfeels like you could go in every singledirection simultaneously? So, did yourtime with those teams do anything toshape the way that you think about thefuture role of the designer, but evenmore specifically

00:33:52like where we shouldinvest and what we will even bring tothe table as professionals in the monthsand years moving forward? I have so muchto say about it. So, there's like inthere there's the sentiment you hearfrom just like the working designers.And there's a lot of sentiment that'sgood and there's some like very There'sthe decelerationists, the There's peoplewho are afraid and there's theaccelerationists. Then on the other sideit's like what are people actuallyfocusing on? What are they becoming?What seems to be happening is morepeople are finding themselves beinggiven responsibility is what I'm seeingand having to lead something, whetherit's like the agentic installation orpeople or even both. I want to see thoseroles starting to have compensation tomatch that. That's kind of That was likemy hypothesis from 2024. Should likethis visual that it was like as thetools get better, the capacity for workstarts to open up. Like you can do morein a day. And I think there's acognitive component of that that wasn'taddressed cuz that's true. But you cando more. It's like you have more time,you have more ability. Like you stackmore things inside of that open space,compensation should increase. That's thehypothesis.And I actually want to run a surveythis quarter tracking where that's at.For teams in like where where I thinkpeople should invest skills or wherethey're looking at it,

00:35:07it's just likebeing able to build intent and zoom inand out. Like the ability to movebetween altitudes,not just of like product thinking andthen like the interaction moment, butalso from workflows. Ben Blumenrose hada really good answer to what makessomeone AI native. And it's not aboutthe tools. Like it it's not about Likethere is some competency in there, sure,but to him it was what do you do withthe question what if? And and I feellike that framing is really where Iwould rather have designers focusbecause I'll have some people who arelike, "Hey, what do you recommend I doright now? Should I create a project?Should I like Should I play with ClaudeCode?" And I would just say like tuneout the noise for a second and think ofsomething that you have as like aproblem. And and like a problem everydesigner has is the blank canvasproblem. It's the Hello World project, Ithink, to get you into this stuff. Andthen just go and figure out how to solvethe blank canvas problem for yourself.Meaning we would do crazy eights. Youknow, we would fold a piece of paper andwe do quick crazy eights in 8 minutes, 1minute per idea to come up with likesome solve some problems, some interfaceproblem. What is the version of thatwith the tooling available?And go and just explore a couple oftools to figure out what is the blankcanvas problem. Oh, maybe you open uppaper and

00:36:23you just like use the new toolto do stuff you're already familiarwith.Or maybe you open up Claude code and youvibe code like eight quick variations.Or maybe you go even deeper and you'relike, "Hey, how do I connect Claude codeto Papers MCP?And then how do I actually create a planin Claude code and have it executed inin Papers MCP? So that I just open upPaper and I have like eight probablyshitty versions of this intense that Ican just kind of like decide what'strash and what's interesting. Just dothat and and do it in a way that's likeinteresting to you. Don't don't look forlike Tommy's newsletter. Don't look forthe Dive Club newsletter and figure outwhat you should do. Just like tune itall out and pick up a couple of thingsthat like you found curious. Likepractice thinking and and just try it.And start to build your own littledatabase of insight around what you'relearning and what you like about yourworkflow. Cuz everybody's workflow, andthis is the the one thing I have seen alot, is different. Like if we were toput screenshots up of everybody'sworkflows Yeah. that are like emerging,they're shapes and sizes and like oblongand weird and they're just very customright now. Even within an org. Yeah,even within an org. The delta betweenworkflows within an org. Like I've beendoing a lot of customer calls and it'sshocking,

00:37:38man. Like obviously everybodyhad their own process and means to tapinto creativity, but people are usingfundamentally different tools, ways ofworking, ways of thinking. Like the thespreading apart of the creative designprocess has gotten so extreme. It ischaotic and and I think there's a lot ofvalue in the people, the teams, theindividuals who are just sharing theirworkflow. And I think it's important toremember that like that's a goodactivity right now. And you know, youdon't have to attach the language to itthat like, "Hey, here's my workflow. Youshould do this." Right? I like strikeshould from the vocabulary right now andjust like compare compare workflows,compare shapes. Because we willeventually converge around some betterpractices around this stuff, but wedon't know what that is yet. And soright now the best thing you can do isif you want to get involved in this thistype of work, to just try to findworkflows that like are are good for youand spend like 80, you know, somepercentage of time on the familiar stuffthat gets the work done, and then somepercentage of the time like where youhave you make room for yourself toexplore a workflow. And so the point forthe people who I said you can wait, likewe're going to converge. We are going toconverge around stuff. If you have theluxury and you don't have to getinvolved yet, you can wait and thereprobably will be like a concentration oflike these are the better workflows.

