Latest / The Koerner Office - Business Ideas and Deep Dives with Chris Koerner / Missed Every Gold Rush? This One Started Just 2 Days Ago | Ep. #161
Transcript
- 0:00The image generation API was just released by Open AI.
- 0:03We're about to see a gold rush. We're about to see a lot of
- 0:05people make a lot of money very quickly.
- 0:06Ads are the perfect output for image generation because they
- 0:10make people money. These marketers, they will pay
- 0:12you a lot of money. It went from unusable but cool
- 0:15to usable really quick. Those things are going to do so
- 0:18many millions of dollars in like the first couple days.
- 0:20It's crazy. What are the biggest
- 0:22opportunities in AI image generation today for someone
- 0:25that has nothing but Internet connection and a laptop?
- 0:27There's like 1,000,000 apps that can be built like that.
- 0:30One of those is going to do incredibly well.
- 0:35All right, I have a super interesting guest today.
- 0:37His name is Jacob Posel and he flew from Australia to San
- 0:40Francisco to build a company inside a company.
- 0:43So he met a guy on Twitter that owns a software business in San
- 0:46Francisco and they want to build AI tools.
- 0:48So they hired Jacob to build tools to fix the problems that
- 0:51they had within their companies, but also wanting to eventually
- 0:55scale and sell that tool to other companies as well.
- 0:59O this guy is an AI hacker at heart.
- 1:01He's especially good at Chad GT's new image generation tool.
- 1:05And as luck would have it, the image generation API just became
- 1:08available today, so you will be hearing business ideas for the
- 1:11first time with technology that was never available before
- 1:14today. I got so fired up, I literally
- 1:16had to buy a domain name as he was talking to me.
- 1:19And I'm going to build at least two different software tools
- 1:22solely based off this conversation.
- 1:24This new image generation API is a defining moment in Chat GPT's
- 1:28history. It will enable so much
- 1:30entrepreneurship. So if you want to hear from
- 1:32someone on the cutting edge of this, we talk about how to build
- 1:34an agency, how to serve ad agencies, how to build ChatGPT
- 1:38wrappers with image generation at its core.
- 1:40You're going to love it. Please share with a friend.
- 1:42Let's get right to it. What are the biggest
- 1:44opportunities in AI image generation today in your
- 1:47opinion, for someone that has nothing but an Internet
- 1:50connection and a laptop and a lot of time?
- 1:53There's two ways to take this that can create a lot of money.
- 1:57And like, OK, so today is the 23rd of April and the image
- 2:01generation API was just released by open AI.
- 2:04And so we're about to see a gold rush.
- 2:05Like we're about to see a lot of people make a lot of money very
- 2:08quickly. And so I would release this
- 2:10quickly so that these people can take advantage of that.
- 2:12The API was released today and there's lots of different ways
- 2:15you can take advantage. So a lot of people who are not
- 2:17technical and who are not nerdy and who are not on Twitter don't
- 2:20know what chat ChatGPT is capable of.
- 2:23I was at an event with some friends recently and we like
- 2:27took a group photo and I was like, oh, I'm going to like
- 2:29giblify this image, which is like a super common thing.
- 2:32And they were like, I have no idea what that means.
- 2:34And I showed it to. Them.
- 2:35And you're in San Francisco talking to people your age,
- 2:38right? Yep.
- 2:39Yep, exactly. They're in the finance world,
- 2:42not in the tech world. And they just, they just didn't
- 2:44know. And there's like 1,000,000 apps
- 2:47that can be built like that, that take whatever, whatever,
- 2:50you know, everyone is loving that Chang CPT does and just
- 2:54turn that into a super focused app like Ghibli me or whatever
- 2:58you want to call it, or like anime me.
- 3:02Like one of those is going to do incredibly well.
- 3:05And so you can charge for it based on like you can, you can
- 3:09figure out the the math to make it make sense where you sell
- 3:12like a subscription and then maybe credits on top to generate
- 3:15more images, look for some virality using the tool like
- 3:18TikTok or something like that. And like those things are going
- 3:22to do so many millions of dollars in like the first couple
- 3:24days. It's crazy.
