Jamey Gannon - Become an AI Creative Director with Nano Banana and Midjourney

00:00:00You don't want my prompts. My my promptshas nothing to do with prompts. AI isjust like any other creative process.Like you have to get good at it. Youhave to know what you want. You have toknow like what tools you prefer. Youhave to practice. It's not somethingthat, you know, you can get overnight.>> Welcome to Dive Club. My name is Rid andthis is where designers never stoplearning. Today's episode is with JamieGannon and we're doing a deep dive intohow you can be an effective creativedirector with AI. This is so clearly amoment in time happening right now whereif you can master these tools, you canstand out in today's market. So, we'regoing to get a behind the scenes ofJamie's process and how she works withMidjourney and just seeing how sheworked is totally changing the way thatI'm thinking about the importance ofprompting. So, let's dive right in.>> I think most people suck at AI for likethree main reasons. One, they lackdirection and they don't know what theywant to make. Two, they're overprompting and not giving AI the rightsignals to make what they want to make.And three, they're using the wrong toolsor models or they're using the modelsincorrectly. Maybe we could go into theprompting piece really quickly becauseto me I almost would have assumed thatyou would have said something close tothe opposite where

00:01:15it's like the themagic is in the detail and the qualityof the prompt. And yet I came across aline that you shared that I found veryinteresting where you said AI is adirection problem, not a prompt problem.So, I'm hoping you can unpack that alittle bit for us. I was like wrestlingreally hard with how to explain this andthen I remembered the like old adage, apicture is worth a thousand words. So, Ithink for a couple reasons a lot of likethe ex-AI influencers, the people thatlike art designers or they're peoplethat um, you know, are really good atlike marketing and stuff so they getinterested in AI to like increase theircreative output there. And because ofthat, they're not very creativelyminded, I find. So, they get reallyexcited about being able to like reverseengineer a prompt and like write theselike crazy prompts and get these likeone shots that don't require anyintuition or mood boarding. I thinkpeople miss how AI is much better atunderstanding a vibe and an image andlike a collection of vibes and imagesmaybe than you will ever be. Somethingthat I have my students do in my course,uh, I have them try and describe thisimage of Hailey Bieber. Now, to mostpeople this, you know, might be easy,you know, it's Hailey Bieber, she's insweatshirt, uh, on a red background,she's posing

00:02:30kind of sexily. Maybe ifyou are like an art director or creativedirector, you know, like it's shot on afilm camera, you can say like lime washon the wall. But, if you give this imageto ChatGPT, Gemini, any other LLM, andyou say like describe this image andlike extreme specificity and give me aprompt, it'll give you like potentiallythree paragraphs. It'll get into likethe age of this sweatshirt, how heavy itis, you know, it might notice her ringsand you don't. It might say that she'stan. It might mention how her hair isfalling and describe the way herexpression is in a way that, you know,sure, you could do as a human with a lotof practice and maybe like running it acouple times. But, you're never going tobeat LLMs at this. So, people that tryand beat AI and like taking all thislatent information and putting it to useare kind of just like fighting a losingbattle in my honest opinion. Real quickmessage and then we can jump back intoit. AI is making it a lot easier tocontribute to the in-flight codebase.So, I'm not going to have polish ticketsall the time. And now, AI agents insideof Raycast make that easier than ever.Like, let's say that I see somethingthat I want to fix. In a single hotkey,I can launch a cursor background agentat a screenshot or any

00:03:45context that'sneeded and then manage the entireprocess directly inside of Raycast. Andyou can ship code as quickly as it wouldtake you to type up a new issue inlinear. It's a big deal for designersand you can start using it today. Justhead to dive.club/raycast.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]of getting feedback just doesn't quitecut it in today's world. That's why I'mexcited to announce that Inflight isofficially in open beta. It's thefeedback tool that I've always wantedand it's built [music] for a world thatmoves at the speed of AI. So, I canshare my prototypes,>> [music]>> give context in video walkthroughs andInflight makes it easy to get the exactfeedback that I need to move forward,whether it's voting on directions ormaybe even getting the green light toship a new idea. And all of this isavailable in a single link that I candrop in a slack or maybe even share withpower users to test out [music] a newprototype. I use Inflight every day andit's totally transformed the way that Ishare work. So, I'm excited for you totry the product and if you ever want tojam about it, just email me atrid@inflight.co. [music]Now, on to the episode. All of

00:05:00whatyou're saying really resonates, but nowI'm like, okay, cool. So, [music] how doI put that into practice? So, the firststep in my process is always startingwith a direction and a mood board. So,that can be like a specific camerastyle, that could be a specific sort ofvibe or attitude. Like in this case, itcould be a specific type of model orscene. And then from there, there's acouple different options that you can dodepending on what you use. For thisexample, I'm just going to show you guysMidjourney. So, uh Midjourney, you'dactually upload this mood board, justcopy and paste it into Midjourney moodboards and then you can just use that asa style. Um Midjourney will make you acode based on that style uh and then youcan use that. Now, what I found and whatI'm showing you guys here is thatsometimes the mood boards like don'tactually work that well. Um Midjourneytends to kind of like average out uh theimages in the mood board, especially ifyou're kind of be a little bit lessspecific on like camera style or modeland things like that. So, you can seehere, if you compare these images I gotback to the mood board, it's it's it'skind of missing the mark. And somethingthat I think is really important in mycreative practice as I've grown as abrand designer, graphic designer, butalso as like an AI creator, whatever wewant to call us, being like radicallyhonest about what

