S15 E10: Why AI Is the Next Industrial Revolution

0:00Nik:Welcome back to Limited Supply, the podcast where we get deep into the tactical and strategic side of e-commerce, digital marketing, and building consumer brands. I'm your host, Nik Sharma. I've spent the last nine years building, scaling, and investing in brands. And through this show and my weekly newsletter at nik.co slash email, I'm here to share everything I've learned. The wins, the losses, the experiments, the tactics, and the insights, all so you can unlock your next$100 ,000 in revenue. Today's episode is a good one, but before we dive in, let me tell you about our chosen sponsor for this week's episode.

0:34Nik:If you're a Shopify brand, pay attention. Roku Ads Manager is built for growth marketers and makes connected TV way easier to test. Go to advertising.roku.com slash limited supply. All right, welcome back to another episode of Limited Supply. Today's going to be a fun one because I'm just going to rant about AI. And we've talked about AI before. We've had different guests on in the past who've talked about AI. We had Dylan recently from Mention Stack who talked about AI SEO or what people now call AEO and GEO.

1:20AI engine optimization or generative engine optimization. and then we've had other people come on and talk about AI Creative. We've had, you know, members from both the SaaS world and the e-com world come. And to be honest, like up until probably, call it a month ago, I think for the most part, everybody was kind of just on the same page about AI. It's like, all right, it exists. We're going to use it for, you know,

1:52Nik:we're going to try to incorporate it, right? We know that AI is here to stay. It's not something that's just like a trend. Like, you know, Fartcoin, for example, or NFTs, right? Those were here. And then they were not here. And while they were here, everybody was like, oh, we got to figure out how we're going to leverage blockchain technology for our brand and create memberships with NFT tokens and all this other kind of like wasted time that was spent on that stuff. So a lot of people are very cautious about what they're calling the AI wave. But realistically, it's just a new technological revolution, right?

2:23The same way there was the industrial revolution. This is just another one. And so I always think back to those old like Gary V videos where he would be

2:32Nik:yelling at somebody and saying, you know, you didn't believe in the television when that came around and you thought radio was going to be forever. And then, you know, when the new, the phone came out, then you thought the television was going to last forever. Or when the computer came out, you thought it was gonna be the television. When the phone came out, you still thought it was going to be the computer. So like, there's all these examples of where people, There's a revolution very obviously happening in front of them, and people don't take advantage of it. So today's podcast episode is actually inspired by a newsletter I wrote last Sunday.

3:01So if you're not on that, go to nick.co.subscribe, nik.co.subscribe. I kind of wrote this rant after really just spending a whole week last week fully immersed in it.

3:14Nik:I mean, I've spent probably the last month and a half, two months really immersed in it, like super, super deep in it. And I'll share some examples of that in a second. But the last week or so, it's just been like, things have gotten so much better. You know, when last year when I hosted the Ecom AI Summit in New York at Webster Hall, we had like probably 400 brands in the room. And the most engaging session there was vibe coding an app from scratch. And when we vibe coded this app from scratch, people loved it because it was essentially like taking something that they would otherwise pay maybe$25, maybe$50, maybe$400 a month for.

3:54Because a lot of Shopify apps scale their pricing based on order volume. So if you're a big store, you're just naturally penalized for doing nothing other than being successful. I'm Indian. You like to save money. So do I. So we thought, all right, let's teach people how to vibe code apps. So we started doing this. You may have mentioned we had Billy on the podcast as well, once or twice. And he walked us through this process of creating apps, right?

4:22Nik:It was like, you start with the idea, you sort of outline the idea, and you worked with different models. At the time, we were mainly talking about OpenAI, and then we were looking at things like Replit, Cursor, Lovable, as the platform to deliver on what we were trying to make. Whether it was a landing page, whether it was a mini app, whether it was something that integrates directly into Shopify, whether it's a script that Shopify uses based on if-then statements with orders. And that was great. A lot of people, I think, got to hear that episode and then figured out how to create a build-your-own-bundle app or create a subscription upsell in the cart or create a regular upsell in the cart or whatever it may be, right?

5:07A back-in-stock app. And we even did another event with Billy

5:10Nik:where I rented out the Shopify space in New York and we had like 100 people come and we all built a back-in-stock app together and then integrated it in Shopify. So it's not like AI has not been here. But when I tell you AI has been here in a completely different way the last two weeks, like if you're not on board at this point, if you're not getting on board, if you're not starting to use AI, you're going to for sure be left behind. I just gave my mom this speech two weeks ago when I was leaving an event. And I was like, Mom, if you're not on Claude right now, in two months when it becomes even more advanced and even more of a black box, which I'm going to talk about in a second, you're going to be left behind.