- 3:25Like it's that's coming. And there's a lot of other
- 3:27really cool use cases, like 1, which gets people really excited
- 3:31is you can combine two people and be like, what would my kid
- 3:34look like? You know, like the system is,
- 3:38yeah, it's like I did that with me and my fiance and I took a
- 3:42photo. I was like, what would our kid
- 3:43look like? And we loved it.
- 3:45It was like, we were like I joked that this is going to
- 3:48solve the fertility problem. You know people are going to
- 3:51want to have kids once they see these.
- 3:53We're all narcissists, right? So if we want to solve the
- 3:56fertility problem and ChatGPT has to make us look better than
- 3:59we actually look, and then we're like, oh, that's a beautiful
- 4:01baby. Let's have some kids.
- 4:02And the kids do look super cute, like it doesn't matter who it is
- 4:06and stuff like that. Like be creative there.
- 4:09Like see what? Like just like off the top of my
- 4:11head, like what am I going to look like when I'm older?
- 4:15That's like another one that you can do when it comes to like
- 4:17feeding into people's Gnosticism and arrogance.
- 4:19Like people want to look at themselves.
- 4:21Like what would I look like with different hairstyles with the
- 4:23beard? What would I look like if I lost
- 4:2430 lbs? Right.
- 4:26I mean, 2/3 of Americans are overweight, so there's your
- 4:28market. Amazing.
- 4:30That's an amazing idea. Yeah.
- 4:32There's like, so many of these, like, feed into people's
- 4:34narcissism. People love looking at versions
- 4:35of themselves. People love taking photos of
- 4:37themselves and seeing what you look like in different under
- 4:40like a different context, like a different in a different
- 4:43environment. So that's like something that is
- 4:46a no brainer. There's also like think of other
- 4:49industries where imagery is important.
- 4:53So like, like for retail, for real estate photography, a
- 4:59really cool 1 is like take an empty room and like fill it up
- 5:01with furniture for me and like help with interior design.
- 5:04Another one is like improving the lighting or the vibe of like
- 5:07an AB, like an Airbnb. I posted about that one and
- 5:10people hated it. But there's got to be a, there's
- 5:13got to be like an ethical way to do that.
- 5:15But yeah, there's like there's so many of these, there's so
- 5:17many ways that we interact with images in our life.
- 5:20And any of that is going to be very important to people.
- 5:24And we'll use this API. So that's like the consumer
- 5:27virality side of it. And I think we're going to see a
- 5:30lot of really creative people come up with a lot of really,
- 5:32really creative things. Jacob, I have to apologize
- 5:35because I told you before I hit record that I can't not launch
- 5:40businesses. So when I said that you know
- 5:42what it what would I look like if I lost 30 lbs and I got your
- 5:45validation that you liked it. I literally just clicked
- 5:48purchase on right size me.com. That's a good domain, isn't?
- 5:53It amazing, amazing. Sizeme.com How much was it?
- 5:56Done. 10 bucks it was open. Amazing.
- 6:00I own it, but like that's totally something that people
- 6:03would would pay for and then they would share and then like
- 6:06that literally wasn't available yesterday.
- 6:09Like it, it would have been very, very hard to build that
- 6:11yesterday. But today the API came out and I
- 6:14think it's hilarious because six days ago I posted a tweet
- 6:18showing like pictures of homes in Texas of like, what would
- 6:22this home look like if it were painted white?
- 6:24What would this home look like if a if a large tree fell on it?
- 6:26And I owned a tree trimming business and I wanted to send
- 6:28personalized images of people's homes with a tree on it to
- 6:32incentivize them to pay for my tree trimming services.
- 6:34And half of the negative comments were like, APIs don't
- 6:37even available. And I'm like, give it a week.
- 6:40That's what I said. I'm like, give it a week.
- 6:42And here it is five days later, and it's available.
- 6:44Like, people look for so many excuses to not launch.
- 6:48They like, oh, there's an excuse.
- 6:49Yeah, I feel better about not launching now.
- 6:51Yeah, yeah. It's so true.
- 6:54People look for a million reasons not to do something.