00:06:15my designs look likeor what uh my outputs feel like and ifthey're actually on track with like uhthe goals of the project aestheticallyor business-wise, it's something you gotto get good at. And I think that's wherea lot of people kind of get tripped up,especially when you're starting with AIor really anything. It's just like thesecond it doesn't look like, you know,six fingers, you're like, this is a win.Like these two images like independentlylook really good and, you know, I wouldmaybe use it for something else. Butthen when you again compare it to themood board, we're missing kind of likethat high contrast, um that's darkness,um a lot of this like color treatment oflike the the neons with the pastels. Wecan do better than that. Uh so, what Itry next is um S ref. So, that'sbasically uploading my mood board imagesas style references in Midjourney. Justagain like copy and pasted into theprompt bar. And then we start gettinglike much better results. So, you cansee the style and kind of like thecamera treatment is much better. Evenjust like the aesthetic elements feelbetter to me. In my opinion, again,going back to that mood board, you know,it's not quite there. To me, it'sfeeling a little bit green. The nextstep for me, I'm going to take away thegreen. So, we have this green eye imagehere. So, I'm going to trygenerating without that. And here we'regetting like much much closer to thingsthat fit in that

00:07:31original mood board.So, that's kind of like my creationprocess and then starting to begin myiteration process. And as I begin toiterate, you know, you could bring indifferent elements. So, in Midjourney,we have this great thing called likeprofile codes. Um profile codes likework very similar to mood boards. Um butinstead of uploading a bunch of images,you rank images and Midjourney'spersonalization feature. So, when I wasmaking this personalization code, I wasthinking like 2025 aesthetic, reallyclean studio photos, iPhone aesthetic,nothing like vintage or or illustrationlike this. So, in this case, I wouldn'thave ranked these. And I basically dothis to kind of infuse a little bit morevariety of input into Midjourney so itdoesn't like continue to kind of makelike the exact same stuff on like kindof the same plane, if that makes sense.Uh and then also infuse some of my kindof like personal aesthetic choices thatI've made. And here is where I thinkI've kind of like reached a pretty goodlike success point with this mood board.You know, the images look crisp and realand high contrast. I really love theskin. I love the amount of color that'sin it. Next steps is just likecontinuing to iterate within this kindof like winning prompt structure that wehave. And if you've noticed[clears throat] like this whole timewhen I'm prompting, I'm doing extremelysimple prompts.

00:08:46One, because we're justtesting. So, we don't need to like betoo specific about scenes or or peopleor types of models. But this is why whenpeople ask me for prompts, I'm like, youdon't you don't want my prompts. My myprompts has nothing to do with prompts.You'd be hard-pressed to like recreatethis image with, you know, reverseengineering the prompts. Like this thisprompt is just astronaut. And these arethe results that I get. So, this processlike one, is not only faster and likemore scalable, especially if I want todeliver this to a client or use it aslike my personal brand. It's also justlike better as a creative because I'mable to interface with ideasthat not only the AI understands better,cuz again, a picture's worth a thousandwords, but also things that I understandbetter. Maybe I'm not as good withwords, but I know that like, you know,this close-up of a woman's face is goingto give me kind of like an edgy,editorial vibe versus like trying toexplain that. I really want to look atclient work. I want to pause reallyquickly though and zoom out for a secondbecause we just showed Midjourney.I am coming to this conversation assomebody who, like I said, like this isnot my wheelhouse. I feel a little bitbehind even in the tooling space. Howmuch of the tooling space actuallymatters to you? Like are you doing thevast majority of your work inMidjourney?

00:10:02Are there other tools thatyou might reach for? What are thosevariables that would kind of what toolsare in your stack?>> There's a difference between like closedmodels and models that can only beinteracted with on like their platformsand then models like VO3 or Nano BananaPro or Kling. These models can beaccessed on like almost any platform.You know, you have Flora, you have CreA,also Higgs Field, which has been like abig controversy recently. So, shortanswer iswhen I'm teaching AI stuff or when I'mgiving recommendations, I almost neversay a platform unless it specificallyhas a really great feature or has aninterface that I really like. So, I useMidjourney and Nano Banana Pro almost90% of the time because Nano Banana Prois a reasoning engine and it's reallygood at listening to instructions andrealism and even text. Like I've hadpeople generate like full slides of textwith like extreme accuracy, like noblurriness, no AI soft, nothing. Andthen Midjourney is an aesthetic engine.So, that's where you can actually getstuff that like feels like something anddoesn't just [clears throat] look likestark AI all the time. That's veryhelpful and also encouraging because Iappreciate the simplicity of the answer.You know, you can choose any tool.There's so many to choose from. Whatactually matters is the models thatyou're

00:11:17selecting to use in them and whenyou're selecting to use them and forwhat task. Okay, so maybe we can diginto that as a part of I kind of justwant to see a little bit more of yourprocess and maybe we could look at someof the projects that you've worked oncuz I can already tell there's still alot to learn. So, I can show you anexample of kind of going from Midjourneyto a different model for a differentpart of the creative process I use. Kindof what the example we were talkingabout a little bit earlier, how to gofrom like a Pinterest or a Cosmos moodboard into like a set of images createdwith Midjourney with like a lot ofconsistency. I went for like iterationflow and I tried a bunch of very simpleprompts and got these images, but someof them are not quite to the qualitythat I would like or I think that I canlike push them a little bit further.Especially if we're thinking about, youknow, like a final output, like being anInstagram post or being something on awebsite. One example I have here, andyou'll get this all the time withMidjourney, and this is why you kind ofneed to have like a repertoire of modelsthat you use. Midjourney, as much as Ilove her, she's not quite good yet atlike hands or like consistency of likecomputers or devices or even likeclothing sometimes. Yeah, you can get itto give you a MacBook with the rightprompting, but a lot of the times you'regoing to get a really good composition,a really good model, a really

00:12:32goodexpression, but there's going to be likeone or two details that are missing. Andinstead of like toiling away inMidjourney trying to get it to listen toyou and like in-painting and and doingall this stuff that it's really bad atright now. What I do is I just take itinto Nano Banana Pro. So, in this case Iuse Flora very often. Again, it's justone of those tools that has a lot ofmodels in it. It's a multi-model tool.Um so, I [clears throat] use it thereand I give it a very simple prompt thesame way that I would, you know, speakto like an intern or something and Ijust tell it, you know, replace thecomputer she's typing on on a 2026midnight MacBook Pro. Great thing aboutNano Banana Pro too as well is that itkind of it has context to like theworld. So, as long as the thing is likea popular thing, you can usually kind ofjust like say what it is. You don't haveto like upload a reference photo. Andthen I've just learned from many timesprompting that a little bit ofinstruction like keep the position thesame, keep the size the same, I mean,what what exactly are we seeing of thecomputer? That tends to just help it getit right the first time. And yeah, andthen we have likea real computer now in the shot. Umalso, this is upscaled to very subtly,so it's like a a 4,000 by 4,000 photoversus like a 1080 by 1080. We get alittle bit more detail in her shirt andher hair, but