00:38:53Ithink something that makes our journeyssimilar, but also unique maybe to theaverage designer who's listening, is weboth made this investment indistribution, putting our ideas outthere, kind of seeing it where it wouldtake us, and at least for myself, I kindof rode this wave of Figma throwing in abunch more technical features. There wasthis new type of design that felt alittle bit more like engineering,honestly. And And it was it was reallyeasy for me to just tinker. I wascurious. I just went for it. I justtried every little different tactic andkept putting ideas out there, and itwasn't like do this. It was more, "Hey,I tried this. This was really cool. I'mgoing to try this again tomorrow."And I think we're in this moment in timewhere that opportunity is wide openagain for people who do kind of want tolike grow within the design community.There's so much hunger and desire forpeople to just see workflows and 100%different experiments that people arerunning and the different learnings thatothers have. And if you just show up,try things, be curious, and thentreat probably Twitter like a littlejournal of the things that you'relearning and the experiments that you'rerunning, like this is the moment in timeright now to reap the rewards of beingpublicly curious. I've been veryprescriptive to the last couple ofpeople who are newer designers

00:40:08and theyasked me like, "Hey, what should I do?"I said, "Okay, do the thing I just saidthere, like pick a pick a project, do anexperiment, go to Twitter, share it. Youdon't have to act like you're a guru.Like this is not about guru. This isabout>> People will sniff In fact, in fact,don't. Yeah. And be building public andjust say, "Here's what I did." Cuzhere's the thing, like the things thatreally pop off, so if you want to talkabout just like algo maxing, it would becreate like vibe code something reallyvisual, something that's like weird andunique, like a 3D graph or, you know,some 3D space and time thing and justshare it. And like that it's so easy tovibe code for content now. But if youwant to make it useful, and that's whatI would recommend, is create somethingthat is specific to a workflow you wantto solve for a projectand then just capture thevisual artifact out of that. There'ssome moment in there that's reallyvisually interesting and just share itand then a few bullet points on what youlearned and like you would just farm.Like that would work so well right nowfor client work, for lead gen. I don'tknow if you're seeing this, very curiousto hear your perspective, too, butsomething that I'm kind of starting tonotice specifically in more of thejob market is that there's this demand,not on paper necessarily, but there'sthis demand

00:41:23for designers who can helplevel up the rest of the org and the waythat they work and the AI systems andthe the person that can just kind ofblaze the trail, you know? And if youcan position yourself as that, whetherin an interview or even just on yourTwitter profile, it's a massive,massive, massive green flag in today'sjob market. You seen that at all? Yeah,the thing that's been really interestingagain, like it's the internal toolbuilding. If you are not a designatedleader at the organization, the absolutebest way to find yourself being leanedon that way is to do something thatBrandon Engineer did at Vercel. Find across-functional use case where designis needed by marketing for a task thatlike you don't have time for, you don'twant to spend time on and build aninternal tool. I think they said theywould was 1,500 vibe coded tools insideof ramp that they had made, right?And the question is like, well, how muchof that is throwaway? Well, most of it.Like most of it is throwaway.And and so the question was like, well,how does like internal adoption happen?Well, it happens just like anything elsehappens with a product. It's useful,someone tries it, and then it starts tospread like, oh, so and so built a thingfor this. Mhm. And and that sort ofthing like you are you are installingsomething new. You are