- 6:55But if you had been preparing for it, you, you could have
- 6:59created like a really, really great outcome for yourself
- 7:01starting today because it was launched today, which is which
- 7:05is pretty amazing. Have you seen any image
- 7:09generation wrappers or tools? I know it just came out today,
- 7:12but have you seen anything that just started printing money
- 7:15right off the bat regarding image generation?
- 7:19No, I mean unless you count on Jet GPT, but like it literally
- 7:22was launched, I would say two hours ago, like literally 2
- 7:26hours ago. So it's like too early to have
- 7:29seen anything that could have. Launched, but what about like
- 7:31with replit or the other a image generation AP?
- 7:35Is the open AI image generation API was like chat, like the GPT
- 7:404 moment, like if people are familiar with that when it comes
- 7:43to image generation, it went from unusable but cool to usable
- 7:47really quick. And so none of the other AP is
- 7:50were were viable to do something like this.
- 7:53And these ones are viable to actually to get past that like
- 7:57churn point. You know, like a lot of AI apps
- 8:00have really high churn if it just can't get past a certain
- 8:02point. And we're there now.
- 8:04And so no, I haven't seen any. But like, I guarantee we will
- 8:07see a lot in the next coming days.
- 8:09Tell me about what you were playing with a couple weeks ago
- 8:13with these ads, like with the image generation tools and now
- 8:17that the API is available, what businesses do you think people
- 8:20people could start with what you were playing around with on
- 8:22Twitter? Yeah.
- 8:24So this is a big one, which is ads are the perfect output for
- 8:29image generation because they make people money.
- 8:33And so I had been playing around with this idea of AI ads for a
- 8:37long time before the image generation API, before the image
- 8:40generation tragic PT was even a thing.
- 8:43This like new update. And then when it was like, Long
- 8:46story short, it was like the first thing I tried because it
- 8:48had something I I had obsessed over and failed at for a long
- 8:51time before. And I had this like wow moment,
- 8:54like, it's there, it's here, it can finally do it.
- 8:57And I started posting about it and that kind of got a lot of
- 9:00traction, A lot of people impressed.
- 9:02And so I think we are going to see a big influx of AI generated
- 9:05ads and there is a very high demand in performance marketers
- 9:11to figure out how to use these ads, how to use AI to generate
- 9:16more ads for them. Because to like simple
- 9:20oversimplify, the ad algorithms are starting to reward volume
- 9:24and choice more than high focus. And so a lot of these marketers
- 9:29are trying to figure out systems to create a lot of high creative
- 9:33output at a low cost. And so the market is moving
- 9:36towards that right now. There's AI can say that with
- 9:39absolute certainty. And if someone can figure out
- 9:41how to make it work for these marketers, they will pay you a
- 9:44lot of money because a single ad unit can make them.
- 9:48If a single ad cost them 100 bucks, they could make like
- 9:50$20,000 from that. You know, like it's the easiest
- 9:53return on investment for them ever.
- 9:55So that said, it's not going to be as easy as people think it's
- 9:59going to be. There's a lot of best practices
- 10:00to be figured out. There's like there's this whole
- 10:03concept of prompt engineering when it comes to regular AI
- 10:06applications, right? Like you can't just go to an AI
- 10:08to an LLM and say, like, do this task for me, that's not the way
- 10:12you do it. You know, there's a specific way
- 10:13you structure it. You give it a role, you give it
- 10:15an output format. You have to give it the right
- 10:17context. So that similar concept is going
- 10:19to be applied to image generation.
- 10:21How do you optimize the output of images and how do you create
- 10:25a system and a process to create the highest proportion of viable
- 10:31ads? And as far as I know, that is
- 10:34currently unsold. And so if someone wants to
- 10:39create a lot of value here for themselves and capture that
- 10:42demand, I would be focusing all my time on figuring that out.
- 10:46Otherwise, you're going to have a massive spike and a massive
- 10:48churn. And maybe you'll like do a
- 10:50little bit well in the beginning, but you're not going
- 10:52to create valuable long, long term business.
- 10:56Yeah. What are some best practices for
- 10:58prompt engineering images that you found?