00:13:47everything else is exactlythe same.Now, if you were trying to do this inMidjourney, it would take a long timeand you probably wouldn't ever end upgetting it perfect either. But uh NanoBanana is like very surgical. It's likesurgical Photoshop. It's like aPhotoshop that we've been watching forour entire lives. So, that's how I tellpeople to use it essentially. It's goodsometimes for likeearly stages of photography as well. Butkind of similar to Midjourney, there'sthings that you have to fight everymodel for. So, with every model, there'sgoing to belike strengths and weaknesses, at leastfor now. So, I found with Nano Banana,while it's really good at realism andkind of like taking that like final stepto make things look like real images, itcan be a little bit AI smooth slop atfirst. Like this photo of these tworunners, while uh you know, I'muploading the style references and I'mdoing like pretty minimal prompting,we're getting an okay result, but forme, like it's a little bit too focused,it's a little bit too editorial. Doesn'treally have like the same vibe of thereferences. And they're also likeexactly in unison, their socks match.So, you have to do a little bit morework afterwards like actually get it tonot look so like fake and robotic.>> Can we dig into that for a second?Because that photo to me

00:15:03perfectlyexemplifies what I do notice in a lot ofAI imagery, which is like it's just tooperfect. Like you said, their socksmatch. Like come on, who's going on arun with the matching highly uniquesocks?How would you approach refining thisimage from this starting spot?Yeah, so if I wanted to refine thisimage,umI would do a couple things. I might tryand change the angle. So in this case Idid. Um trying to make them like astraight-on shot uh to me sort of likerelax the frame a little bit. But thenthe next thing I would probably try ischanging their outfits. So I would workin steps and I would just say, you know,change the man'sshoes and socksto black. It'll just do like a quickgeneration here. Um and the reason I dothis in steps is it's just a lot easierfor the model to like listen to onething at a time. You can do two or threethings depending on what it is, but Ifind that this is just going to get youa lot better results, especially whatI've seen like working with other AIcreators and my students is that a lotof times when they're getting reallytripped up and they're like, "I don'tunderstand why this isn't working." It'scuz they're just like expecting a lotout of the model. So moving in steps.Yes,

00:16:18over-prompting. Yeah, so moving insteps can be really helpful. Again,>> gone too far into Flora yet, but justthe way that you're laying things out inthis canvas everything's so spatial andyou get your iteration tree. Kind oflike it. It's pretty compelling.>> Yeah. Okay, here's a good one. Umanother example too is like this image Iended up like getting correct correctlygot in a nanana um after a lot of likeprompting and trying differentreferences and and figuring out whatreference would make it kind of likelisten to the vibe that I wanted. But inthis case, you know, again we have thatkind of AI perfect matching um and Iwanted to avoid that.So I gave it um a shoe.Let me see if I can find this shoe.UmOh, and for this client we didn't wantto have any logos, so I just told it toremove the logo from the shoe verysimply. And then I just funnel it backin um and I say replace the man's shoes.Um I think like keep everything else thesame, and then same for the next thing.I didn't want them to be supermatchy-matchy, so I replaced the man'sshorts with a pair of gray shorts that Ialso removed the logo from. So, workinglike in this very simple, verycontrolled, very calm, step-by-stepprocess is

00:17:33going to save you a lot ofstrife and give you yeah, just morecontrol over the model and what you'reworking with versus like kind of likerolling the dice and throwing like eightHail Mary's on it and hoping that itlike understands, you know, the shoesare white, but the shorts are gray, andand don't change the woman, but changethe man. Over-prompting is like whatkills generations, in my opinion. Yeah,this is way more granular thanI would have done out of the box.It kind of then makes me wonder likewhen you reflect on your journeylearning these tools and making mistakesand figuring out what pitfalls to avoid,are there any other ways that you'reable to identify a clear shift in theway that you approach these tools ordifferent lessons learned that you thinkpeople can benefit from? Yeah, I thinkfirst of all, like any creative process,you sort of grow an intuition for likewhat isn't working. Um, like forexample, with these images here, like Iknow uh because I've done this like justhundreds, probably hundreds of thousandsof times now, that the reason we'regetting this really cool like rainbowgradient is probably because of thispicture. So, if we take that pictureout, umwe're probably going to get less of thatstyle. And same thing for this like

00:18:49really nice kind of like almost PNGeditorial like gray background, we'regetting thatfrom this image here where there's likeno shadow and it's placed on. So, Ithink intuition and just like AI, a lotof peoplemorally, ethically could talk a lotabout like is it art? Like what does itmean for creatives? I'm definitely inthe camp that it is, you know, an artform and that you can't create the samething that I can create, you know, fromscratch, essentially. Um if you don'thave my process, if you don't have myreferences, if you don't have the waythat I prompt, if you don't know theorder in which I do things, you're goingto have like a really difficult timegetting it, which is why, you know, Ihave a course on this. Uh I was almostfor a while in the camp like, yeah,maybe it's not art. Like And yeah,you're right. Like anyone can recreatethis until I started posting my images.I was like, oh, like nobody can createthis. Nobody knows how to do it. AI isjust like any other creative process.Like you have to get good at it. Youhave to know what you want. You have toknow like what tools you prefer. Youhave to practice. It's not somethingthat, you know, you can get itovernight. Now, is it easier to get goodat it than painting or the violin? Yes,obviously. But it's still a creativeskill, which is why the people that wesee that are really good at AI uh happento be designers and creative directors