00:42:39now the subjectmatter expert on that thing that'sgetting installed. People are going tostart looking at you and saying, hey,like that was really interesting. How doI do that?And they'll start making time for you.Like that is almost a way where if youwork at a company that doesn't have alot of time for this stuff, but you youdo have the ability to use the toolingthat way, which is another issue too,right? That's a fantastic[clears throat] way to do that. Andthat's what I would recommend. Multipleepisodes are coming to mind where peoplewho have built the early prototypingplaygrounds atNotion, Atlassian, Stripe, like theywere catapulted internally in the org.You know, like it became a it became abig deal all of a sudden where differentdesigners were relying on them andthere's just so much to figure out anddesigners we are problem solvers.And it's interesting to think about howa lot more of the problems that we willbe solving in our day-to-day roles mightbe internal rather than user-facing. Andthere's actually more bandwidth for thatnow too as we can accomplish theuser-facing tasks much more quickly.People are like, why would you createtools that would replace yourself?You're just replacing yourself. You'reyou know, this is a self-own. And youknow, at ramp there's two things thatstuck out to me which wasthey really want people designers tofocus on like creating 10 out of 10experiences on like the friction pointsthat matter. You don't need to becreating the blog post image formarketing. You know, like that'sprobably not the best use of your timeas a product designer.

00:43:54The other thingthat was really interesting to me wasthe tools that do get adopted internallylike they look really cool, man. Likeand I'm like, whoa, that's weird becauseanytime I've ever worked on a aninternal tool, I remember one forQuantcast like our account managersneeded something. It was just like makethe thing work. It was like bootstrap.It was a bunch ugly things strappedtogether. It just worked. And then evenwhen it got leveled up at some point, itwas still just sub par. Because now I'mlooking at like there was a tool. Ican't remember what it was. It wasn'tglass at ramp. And Diego and Simon Coreyhad been working on like this drop capletter. And and it was like you weregoing to get this summary every day. Andthe drop cap letter was just like thistapestry with the old like you know,Renaissance language. And it just reallybeautiful. And I was like that's crazythat that much attention to detail wentinto this internal tool. I get it. Likethat makes a lot of sense. And I thinkthat that'spotentially I mean like I just I justthink it continues to grow on thedirectory like that. I'm not thinkingabout Shopify's artifact tool. And theyhad that tweet and it's like one of thebest design products of the year and itwas just an internal tool. Yeah. Ireceived so many DMs in the weeksafterwards. People being like, can youjust build this for invite? Like this iswhat we want for invite, you know? Andmy head I'm like, yes. But then theother part of me was like, I don't knowif I can design a UI that good. Likelegitimately. I that internal

00:45:09tool mightjust be beyond what I am capable ofmaking, which is insane because you'reright. Like if you would see somebodyeven that had a resume where they'relike, I've been building internal toolsat X company for the last few years.Your brain immediately is like, well,they probably don't have the visualskills, you know? They're probably justbuilding really gray box functional UI.But that is not the case anymore. No,it's it's it's really interesting. Andand I you know, I'm cautiouslyoptimistic. I'm not like full-blown allI there are some people who think moveeverything out of the way. Let theacceleration of this happen as quicklyas possible because we got to getthrough the painful middle as fast aspossible. I've heard that take.And I mostly align with that take. But Ido still think don't go 100 miles anhour. You probably go like 60 miles anhour. Like let's go the speed limit andlet's pay attention to like what we'reseeing as we get there.But let's get let's try to get there. Ithink getting there is a worthwhilecause. All right, let's talk to a veryspecific listener for a second who theyare working at a Midwest mid-tieragency. They've been there for three orfour years. They did some randomfreelancing before that. They're notgoing to win people over with theirresume, but they believe that they havethe skills to take a big leap and toland one of these roles that people careabout and kind of puts them on the map,per se. Help them think about where topoint.