- 11:00Do you use reference images? What like do you ask it to do it
- 11:05in a certain style? Like what are your go TOS?
- 11:08Yeah, so an image is worth 1000 words.
- 11:11Same thing applies here. If you can provide a single
- 11:13reference image, that gives the AI so much more context into
- 11:17what you want. Explain to everyone what a
- 11:19reference image is, or an example of 1.
- 11:22First of all, you can, you can give it as many images as you
- 11:24want. That's something to, that's
- 11:27something to take into account specifically when it comes to
- 11:30ads. That reference image might be an
- 11:32ad that you want to reproduce. So you have a company, a
- 11:36competitor, and you really like one of the ads, but you want
- 11:39this applied to your brand. So the reference image can be 1,
- 11:43the reference ad, which is just a screenshot of the ad that you
- 11:46like, and two can be your product.
- 11:49And then you can get the AI to combine those together and turn
- 11:52it into an ad for your product. So just clearly tell it that the
- 11:56ad is a reference ad to reproduce and the image is your
- 12:00product image that you want within that reference ad.
- 12:03Just be clear about that. That will improve the quality of
- 12:06the output like orders of magnitude more than if you were
- 12:09to try and describe that image to the AI.
- 12:12Gotcha. OK, what's a like an example of
- 12:15doing that? Great example would be you have
- 12:19a lot of people try and a lot of people when they try and make
- 12:23ads, they take a winner and they want to create a new variant of
- 12:28that ad, a diverse variant of that ad.
- 12:30Because once an ad has been inside of a inside of an ad
- 12:35network for long enough, it becomes fatigued.
- 12:36So you take a winning ad and you try and create a variant of
- 12:40that. So a really easy example is take
- 12:45that winning ad, give it to give it to the LLM and ask it to
- 12:49create a new variant, A diverse variant of that ad.
- 12:53Even better, give it to Soro and get it to create 4 images, 4
- 12:58variants. I guess I'm kind of getting
- 13:00ahead of myself here. But like the API is going to let
- 13:02you create like 50 variants of that because now the API is
- 13:07available. But yeah, so you're giving it a
- 13:08reference, you're getting a new variant of it, and you can pick
- 13:11the best one and you can use that to refresh your account.
- 13:15What is the benefit of using Sora for image generation versus
- 13:18just within ChatGPT? So this take might be updated
- 13:21with the API coming out, which is something to consider, but a
- 13:25couple there. I was going to say like most
- 13:27people aren't going to use the API that are listening to this
- 13:29in my opinion, so maybe not. An Apple come out I think that
- 13:32does things better than Sora is the only reason why I say that.
- 13:36So like Sora is limited to creating 4 images at once.
- 13:40With the API you can create more than that, but with ChatGPT you
- 13:43can only create one image. So Sora is immediately better
- 13:48because it creates 4 images in parallel which is which means
- 13:52you have more choice when selecting which one you want to
- 13:55use. So store is great because of
- 13:59that functionality, because the images, it's unlikely you're
- 14:02always going to get the output that you want first try.
- 14:05It's going to take a couple of attempts to do it.
- 14:07So is also great because it has a library that you can reference
- 14:11of like top ranked images and you can click on it and you can
- 14:14see the prompt. And that's really amazing for
- 14:16for like learning and for seeing what's working out there and for
- 14:20getting ideas. And this is going to be this is
- 14:24unconfirmed. This is just a little personal
- 14:26intuition. It could be completely wrong,
- 14:28but I think what happened was Open AI had two versions of the
- 14:33image generation model. They had like the main one which
- 14:36was initially released, and then they had a smaller one.
- 14:39It might have been like an earlier checkpoint or like
- 14:42there's a lot of different ways that these engineers can
- 14:44decrease the size of their models, but that reduces
- 14:48quality. It it reduces compute demand and
- 14:51it lets them, it lets them output the images at a cheaper
- 14:53cost, but it reduces quality. And I think that they switched
- 14:57their ChatGPT model to a smaller one, the one that was used in
- 15:01ChatGPT. And a lot of people noticed that
- 15:04the output quality had decreased, but I think they kept
- 15:07it in. I think they kept the original
- 15:09one in Sora. When I was like testing ChatGPT
- 15:12by Sora with the same prompt, the Sora outputs were a lot
- 15:16better. And so I think Sora does create
- 15:19high quality images, which may be just like, it may be wrong,
- 15:23but it's just a little bit of intuition that I have about it.