00:20:05or people that have spent a long time uharound photography or whatever they'retrying to make. You know, we're notseeing people that are just, you know,fresh off the street, uh you know,developer types or business types likegetting really good at it uh withoutpractice. Another thing that's reallyhelped my process is like understandingthere is a process. I think like many ofus, uh and as you were saying a littlebit earlier, is like you didn't realizehow granular it was. And I also didn'trealize either. I think I was treatingit up until even just a couple monthsago like a slot machine. And like notreally understanding how to likereinforce what I wanted and telling itwhat I wanted. But when I realized thatthere were actually steps that I couldtake like 80% of the time to kind ofsketch out like, okay, these are myreferences and this is my mood board.And this is how I want to prompt. Andhere's how I should approach likebringing in these profile codes. It madethings feel a lot less stressful and alot like I was just like praying and andand getting randomness. And now I feel alot more confident too that, you know,if I'm getting a client and and wantlike a specific style, I'm not like, oh,I've never done that. I don't know if Ican do that. I know that there's likethree to four different paths I can takefrom any point to kind of try to nailthat style, whether it's um like I'veshown you guys with the S reps,

00:21:20whetherit's maybe starting in Niji Banana Probecause Midjourney is not that good atphotorealism. Maybe that starts withlikecreating a better personalization code.Um there's a lot of different leversthat you can pull, but knowing whatlevers to pull and when and for whattask is where you kind of start toreally be uh an or an orchestrator ofthese tools. Yeah, it makes sense. Andhonestly, something that you said thatresonated was thelike slot machine metaphor. Because tome, as someone who's quite novice in allof this, it has felt like a slotmachine. Yeah. It's felt like a slotmachine and it's felt likeI didn't haveenough language under my belt in orderto achieve any semblance of consistencyin Yeah. You know, I'm even thinkingabout you know, brand building andcreating this universe around like thisproduct that I'm working on andI don't know, I haven't went as far intothis world as I would like because Ihonestly just didn't think that I wouldbe able to consistently achieve a stylethat worked. Consistency is probablythe marker of someone that's actuallyvery good at these tools. I think a lotof people on X especially go viral formaking like very beautiful like art orlike fantasy stuff, but the second thatyou try and like apply that to,

00:22:36youknow, a startup brand or, you know, abrand that maybe is like lifestyle, theyhave to showcase a lot of differentaudiences that want different things, itcan start to fall apart. So, having likeelements that you're able to swap isalso helpful, too. So, something that Ido often, um I have a many mood boardsthat kind of speak either to moregeneral stuff like like this mood boardI tried here or like specific functionsofyou know, fromSo, in this case I have like what isthis called? Real skin. Um so, this isall textures of real skin and wrinklesand freckles and stuff like that. So,what I can do is, you know, when I havethis style here, this like very kind ofedgy, blurry style, if I want to addthis real skin in, it'll kind of helpreinforce that even though I have thisstyle set. And this, of course, isalways an experiment, so we'll see ittogether like what happens. But, that'slike one way to get consistency.There's one question that I can't stopasking myself. What if companies appliedto talk to you rather than the other wayaround? And that question is thefoundation for the all new Dive TalentNetwork, and it's working. Like rightnow, I'm helping many of the mostexciting startups that I know to hirethe designers and builders who listen tothis

00:23:51show. So, if you're curious whatmight be out there, and maybe you wantto get on my list, or maybe you're evenlooking for your next design hire, headto dive.club/talentto join today.Is there an example that you can pointat to show us what consistency and theskill of consistency looks like inpractice?Yeah, so I think something that's likereally important [music]for designers that are starting to getto get good at this stuff is like, okay,great. I can make like really beautifulstuff. It's really cool. It's cool inthe brand presentations, but how do Iactually give this to a brand to use it?And to be honest, I think this is goingto be like a tool space that isevolving. Right now in Midjourney, forexample, there is no kind of like shareprompts or like share mood board. Thebest you can get is like having thebrand create a Midjourney account, andthen you log in, do a bunch of stuff,and then that's part of the deliverable.But, currently,what I try and get to is not necessarilylike a one-shot.Um well, you could maybe call it that. Itry and get to like a master prompt. So,that's like all the parameters that Ihave, the reference photos that I wantto use, maybe it's an image reference aswell. Uh and then also any profile codesI'm using. Since we're going to havereal match path. It's going to be comingout probably in like May or June.And I was able to create all of theseimages with this master prompt here. And

00:25:06the great thing about this is while, youknow, sometimes I might want to useslightly different prompts, longerprompts, maybe pull in like morespecific mood boards to enforce acertain style. If they want a picture oflike a very specific model, you know,maybe they want to add like a skin moodboard to it. What they have is just aprompt that you can literally copy andpasteor, you know, use in Midjourney. Andthey can say like one word and it willget them something that is likeprobably pretty good. So, I'm writingcat right now.But this is an example. Literally myprompt is just dog.Mhm. And it's likeI mean, you know, this one's not thatgreat. This one probably wouldn't makeit into the brand. ButI I think this is fantastic. You havelike the cool editorial lighting. Youhave theuh the plain background. It's even likeextremely detailed already before weeven upscale. The fact that you got thatfrom cat and that from dog is giving mehope right now.Yeah, I mean, these are insane. Like howwould you even prompt this? Like Yeah.You know, even if you ask me>> that's the key. I I think this is likeone of my biggest takeaways is you'releaning so far into references andimagery. And that's been the aha momentfor me because I've just been sointimidated by my inability to

00:26:21describethings. Like you were you brought up theoriginal picture.I wouldn't even gotten half as far asyou got in the baseline. And but I canspend too much time on Pinterest. LikeI'm capable of that. And it it reallycompounds like over time. Like here isan example of we're using the sameprompt here. So, live interface isprompting fingers touching future sixscreens. I think I was justexperimenting here cuz I think thisprompt is cool. And this is using all ofthe parameters. I think all the the formatch path and the S refs. And as youcan see here, almost no matter what wedo, we're getting those qualities, thehigh contrast, the like blackbackgrounds, the rainbow. And then if weuse that same prompt, in this case withjust without the references, you know,we're getting a different style, likestill cool,um but these mood boards that I'm usinghere, and these profile codes don't haveall that rainbow in the background, andthe contrast, and all these like uniqueelements. That's like all I've I've justbeen saying that to the podcastcommunity. If you If you have to takeone thing from uh anything I post aboutAI or this podcast or my course, it'sjust stop prompting and start usingreference photos. It'll save you a lotof time, a lot of credits, and a lot ofstress. I want to zoom out for a second.How