00:46:24What are the types of ways thatthey should be spending their time? Youknow, you only have so much time outsideof day-to-day work responsibilities.What artifacts or processes or pieces ofknowledge, what can they do to equipthemselves to take that leap and play inthe big leagues? I think the biggestthing you can do to add value right now,I I really think in not every designerwants to be a systems designer, likesomeone who's thinking about systems,and I I can appreciate that.Um I do think there's a world where alot of very visual specific folks arestill valued, but I think right now, thebiggest thing you can do is to go in andto try and help cross-functional peopledo more design work easily. Like partsof the process. Enable your marketingteam to create that blog image withoutneeding you, right? Like create a designfunction that is enablingcross-functional partners. I truly don'tthink you're going to cannibalizeyourself. I don't think that's what'shappening. I don't think you're going tobuild these processes and then they'regoing to be like, "All right, thanks.See you."I think like they're going to find moreand more things that need that sort ofsystem thinking. I think there's goingto be more things that need like actualproduct contributions that aren't thethings that cross-functional

00:47:40partnersshould be focusing on, and you're goingto have a lot more time to do that. Whenyou have a lot more time to do that, Ithink of someone like Andrew Wendling.He's this wonderful designer over atIverson Studio.And now he's living in this wildgenerative image video world and andcombining those with coding projectsand and getting these like really richexperiences. Amelia Wattenberger sharedsome of her like just blog posts thatare these combinations of interactionsmade between different generative imagesin like really thoughtful philosophizingposts about the future of interfacework. And it's like those things becomepossible because you now have timebecause you've enabled these systemsthat remove the toil of what you have towork on. Let me push on that a littlebit because you could make the mosthigh-impact systems at this old companythat you're trying to leave imaginable.It's a whole other set of challenges tofigure out how to stand out in themarket on top of that skillset,demonstrate that value,do it in a way that gets people to spendmore than 10 seconds on you as a person.So, how would you go from there becauseI think the fact is that's not as easyto sell as the kind of

00:48:55goofy machinethat you were talking about gets playedon Twitter. I don't have the perfectanswer. I don't want to act like I'm asubject expert on how to get hiredtoday. I'll tell you what I'm saying.Intern programs that still havewhiteboarding but like before that theydo 20-minute exercises of live coding.Thought that was really interesting. Youknow, they want to test thosecapabilities. Yeah, I I was like theysaid they didn't want to see like Figma,they wanted to see like hey, live codethis. And it was for product designrole. There's like people who are tryingto figure out how to evaluate AIfluency.And I think like AI fluency is going toeventually just become like an ambientthing. I don't think we're going to likehighlight that language forever. But Ithink right now it is important languagecuz a lot of companies are like yeah, wewant this designer who can come inand think. We want someone who knowswhat to do with the question what if ina way that doesn't feel like it's from2012, 2015, you know, 2018. Thatquestion seems to be being answered bylike what are you showing that you'resharing? Like social media is the newdiscovery engine right now. Like that isthe way to for better or worse, that isthe way to prove that you are someonewho is experimenting and that you'reexploring what to do with the questionwhat if in ways that likea lot of other people aren't. Let's saythat you're the hiring manager andyou're talking to this hypotheticalperson that I've been bringing up thelast couple questions and you are trying

00:50:10to figure out specific questions ormeasuring sticks to get at this level ofAI fluency. Mhm. What are you reachingfor and what would what signals wouldyou hope to see? Right now what's reallyI mean this is I think the biggestquestion that I want to know is likeshow me your workflow. Show me how youdesign something, you know, take methrough an idea that you've created andand that is something that I want toknow. I want to watch and pick apart andask why and see which tools are involvedin that. That's a challenging thingbecause I don't think people are likewhat do you mean my workflow? Like well,I went over here, I asked somecustomers, I went over to Figma, I madethis thing. I think the things that Iwant to know is okay, cool, yeah,totally get that. Is there anythingyou've played with and like why did itwork or why didn't it work? Oh, likewell, normally I would pull up superwhisper and I or whisper flow and Iwould just be like shh shh shh shh shhand then it would like look like this,but I don't I don't want to show youguys that cuz that's like a much wasteof time. I'm like no, no, show me that.Like actually show me like what are youwhy did you decide to do voice?Well, because like I'm kind of rambling,I'm not a great communicator. If I justtalk, but AI's like really helped mecondense my thinking and so then I cantake that and I can plug it into thisprompt over here and I found that itcould actually generate this planbetter. I'm like okay, cool, now we'reonto