- 15:26There's just, I mean, there's just no point to use ChatGPT for
- 15:29image generation. You you can move at least four
- 15:31times faster. It likely, maybe not definitely,
- 15:35but likely has a lower image quality output.
- 15:37So you might as well use Sora. Like if you had to choose gun to
- 15:42your head, you can only generate images on Sora or you can only
- 15:46generate videos on Sora. What do you think is the bigger
- 15:49opportunity? Which would you choose?
- 15:51Images for sure. Why's that?
- 15:53Videos. Videos still haven't had that
- 15:55moment we spoke about before. They're still unusable.
- 15:58They're still cool, but unusable.
- 16:00It's a gimmick. So yeah, exactly that.
- 16:03It's fully a gimmick. So video is cool, but you know
- 16:05what's better? Long form audio via podcast and
- 16:08my newsletter, TKO pod.com. Go there to subscribe for free
- 16:12to my newsletter. It's one e-mail a week, very
- 16:14tactical. And then go to my audio podcast,
- 16:163 episodes a week. Stuff like this.
- 16:18You're going to love it all free, no sleazy sales bitch.
- 16:20TKO pod.com you. Can think of it like this if you
- 16:23use. If you use text, like I would
- 16:26say 90% of outputs are going to be good.
- 16:29If you use images, 60% of outputs are going to be good.
- 16:34If you use video, 5% of outputs are going to be good.
- 16:37So that's until video can get to the that certain threshold where
- 16:42it's useful. I wouldn't really invest a lot
- 16:45of time though. Let the let the guys with like a
- 16:47ton of money training these models, invest their time there
- 16:50and when the opportunity comes, take advantage of it.
- 16:53Do you think there's an opportunity for people to kind
- 16:55of surf the, the Sora library, sort by most liked, most viewed
- 17:00and think of like rapper ideas or business ideas based on what
- 17:03those prompts are? Cause for those who haven't
- 17:05played around with Sora like it will show you what the prompt
- 17:07was exactly to generate that image.
- 17:09Do you think there's anything there?
- 17:11Yeah, there's so many ways people get can get creative.
- 17:14You Sora is a great place to go to just like get the get the
- 17:18brain juices flowing and just thinking about what's possible
- 17:22because people are so creative, they're coming up with the most
- 17:24creative things. I highly recommend anyone
- 17:27listening to this just scroll through Sora and just take a
- 17:29look at their explore page. Anything that was like a
- 17:33pre-existing filter that people use, I would take a look at.
- 17:36There's so many ways to get creative here.
- 17:38It's just unbelievable. I'm really excited to see what
- 17:40people come up with. Yeah, like I was looking at the
- 17:42the explore page and there was a picture of, like, Dumbledore,
- 17:46Hermione and Harry Potter in the gym.
- 17:48And they were all like, jacked. And it was this black and white
- 17:51and it had the most likes. And I'm like, you could build a
- 17:53rapper out of that, out of that one prompt, right?
- 17:56There's a billion people that love Harry Potter.
- 17:59There's just so many ideas. Is the prompt the most important
- 18:02part about all this, or is there something else that that people
- 18:06are missing? So it's task dependent.
- 18:09If it's creating ads, it's the reference image for sure.
- 18:13But think. Think of the reference image as
- 18:16a part of the prompt. It's all the.
- 18:19It's not actually truly separate in the way that the AI ingests
- 18:24it. It can take two different
- 18:25modalities and use them both to create an output and so if you
- 18:31are trying to reproduce something, then the image is the
- 18:35most important thing. If you were trying to create
- 18:37something completely new and creative that doesn't require
- 18:41like reproducing something real, then the text the prompts is the
- 18:45most important thing. OK, what's an example of when
- 18:50the prompt would be the most important thing?