00:27:37far have you leaned into AI in terms ofhow you're even positioning yourself?It's one thing to know how to use thesetools, and it's another to have successin the competitive market of independentdesign. So, help us cross that jump.Like, what are some of the things thatyou've been thinking about over the lastyear or two even as you figure out howto evolve your own business? Yeah. So, Ithink for the next probably like 6 to 12months, maybe touting yourself as likean AI designer will become official,especially because the people that areinterested in AI tools right now, or AIdesign work, they're interested in itbecause they know about it, and theyknow it'll allow them to do things thatwere either previously impossible, or,you know, would save them a lot of moneyand time. So, like people that want AIproduct photography, they just want tolike not have to pay a productphotographer to do that. Or, they mighthave the constraint where it's like,"Hey, we're getting our packaging fromChina. It seems to take 6 weeks, but Iwant the website up tomorrow. Can you dothis?" And before AI, you know, maybe 3Drenders and stuff like that, it could bedone, but with AI product photography inparticular, it's extremely easy, and youdon't even need a 3D render to getstarted with that. So, I would say ifyou are using AI right now and you'revery good at it and you want to sell to

00:28:52clients, it's not a bad idea to mentionit somewhere in your like headline ofyour site or maybe even in your bio.But I think eventually, like anytechnology, saying that you're an AIdesigner is going to be as silly assaying you're like a food chef. Well, ofcourse. You know, it's like saying I'm aI'm a Figma designer. Well, Figma's justa standard now. You don't even need tomention it anymore. Basically, I thinkAI is going to become the standard.We're already seeing that in productdesign, like way more extreme thanbranding. I mean, I'd love to hear yourthoughts on that, too. I think mythoughts are that you're spot on thatthere is a window in time right now. AndI think that almost every application ofAI does have a window in time and it'shas varying degrees of length, you know,like and I saw that in product design100% where people were putting AIdesigner. AndI think looking back, it looks maybe alittle bit silly, but it worked. Like ifyou were able to convince somebody thatyou are an AI native product designer,you've had some success in the marketover the last like 6 to 12 months. And Idon't even think we're fully at the endof that cycle. I think even looking atyour own success on Twitter,you've obviously crushed by being earlyto this, but the window is not closed. Ithink because there's so few people, Imean, there's probably only 100

00:30:07peopleactually offering like AI services.And that's pretty generous. But I havepeople ask me all the time, like, doclients actually ask for AI work? Andit's like, well, yes, sometimes rightnow,but also they don't they don't carewhere it comes from. They don't care ifit's AI. As long as you can get it tothem for better, cheaper, and faster,that's what that's why they're going tochoose you. For someone who's listeningwho's inspired by the things that you'resaying and, like me, they agree there'sthis moment in time where you really cancapitalize on these skills and maybethey want to be like, you know, one ofthose 100 or so people that are activelymaking a living bringing this kind ofwork to the table. You do have a coursefor people that want to go deeper. So,I'm curious if you could share a bitabout you know, what is it? Who's itfor? What can people expect to get outof it?>> Yeah, so the course is called the AIcreative director, and I basically madeit to be a step-by-step system forpeople that want to make consistent,high-quality, sort of like client-readywork, whether it's, you know, for yourown internal team, or if you're afreelancer. Something that can actuallygo out into the world and not look likeslop. Kind of the impetus behind thiscourse is like, I had people, even likeyou, Brit, that have like been in thegame

00:31:22for like much longer than I have.I'm like 23, you know, I've I've beenworking a long time, but, you know,people that have done some of like thebiggest startup brands in the world, orlike worked on campaigns for Nike andall these places, they were commentingon my AI posts like, "How did you dothis?" And I'm like, "What do you mean?Like, you actually are like a workingcreative director uh for these hugethings, and you're you're so like rawlike so much raw talent." After kind oflike looking into it, I realized theyjust don't know how to use these tools.I think they've been led astray, kind ofby like the tools themselves. Like,obviously, when they're marketing, theywant to, you know, say, "It's super easyto use. Like, it's just one click." Andthen you have kind of like what I callthe prompt bros, kind of mirroring thesame thing. Like, if you just buy mycourse and follow my process, you know,it'll be so easy. And I think that'slike totally a myth. Like, I I'vementioned so many times now, like, I Ihate prompts. I don't use them thatmuch. So, yeah, it's a course meant forpeople that have like good taste, andprobably a creative background, orsomeone that's like been around creativeworkflows a lot in the past. It's notsort of like a design101 course. I also tried to make itpretty tool agnostic, so it's not evennecessarily like a a model 101 course,cuz I'm hoping that, you know, ifMidjourney decides to like change itsentire thing, or

00:32:37a new model comes up inthe next like 6 months that, you know,these skills are are things you canactually use. It's also has like fourlive sessions, too. So, like I encouragepeople to like just come with like yourquestion and like show me your promptsand we'll walk through it together, andI think that's where kind of a lot ofthe value comes in cuz everyone's usecase is different. And then, hopefully,I'm going to train an army of like acouple hundred people in my exactprocess that are really good at AI, andthen I can, you know, offer us as like asort of team package to companies thatwant to create a lot of these assets.Hey, real quick, if that sounds likeit's up your alley, then I asked Jamieif we could share a discount code forDive Club listeners, and she agreed. So,just head to dive.club/jamieto get the special offer. That'sdive.club/ja m e y. Okay, back to the episode. Atakeaway that I have from just watchingyour process is, yeah, there'sdefinitely tool mastery and lessonslearned, and you're understanding how totake different steps to achieve thisconsistency, but at the end of the day,it really is just good creativedirection. Like, you understand whatmakes a good brand, and you put a ton ofthought into the references. Like, thereferences, you don't get those forfree, you know? References come froman understanding of what makes greatcreative direction. And so, if anything,