00:51:25something and like let's dig intothat. You know, and not everybody hasthat workflow, but that's what I'mlooking for is like have you played withworkflows like this? How are youthinking about it? Like why did youdecide to do that? If I'm looking forlike AI fluency, I'm looking for I'mlooking for that. I'm looking for peoplein the playground of not just like youhave a playground where you vibe coded,but you've also like you're treatingyour workflow a little bit like aplayground. Whatever your alt-tab is onMac looks different at any given week.I'm going to pull back the thing thatyou said earlier, which I feel like is alittle bit in tension with this, whichis it's okay to just wait. Because whatyou just said is totally what I'm sayingas well. And I have this [clears throat]in the talent network, we have thisproduct that kind of powers everythingbehind the scenes called decimals, and Ihave this thing I can hover over and itshows the green and red flags that ahiring manager is looking for in anygiven role.And it's really interesting to look atthe trends across those green and redflags. A red flag that I by far, thenumber one red flag that I see, is thata candidate has not demonstrated anyinterest or curiosity in new AIprocesses or tools. What you'redescribing is kind of that's the that'sthe opposite, right? Like that's thegreen flag. If you can show up and belike, man, I've just been exploring inevery different direction, and I didthis one thing and that it really

00:52:41thisworked kind of over here, and I'mplaying with this new tool. Like, hiringmanagers love that. They love that. Idon't have the perfect answer for thebalance. I'm trying to be responsiblewith language and not just crush peopleinto fear about this. Again,60 miles an hour versus 100 miles anhour. I do think at some point, cuz youhave to understand what it says aboutyou as a designer though. If you're notexperimenting, it says that you mightjust be too damn busy. You may not havethe luxury, and that's very real. It's avery real thing. You work at a companyor at Oracle where like the mandate islike, we are we we do not think thefuture of design work looks like that.And they might be right. But I from myunderstanding, Oracle is like very, veryanti that direction.And maybe that's the situation thatyou're in, and there's not much you cando about that. And it's probably goingto be okay cuz like you'll figure that II do think there's going to be a asecond adoption wave that occurs. But Ido think if you're somebody who's likelooking for a job right now, and youwant to stand out, like if that's that'swhat we're talking about, is standingout, having a shorter loop to gettingthat next piece of work, it's a reallypositive signal that you're exploringthese tools because it says it's it's100% says you give a It doesn'tmean you don't give a if you're notdoing it, but it's 100% means

00:53:56you give a if you are doing it. And andsometimes the biggest differentiatorbetween two people is like how much gasthey have in the tank. How much give a do you have? If you're going tokeep up with this stuff and if you'regoing to do really good work, you haveto care a lot. You have to care a lotabout this, you know? I'm going to putyou on the spot here. The chances of usgetting this right are almost nothing,but let's say that we have thisconversation 6 months from now. We runit back on an episode. Yeah. What topicdo you think that would be much more inthe zeitgeist that we would be talkingabout that, you know, isn't as top ofmind today? Well, I I mean I think thatcontext layer of headless design. Ithink that's where this is going. I Iwill I like making predictions and Ilike fact-checking my predictionsbecause it helps me like hone myworldview. Like, am I full of or amI like paying attention to the rightthings?I've been on a little bit of a run. Ithasn't always been the case, but I feellike feel like I'm I'm plugged in. And Ido I I think we're going to see thislike headless design topic, this thisI'm I'm so early. We're so early talkingabout open claw installations and likeramp building glass. And I think overthe next few quarters we're going to seethat really accelerate. I think a lot ofcompanies are going to realize like,"Oh, agents are using tools. What doesthat look like and what does that mean?"Dan Shipper just launched um