- 18:52No reference image necessary or as relevant.
- 18:56So like for example, if you're trying to create, if you're
- 18:59trying to create that like Harry Potter scene, right?
- 19:01If you're trying to create something truly creative and
- 19:03truly novel or like something like a fantasy using characters
- 19:09or themes that are very, very well known on the Internet.
- 19:12So like for example, Harry Potter is very well known on the
- 19:15Internet. And so the AI already has a very
- 19:18good understanding of what that is, then no reference image is
- 19:21necessary. Like by providing a picture of
- 19:23Harry Potter, you're not really moving the needle.
- 19:25Because you're not going to move the needle.
- 19:26Yeah. Yeah, exactly.
- 19:28And so in that case, it's the most important, like for
- 19:31example, with creating like a theme, like using a theme like
- 19:34the Ghibli theme, like the prompt there is the most, well,
- 19:37actually, no, because you're providing a reference of
- 19:39yourself. But yeah, it's, it's difficult
- 19:42for me to think of A use case where you're not providing a
- 19:45reference image because everything I've done includes
- 19:47reference images because it's all some kind of edit of an
- 19:50existing image that exists. I think there's going to be a
- 19:52lot of winning and like stock imagery here as well.
- 19:54Like you can create any stock image that you could possibly
- 19:57think of. And so that wouldn't require a
- 19:59reference image, that will just be a font.
- 20:01Yeah, it's it's super case dependent.
- 20:04Do you think there are opportunities for people that
- 20:07literally their only skill is just knowing their way around
- 20:09Sora and using reference images and using good prompting?
- 20:12Let's say that's all they know. What would be your advice for
- 20:16someone like that to go start an ad agency or to go do like a
- 20:19permissionless marketing strategy and go find like a like
- 20:23a Casper or AD to C like a Ridge wallet brand out there?
- 20:27Take their ads and those might be bad examples because they're
- 20:30probably on the cutting edge of this, but you know what I mean.
- 20:32Any company and like without permission, make gorgeous ads
- 20:36for them, send them to them in a cold e-mail and like start
- 20:39charging. Like how would you suggest
- 20:42someone approach that? What would they charge?
- 20:43Like how could they reach out? What what are your thoughts on
- 20:46that? If you had to start that
- 20:47tomorrow to pay the bills, what would you do?
- 20:49I would probably partner with existing ad agencies and I would
- 20:52say I'm going to help you to offer a service to your
- 20:55customers. You already have customers, they
- 20:57want this. Let me just just whitelist my
- 21:00service and I will. I have the skill set to create
- 21:04high quality images and I have the system to create high
- 21:07quality images and I'll do it for you.
- 21:09And I think if you DM D any of these agencies on Twitter, they
- 21:12would probably give you a shot because the way that their
- 21:15business is run is it's all it's all like arbitrage on another
- 21:19service. Like it's either labor arbitrage
- 21:22or it's like software arbitrage or it's like another service
- 21:24arbitrage. And so if you can deliver the
- 21:27service to them and they can sell it for more than they're
- 21:29going to say yes. And so I would just find as many
- 21:32as you can. I would offer this and I would
- 21:34figure out. I would dedicate my time to
- 21:36uncovering those best practices and understanding what works and
- 21:39what doesn't. And so you basically, I like to
- 21:42call it piggybacking. You'd piggyback off someone that
- 21:44already has a whole portfolio of not just customers, but trusted
- 21:49customers. They already have their foot in
- 21:51the door. They're already paying them.
- 21:53Even if they're dormant customers, they can reactivate
- 21:55them with this. How do you think you would
- 21:58charge? Like would you just say I want
- 22:00to take a third of whatever you charge?
- 22:02Like you white label this to me? Like here's how you would sell
- 22:05it, here's how you'd promote it, here's some examples, Outsource
- 22:08it to me and we'll split the revenue 5050.
- 22:10Like what do you think that looks like?
- 22:12Difficult to say. The way that ads are usually
- 22:15charged in marketing agencies is like per ad, like there's a
- 22:18certain cost per ad, value per ad.