00:33:52I think this has been a giant salespitch for the value of somebody thatunderstands brand, and that AI isn'treally going to swallow that anytimesoon. No, I mean, until AI can gothrough Cosmos and like pick the imagesand the visual direction for you, whichI thinkcertainly there will be a time and amarket that that that would make sensefor, you're going to have to be acreative director in order to get thesekind of results. Like, the better youare at the visual stuff, the better youare going to be commanding AI to makethat visual stuff. And like all thesereferences I have here for this MagicPat, you know, style, a lot of thesecame from my mood board. Like, this isfrom Pinterest here. And then I doubledown as I started to create images andgot a better sense of like, okay, youknow, this prompt is working.I can show you guys. I start to kind oflike dog food my own style. So I putimages that I made into the stylereferences to like reinforcethat style. Um so this one was one thatI made in AI like very early in theprocess and then I loved it so much thatI was like, hey AI, I want everything tohave elements of this. And thatobviously was like super fruitful forconsistency and then

00:35:07also just nailingthe vibe that I wanted. Is that prettytypical that your output would slowlytake over the initial references? Yeah,especially for stuff that's like veryabstract and and and juicy and verymidjourney like this, definitely. That'sjust the best way to like reinforce yourprompts is to give it stuff that you'vealready made that you like. And thenalso too, like for anyone that'sconcerned about like, oh, I'm usingreference photos of like,you know, it's not my work or the stockphoto, I don't want it to be generic orI don't want to like maybe be copyingsomeone. As you start to do this, youalso 100% go to your own unique style.Like this imageis solely mine. Like this was not taken,you know, from Pinterest or anything.Same with this. Like I made this imageand this one. So now, even though I'musing style references from Pinterestum or stock photos, I'm pulling in myown style as well. So it is totallyunique to me and then also the clientsget that benefit, too, that they're not,you know, potentially looking exactlylike some other brands. Cuz it'simpossible. I know the topic of taste ispretty wishy-washy and a little bitharder to talk about sometimes, but youknow, I look at that photo with therainbow gradient cubes

00:36:23and>> Mhm.I mean, that's taste, right? Like it'snotIt's not obvious to me that this wouldeven be great, you know? Like, you haveto have an understanding of what you'retrying to accomplish and how you want tostand out. And I'm sure I'm not alone inlooking at that and being like, man,yeah, I have a little bit of room togrow there in terms of growing thismuscle of of taste and bringing more tothe table in that regard. And I knowit's difficult to talk about, but like,how practical can we make this? Like, ifsomebody comes to you and you're like,what are some steps that I can take togrow my own taste and invest in this asa very core part of how I'm able todifferentiate myself as a designer? Doyou have any advice for that person?What I recommend to anyone in studentsof my course is, you need to have adaily taste practice, and taste has tobe a part of your life. If if you wantto be someone that makes visuals for aliving, you you got to think about thethings around you, the things you'reconsuming, even the things you'rewearing, the things you're eating. Umjust anything that you consume, youknow, visually or culturally. Andthere's a couple things that I do. So,like, number one, I'm like always moodboarding. You can see here, I have likethe Pinterest plugin, I got the Cosmosplugin. I have it downloaded on myphone. So, anytime I'm like shopping,

00:37:38oranytime I'm even on X, I'm likeconstantlyable to just like quickly share images.Um So, let's see. As you can see, like,yeah, here's Joe's courses. I can savethis frame to Pinterest with the littleplugin that just popped up. Or eventhis, if I, you know, this is a someonetalking about their uh protein icecream. This is not a design post, butthe design's really cool. So, I'm goingto save that as a design. And then, whatI do as well, as something I used tostruggle with all the time, was I usedto have like really specific moodboards. Like, here's my like UI designmood board, and here's my vintagebranding mood board, and and here's mylike winter fashion mood board. Butobviously, things overlap. What if it'slike a light jacket, does it go in fallor winter? I like to just separate themmore into like way bigger categories.So, like I noticed like over time I wassaving a lot of things to my designinspiration that could probably just beits own mood board as productphotography. As you can see here, Iprobably need to switch the system up alittle bit cuz I have 7,000 sideinspiration bins.Uh maybe I need to do like a yearly one,but I noticed okay, like I specificallywant to look for just productphotography, just logos, so I split itinto that. But otherwise, I try and notto overcomplicate it because thatdecision fatigue when you're savingstuff is going to

00:38:53prevent you fromactually saving it and just like gettingthat into the archive. So, that's mostimportant. So, that's kind of about thefunction of it. So, you got to find aplace where you can save yourinspiration. My mind, Cosmos, Pinterest,you know, even Notion, you know, ifyou're saving maybe writings, too. Andthen next, I would say I think I tweetedthis a while ago, but the worst thingyou can do as a designer is learn moreabout design. And I think personally,like some of the best things I've doneto help me in projects, like you canthink of the randomest things. Like Ione time had like a handbag project thatwas like really alternative and kind ofedgy, and I had just happened to watch aYouTube video on this fashion show likeLanvin the rich premiere and all thesecrazy brands, like super edgy, likeheroin chic aesthetics. And had I notbeen watching these fashion videos, Iwouldn't have had the like referencerepertoire to be able to execute like sogreat on this like edgy handbag project.And that works like all the time. Like Ilove to cook, I watch a lot of cookingvideos. If I ever got like a cookingbrand, I'd be able to know like, "Ohyeah, like Julia Child or maybe know whoAnthony Bourdain and I have this kind oflike cultural reference that helps meyou know, either very specifically. Andthen even more broadly, too, like as I'mmaking AI prompting, like watchingmovies, like knowing like, "Oh, like Ireally want this like kind of quirky