00:55:11thatwriting tool. That's right. And that wasa really That was a really likecompelling case for working alongside anawakened AI that's just payingattention. That's like, "Hey, actuallyyou should do this over here." Oractually this thing this insight came inlike maybe you can consider this. Like Ithink there will be the firstconversation is the installation of thatbriefcase, moving tool to tool, and thenafter that I think it's going to belike, "Okay, how are people actuallyworking with an AI that is like alwayson in the tool with you? I think that'sgoing to be a real conversation thatcomes up. And that sounds like sci-fi tosome people. Like we're there already.Like we are there. It just hasn't hasnot yet. There's just a lot of frictionto creating and managing the briefcase.Correct. The onboarding is terribleright now. The onboarding is like not,you know, but that's I think it's goingto be solved very quickly. You thinkabout the design of the interior of thebriefcase and where are the folders?What goes in where? What's the taxonomyof the little file paper I I've neverhad a briefcase, but I would imagine youknow, there's like little compartmentsinside of it. The design of that isgoing to fall as much on designers asanybody else. 100% It's a wholedifferent type of problem to solve.>> Yeah, and that's why you see like Googlestill launching like the design.md file.You're going to see a lot of I thinkinitiatives trying to

00:56:26standardize whatthat means. Like how do you standardizeintent? And and we're we're moving inthat direction and it's going to lookprobably very different a year from now.And how do you share the lessons derivedand the best practices derived across anorg too? Because something that we saidin the very beginning is we're not evenreading our markdown files. I don't evenread my markdown files. So how isanother designer going to have any cluewhat is in there, you know? Like there'slike a visibility challenge that's goingto be really interesting to think too.>> That's that's a really good point. Ihaven't thought about that. Aselaborating, collaborating around thoseblack boxes is going to be a challengethat needs to get solved. Totally randomquestion Yeah.>> left field before I let you go.If you had to take a W-2 role at acompany tomorrow,where would you want to work and why?>> Oh man, I think it would be Vercel,dude. Yeah.>> Really? I left that like they're notfull of Like they're they'rereally working in this way and I am justa builder and I really think the futurebelongs to builders. There's likesomething I said the other day which waslike I think the future is bright, butit has to be built. And I love beingsurrounded by people who are just likebuilding, who just like are obsessivelyabout building in a direction thatbetters things. And and I

00:57:41that was oneof the coolest cultures I've seen.There's a gravitational pull towardsmore technical products right now that Ithink gets kind of the cream of the cropof the designers that skew builders aremore like the early adopters, you know?And so it's no surprise to me, I guess,that those cultures are attracting thelevel of talent that they are. Becauseanybody freaking out right now that allof a sudden they can build their ideaswants to be in that kind of environmentwhere they're told to just run crazy.And so yeah, there is a little bit moreof a separation maybe in terms of uh thethe different tiers of orgs and whatthat makes possible for a designer. So Iget it. Anything that we haven't talkedabout before I let you go that uh you'reQuick plug, we got a party coming up.>> We do have a party coming up. Well, ifyou're listening to this, come to SanFrancisco for Config. And the last nightis Detach. It's going to be so fun.Detach is this um idea that JesseShowalter and I had two Configs ago.It's where you and I met, I think. Ithink it was like 150 people. It wasjust like let's just invite like oncethe conference is done, let's not have amini conference. Let's like just I Itold Jesse I was like I just want to getdrunk with all the designers that I seeonline and hang, you know? And and uh wedid. We had it turned out great. Thenlast year it was like a little bitbigger. I think we had like 300. Andwe've been like selling out every year.And then so this year I'm like cool, wehave like a three-story at 25 Lusk.Really cool

00:58:56venue. It's got like abeautiful roof rooftop. And we got like400 people. And it's like the people whoare hitting me up who are coming likeVercel's team's coming. Brett fromDesignjoy, who never leaves. He neverleaves out of his house. He's like I'llbe there. And I'm like heck yeah, youknow. Josh Puckett. And it's just somany cool people. And I'm telling you ifif you have an interest in like reallynetworking in a way that's like notpretentious or anything, come to theparty. Come hang out. Like I want tohave a drink with you. You get to alittle taste of how amazing thiscommunity is. And Tommy, you've played aa pretty outsized role in contributingto it, shaping it, and super appreciateyou coming on and sharing a little bitabout what's on your mind and whatyou're seeing today. I think we shouldmake a regular practice out of thesebecause that vantage point is a good oneand things are moving way too quickly tohave a full year in between episodes. Weshould have like a bi-annual state ofthe state of perspective, you know?>> I [music] think that I think this is thefirst one of those and we'll have more,so.>> [music]>> Inspired by you, by the way, so that'swhere it all came from. I got you. Allright, but I got you.