- 22:21And so I would determine some kind of value per ad, which is
- 22:27reasonable. And then yeah, like I would
- 22:30split the profits probably or I would deliver to that.
- 22:33Like I would understand what they would be able to sell it at
- 22:36to their customers and then provide a reasonable mock up
- 22:39with them. And then it becomes a really
- 22:41nice volume game, right? Like you just, everyone's
- 22:43incentivized to just create more ads and there's almost no cap
- 22:47that the brand will buy from the agency if the ads are ready to
- 22:51go into the account because it's very unlikely that they're going
- 22:54to lose money on that in the long term.
- 22:56And so I'd probably do it by ad or you could figure out like a
- 22:59cost that makes sense as like a base plan, right?
- 23:02And like they just charge you that as like a retainer if
- 23:04they're willing to do that, like regardless of how many ads are
- 23:07accepted or not, I think that would really work as well.
- 23:10But that like is a little bit that reduces your upside but
- 23:13decreases your downside or or do some combination where it's like
- 23:17a base retainer plus number of ads accepted by customers and
- 23:20launched into the account, something like that.
- 23:22Dude, I like that. Here's another idea that just
- 23:25came to me. So a lot of these ad agencies
- 23:27list out their customers, right? They just make it too easy for
- 23:30growth hackers like us. You go scrape a list of all
- 23:33these ad agencies, preferably ones that aren't on the cutting
- 23:36edge, aren't already doing this. You build a wrapper with Replit
- 23:39that just can basically do what Sora does right, But you brand
- 23:43it towards ad agencies only. You're not going towards, you're
- 23:47not trying to get brands or companies.
- 23:48You want to go through the ad agencies and then you go to
- 23:51their portfolio page and you see, all right, John, ad agency
- 23:55has Athletic Greens as a client. OK, I'm going to go remake 6
- 24:00Athletic Greens ads and I'm going to send them to him
- 24:02proactively with a link to my wrapper that I built in Replit
- 24:06that allows them to do this themselves.
- 24:08You get them using it, you get them on the interface loyalty,
- 24:11it's in their workflow, they're sharing it with their account
- 24:14managers, etcetera. And then you charge 200 bucks a
- 24:17month for it. And it's just a tool.
- 24:19It's just a tool that they use for unlimited image generation.
- 24:22Now you'd have to watch the credits, right?
- 24:23Depending on how much they use it.
- 24:25But I like that model over like charging per ad because usually
- 24:31brands want to know like, what is this going to cost me per
- 24:33month? Like just, I just want to know.
- 24:35I don't want to be surprised with a big bill.
- 24:36What's it going to cost me? So really we're just talking
- 24:38about like a wrapper for AD agencies.
- 24:40What do you think about that? I think in the beginning you
- 24:45need to do it for the agencies I think like.
- 24:48Remove the friction. Yeah, I think tools are good,
- 24:51but it's really difficult to get people to use tools like they
- 24:54actually use it themselves, even if there's really obvious value
- 24:57there. And like the dream scenario for
- 25:00an agency is they don't need to think about it and they make
- 25:03money. And so while you can, I would
- 25:06deliver the service for them. Like I would literally send them
- 25:09Google Drive folders of just completed ads.
- 25:11I would ask them what format their client expects ads in.
- 25:14And I wouldn't overthink it. I wouldn't overthink the
- 25:17software. I would just be like, how do you
- 25:18deliver your ads to your client? OK, I'm going to build you ads
- 25:20and I'm going to put in that format for you and I'll send it
- 25:23to you. And all you need to do is pass
- 25:24it off to them. And I think that behavior of an
- 25:27ad of a brand approving ads from an agency and the agency and
- 25:31then charging based on the number of ads they approve, I
- 25:33think that's normal. And they're not going to be hit
- 25:35with a massive bill. You know, like they could be
- 25:37provided ads and they don't need to accept all of them.
- 25:40Like they're very familiar. They're aware of how much
- 25:42they're going to be paying for these, you know, Oh.
- 25:46Man, my mind is like buzzing right now.