00:40:08feel." I'm going to say like WesAnderson, like Wes Anderson style. Ormaybe I'll find a Wes Anderson frame toput into the prompt as a reference or animage prompt. Versus, you know, if Ijust like toiled away only savingPinterest design inspiration and andnever tried to like go outside of thebox or something that happened to me alot is I I had goals. I wanted to becomea better designer. I'm just going tolike design design design. But actuallywatching movies and you know, maybetaking breaks or traveling has actuallygiven me that kind of like space andinspiration that has been like verytangibly useful to me. So, I would saythat. Like if you are maybe like adeveloper guy or like yeah, a girl thatcomes from like business or marketing orsales. Like I had a student reach outthat was taking the course and asked mefor books on like being a creativedirector. But I didn't really have anygreat recommendations. I you know, sentthem some like Virgil Abloh videos, butI think any book that you read on acreative director is not going to be asgood as like reading just goodliteratureor watching a good movie or, you know,making a playlist and tapping into theSpotifys and learning about the artistsa little bit and checking out theirInstagrams and seeing how they brandthemselves. Like that is where youactually get good at creative direction.Um like finding those references more

00:41:24sothan reading a book that's like tellingmaybe telling you to do that probably. Idon't know. I don't read creativedirector books. But that creative act umlike Rick Rubin. It's very If you'vetried to read that book, it's extremelymeta.It's kind of a lot of nonsense anywayand it's I think he kind of reiterateslike it's about you. It's about yourtaste. Like I'm making what I make is sogood because I like it. And I have allmy life experience like putting intothis thing. Yeah, so there's no textbookon being a creative directorunfortunately. You have to just be likea cool personand creative.And you have to put in the reps. I mean,the biggest thing that I'm>> Oh, yeah. honestly taking away is thisall pins. Like you have over 15,000.That's a really big number. Like you'vesaved a lot of stuff.>> I think another great example, too, iswhen I first moved to New York, I wouldsee all these like influencer girls, andyou know, they're like incrediblefashion, incredible Instagrams, and I'mlike, oh, like how did they do that?Like are they just like rich, whatever?They're just like born with it. And thenyou spend time around these girls, yougo on vacation with them, it's like, no.When they're making an Instagram post,like they're spending 5 hours doing it.They're They're posing. They're watchingvideos. They're scouting locations. Likethey're spending time shopping. Likenothing comes for free, basically. Andthat's the same thing with like reallygood creative

00:42:39directed images, whetheryou're making them for real or in AI. Umyou got to you got to put in those thattime and that reps, and it does pay off.It It's slow. It's kind of like the gyma little bit. I definitely noticed likeeven just with my own like fashion, forexample, likeyou know, I'm kind of like coming of ageright now, so I think it's natural thatlike my style is evolving, but I've beenlike really focused on it for like 2years, and I'm just now getting to thepoint where like almost anything I pullout of my closet I love,but it's still always an effort. Andthat's the same thing for design, whichI'm sure you feel as like a productdesigner as well. You know, you don'treally hit your stride for like 5 or 6years, I feel like. It's very easy tokind of go from like being a Canvadesigner to being a Figma designer, andlike you know, being better thaneveryone you know, but going from likebeing good to great takes like forever.And it's because it's all about thetaste and and stuff. You kind of likelive and breathe>> And And you've got to have enough repswith the bad taste to get it out of yoursystem, too. Like I'm sure you couldscroll back in those 15,000 pins, andyou probably hate that first 20%. Like Ido that I have a Notion database, and Ihave>> Yeah. a ton, not 15,000, but I have alot of UIs saved in there. And if Iscroll to the bottom of that list, ormaybe I apply like too specific of afilter where I get something from

00:43:54like 5years ago, and I look at this, and I'mlike, that's trash. I would never savethat today. In some sense, it almostfeels good cuz I can sense growth alittle bit, but you really do just haveto put in the reps. Yeah, this is fromlike 104 weeks ago. So, what is that?Like this is like barely 2 years, 3years. This is the AI I was making.You know, not very long ago. Part ofthat is obviously the tools improved,but most of this islike my understanding of it and mytaste. And I thought this was good.Something I was talking about before isbeing radically honest with yourself.Because I had never used AI before, I'mlike, "Oh my god, I was able to createan image that it kind of listened to me.It like looks kind of like what I want."I was like, "This is so cool." Like,"Wow, it's like neon and it's likeoverlaid." But, like if you're honestwith yourself, you're like, "No, thislooks like Like keep keep trying.Don't settle." Yeah.>> Imposter Imposter syndrome is deservedmost of the time. I think the lack[clears throat] of imposter syndrome isstagnation. And that if you don'texperience it imposter syndrome everyonce in a while, then that's actuallyquite a negative sign. You should belooking at other people's portfoliosevery day and comparing yourself. Ithink it's a reallygood practice, whether it's with like AIimages or even just like branding stuff.Like take a screenshot, like go

00:45:09fullpage, that that plug-in. Take ascreenshot of like a big agency's casestudy and then put it next to your ownand you will quickly see you are notthat good. You have a lot of work to do.And that doesn't come from, you know, itcomes from like with brand designer logodesign, like literally putting in thereps, physically knowing how to like doit and what looks good. That takes time,but the references you're using, themood boards you're using, why you'remaking the decisions you're makingmatters a lot more, which I know a lotof junior designers don't want to hear,but sometimes timeis like the only thing that's going tomake you better. The one thing that Iwill say that is very pro juniordesigner though is, my god, you can makereps much more quickly than you couldyears ago. You know, we're not doingpainstaking things in Photoshop fordesign, you know, it's not like we'redesigning apps in Illustrator anymore,you know, they're using tools like MagicPath where you can get reps 100 timesfaster than I could when I was a coupleyears into my career. So, I I I'm prettybullish on early career people that aregrowing up natively with these tools.Yeah, I mean I have no idea like whatthat's going to look like. Like I I knowthere's kind of probably not mygeneration designer, but a little bitolder. Um like people in their 40s and50s, that kind of designer that was likemaking the first like apps, you know,early Facebook people, early Shopifypeople, whatever. They're such aspecific kind of like builder, tinkerer,artist, quirky

00:46:24personality. I'm I'mreally interested to see like what itlooks like, what the designer role lookslike in the future cuz I think thatwe're kind of converging. There's goingto be like three jobs. There's going tobe like ideas guys, like CEOs, likepeople managers, you have this onecalled like finance and stuff like that.And then there's going to be brandpeople. So that's anything that you see.So marketing, websites, even events,logo design. That's all going to be justlike brand. It's like my job in thefuture, you know, I I do brand strategy,I can do a little bit of marketing on10x, but like part if I were to gethired, you know, in 5 years, I think Iwould likely be, you know, doing thebrand design, probably running socialmedia, doing the marketing strategy,running community, whatever that lookslike. And the third job is just going tobe product. If you can click on it, it'sgoing to be like one type person's job.There's not going to be engineers anddesigners anymore, PMs anymore. It'sjust going to be one person. And they'regoing to managethemselves and, you know, the PRDs andthe road maps and the development andthe front end. I think I agree. I wasjust texting my co-founder a few hoursago and we're like, is this just productbuilder? Like is that the title? Likewhat is happening right now? Something Isaw on Twitter recently that I thoughtwas quite interesting is

00:47:40they werehiring for a launch designer.And it was this>> I saw that. Right? It's very It's likeas soon as I saw it I was like, oh, thatmakes so much sense cuz you're right.You're owningall of the visual imagery. You'reprobably owning a bit of motion. You'reowning the copywriting. You're owningthe strategy of like how are we What'sthe story that we're telling? Likethat's areal generalist role and I do think itkind of speaks to where this is allgoing, where it's just like, "Yeah, youyou're just going to have to becomfortable wearing a lot of hats." Ithink that's definitely the future. Andif you don't think it is, then you'regoing to be competing with people likeme that can do it all. You know, I mean,this started early for me like anydesigner that posts on social media, uhI'm sure you understand this idea. Youcould also theoretically be hired forlike media stuff. You know, someone likeFigma who probably has hired you. Imean, you get sponsorships, obviously.You know, they're hiring you to domarketing stuff. They're hiring you toreach an audience and and do strategy.Like that was like kind of like thefirst leg up is like the unicorndesigners. Like I'm a designer thatunderstands how to talk to people andhow to get attention. Uh and that's justgoing to like expand to be like, I knowhow to I'm a designer that knows how totalk to people and get attention and getpeople to buy and like keep themretained,

00:48:55you know, and and grow thisthis whole thing. Have you ever feltpressure to expand the team, to hirepeople underneath you, to go more of theagency route, andhas that changed with all this AI stuff?Oh, yeah. I mean, people I felt pressurejust from like seeing, you know, otherpeople's success having an agency andeven just like people in my own life.They're just like, "Why don't you scale?Like you could like take on moreclients." And I think for the longesttime I've been so against it becauseas a brand designer, it's really hard tofind a junior brand designer.Like if I was going to hire someone,which I actually now have like anassistant creative director, but in thepast like if I was going to hiresomeone, for it to be worth theirweight, it might as well be like 50% ofthe contract value. My contract valuesup until like a couple months ago weresometimes only like 10, 12, 15K. So atthat point like getting help would justmeant like splitting the project. Andyou know, when you're a freelancer onlydoing 10 to 15K projects, that lookslike 250K a year and splitting that inhalf just you know, doesn't make sense.Why would you freelance? Like what whywould you have the lifestyle for? It'skind of just a trap. Like there's a lotof happy agency owners. I would saymaybe 20% of them. They love doing thesales. They love having a big team. Theylove that they're able to like land bigclients,

00:50:10but 80% of people are not goingto going to get that. You know, there'sthe 80/20 rule for a reason. You know,they're not going to get the bigclients. They're not going to get thebest talent. There's only so many Aplayers out there. So you're going to beworking with B players. More money, moreproblems in a sense. Like if you're havemore people on your payroll, then youhave to get more clients, then you haveto have more project managers, andthere's more stress, and you have lesscontrol, and you basically just becomelike a manager essentially instead of adesigner. And often times you're alsomaking a lot less money. If you can geta 100k client because you have an agencyand you have enough to support like amotion design and a brand designer and aproject manager, you got to split thatproject, you know, maybe not evenlyamong them, but like certainly a lotversus if you were just really lean, andyou know, maybe have like one contractorhelp you or even just yourself with theeye and maybe now as a brand designertake like a 50k gig, you get to keepalmost all of that money. And then youdon't have to worry about who is on yourpayroll, and you don't have toconstantly have work coming in as well.See, I honestly think the agency kind ofnarrative is kind of a scam for a lot ofpeople because I'm certainly like aliving, breathing example of that. LikeI actually saw I won't like say theirname, so I think I saw someone kind oflike publish their like agency perhapsfor the year, like how much money theymade

00:51:25with like two people with likepretty cool, like big clients, and Imade more.And you know, I'm working with likesure, you know, I'm I'm not taking onlike the Google, you know, commercialprojects or whatever, and my projectvalues are like much less, and maybe Igot to I chase my leads down a littlebit more, but the money and thelifestyle for most designers, which Ithink a lot of us our goal is just like,I want to create cool stuff, want tohave a good living, I want to be able tolike close my laptop at some pointduring the day and hang out with myfamily or do my you know passionproject. And I don't think agency lifeis conducive to that for most people.Well, Jamie, I appreciate you coming ontoday and this episode has really movedthe needle both in like the the granularpractical tactics, but also just a lotof like career strategy. It's like verydifferent ball games about you know howto learn these tools versus how to makesense of the way they fit into yourjourney. You know, years from now and II I feel like you've shared a lot ofinsights on both fronts. So, Iappreciate you taking the time today.Well, I'm glad. Yeah, this is fun.Before I let you go, I want to take just[music] 1 minute to run you through myfavorite products because I'm constantlyasked what's in my stack. Framer is howI build

00:52:40websites. Gen way is how I doresearch. Granola is how I take notesduring crit. Jitter is how I animate mydesigns. Lovable is how I build my ideasin code. Mobbin is how I find designinspiration. Paper is how I design likea creative and Raycast is my shortcutevery step of the way. Now, I'vehand-selected these companies so that Ican do [music] these episodes full-time.So, by far, the number one way tosupport the show is to check them out.You can find the full list atdive.club/partners.