- 25:48We I saw you either on Twitter or on another YouTube video
- 25:52talking about micro personalization with these image
- 25:55generation tools. What What are your thoughts on
- 25:57that? I think it's kind of amazing how
- 26:00marketing can change now where the bottleneck and how you
- 26:05managed almost everything was the creative and like the dream
- 26:11of marketing are these hyper personalized, hyper targeted, 1
- 26:15to 1 interactions with your customer.
- 26:18And so like that's obviously impossible.
- 26:20So companies think of like different personas and they try
- 26:24and build towards those personas.
- 26:26And let's just say based on the bottle, based on your creative
- 26:29bottleneck, you could only target 4 personas a month right?
- 26:33Now if you can create so much more creative, then the
- 26:35floodgates open on the ability that you can market yourself.
- 26:40So rather than trying to market middle-aged women in America,
- 26:44now you can target upper middle class women who grew up in a
- 26:50certain post code in southern Florida.
- 26:53You know, like there's like ways that you can really nail down
- 26:57your personas because now you're able to actually create, create
- 27:01a for them that targets them at a scale which and at a cost,
- 27:04which makes sense for you. I think that like that's going
- 27:07to really change marketing. I'm really excited about that
- 27:09too. You know, like I spoke to an
- 27:10events company that runs events and it was really difficult for
- 27:14them to market every single event because how they're going
- 27:16to create an ad for every event and how they're going to create
- 27:19an ad for every person who might go to an event, you know, and
- 27:21that just wasn't possible, but now it is.
- 27:23And so there's a lot of different ways that marketing is
- 27:26going to change and I'm hoping it results in higher quality
- 27:30targeted ads so you have less just like slot on your feed, but
- 27:35it might start the other way first.
- 27:37But it's exciting. It's exciting to where it could
- 27:39go. OK, so you had mentioned like a
- 27:43wrapper that takes two people that makes them, shows them what
- 27:46their baby would look like, right?
- 27:47Yeah, I'm going to share my screen.
- 27:49I just did that. Are you ready?
- 27:51Please. So I went you don't know what my
- 27:54wife looks like or my kid looks like, but this is my prompt.
- 27:58OK so I took a blurry image like this was like it was very
- 28:03blurry. I just screenshot of it.
- 28:04That's me and that's my wife. And I said such a.
- 28:08Cute kid. What would this couple's baby
- 28:10look like? OK, and you see the baby, It's a
- 28:13cute baby. And then I went to my Google
- 28:16Photos and it looks like about like a one year old and it
- 28:20reminds me of my youngest son. And I got this right there.
- 28:27You see it? Yeah, I see that for sure.
- 28:31It's weird. That's crazy.
- 28:34I'm going to build a wrapper that does this like by the end
- 28:38of the week. Please, please do it.
- 28:41Please. It's.
- 28:41So it's so good. I love that.
- 28:44How do you stay focused on this one project you're working on
- 28:47with all this crazy stuff going on?
- 28:49It's. Tough.
- 28:50It is really tough, but I just know where my strengths and
- 28:55weaknesses are. When I get to spread thin, then
- 28:59I kind of don't really win at anything.
- 29:02Yeah. And so I've made the very
- 29:04conscious effort to drop everything else and only focus
- 29:07on one thing. But better or worse, you know
- 29:10that I may end up regretting that, but I'm going to give
- 29:12myself that opportunity for now and see where that takes me.
- 29:16I love it. Well, Jacob, this was amazing.
- 29:19Thank you for your time. Where can everyone find you if
- 29:22they want to reach out? Follow me on Twitter, Jacob
- 29:25under score puzzle. I also post on LinkedIn
- 29:27sometimes if you want to follow me there.
- 29:30But yeah, if anyone builds anything cool, please share it
- 29:34with me. I would love to see it, love to
- 29:36take a look. I really enjoy talking to people
- 29:39who are creative and who are building and who enjoy building
- 29:42because that's my obsession. So if that's you, please reach
- 29:45out and just we can, we can talk through these things.
- 29:48OK. Thanks man.
- 29:50Appreciate it. OK, please share with a friend.
- 29:52Leave me a five star review. I would greatly appreciate it.
- 29:54We'll see you next time on the Kerner office.