5:54You're not going to know how something works. And therefore, when you ask a prompt and you see the end result,

5:59Nik:you're not going to know what took place behind the scenes. Because right now what's happening is as these AI tools are getting stronger, and I'm talking about every two to three weeks, there's a new model that drops. And I'll explain models in a second. When these models drop, they do more and more things without you being able to see what's being done because that's just how they progress. They're just getting smarter and smarter. They're learning from these millions of users, hundreds of millions of users, I think maybe, who are already using AI and using it to build new things or create new things or speak about different things differently.

6:37So when these models drop and you can't see what's going on behind the scenes, it's like the equivalent of running ads in 2017 or 2014. That's when I started running ads. 2014, using Power Editor, the jankiest of the jank, still worked. Maybe better than today's ads manager, actually more uptime.

6:57Nik:But everything was manual. You had to choose your audience. You chose your targeting. You chose your placements. You chose your bidding strategy. You chose, you know, literally every little nook and cranny was chosen by you. And when you talk to a media buyer from 2014, they have a much different approach and a different respect to the ads manager as a whole, right? Most people who are buying media in 2014, they remember that kind of that feeling, like you get that gut feeling as a media buyer because you have so much control and you can have your fingers in a bunch of different pieces of the action versus today, it's like, hey, I want to run this post as an ad or I want to run this creative as an ad.

7:36And it's relatively simple. There's a couple things you got to do. For the most part, it's up. Maybe you're changing a few things, but for the most part, you're kind of relying on Meta's algorithm, whether you use a different algorithm. You know what I mean. I'm not going to even get into it and try to sound smarter than I am. So that same approach is what's happening now in AI.

7:58Nik:and new models are coming out every two weeks. What's a model? Well, there's a few different companies that are kind of like the big players in the space, right? So first of all, we're going to talk about models and then we're going to talk about wrappers. Models are, and we're not talking about models and wrappers like in the club, that's funny. But models and wrappers, it's a good event name actually. So models are put out by companies like OpenAI, Anthropic, Perplexity. What are some of the other ones? There's some companies in China that have made some good models, whether it's for video or for graphics or for text.

8:34There's a bunch of different models, right? And so these companies put them out and then they name them.

8:41Nik:So you might have like Claude Sonnet. You might have Claude Opus. You might have OpenAI, you know, ChatGBT 5.4, 5.6. These are all different models. And while they're all still accessible, the newest ones obviously cost the most to use token wise or credits wise, depending on what the platform you're using quantifies them as. And they get smarter, more efficient, faster. They start to do things without you having to ask. So, you know, a previous version of a model might be like with when Billy was on, right, it might be like, hey, let me use this model to help me define the prompt. Then let me take this model and put it into this different software, which might be a wrapper, which I'll talk about in a second.

9:26Nik:And let me build the landing page there or let me build the module of the Shopify site there. And then let me use something else to actually integrate it into my site. Now, it's more like, hey, I want to build this model or this module in my website that explains how my product works, shows three to four GIFs as a side scrolling slideshow for desktop and mobile. put it in my Shopify theme put it in this Shopify theme which is currently in staging and send me a preview link and boom, it just happens now so that whole reason right there seeing the comparison of going from let me build a prompt to let me use one thing let me take that output, let me then put it somewhere else take that output, put it somewhere else that's kind of the old way of working, right?

10:16Nik:It's basically like you have AI employees to some degree or like AI building things for you. And then you are kind of managing. You're the project manager. Now the project management layer is almost getting taken over by AI. And depending on how much you are constantly feeding in, educating the bot. To be honest, what I do, because this sounds weird, right? You're saying like, oh, I got to teach my bot. I just name them. So I've got four. My first one, I named him Sanjay just because, you know, the homie Sanjay. The second one, I named him Jet. The third one was Geo. And the fourth one was Cole.

10:51Nik:So now I've got names for them. They've all got kind of personalities that I've given them, inspired by different TV show characters. They're all sitting right in front of me. I'm looking at three Mac minis and one MacBook Pro here at my desk. The only cable connected in is a power cable. And these are all running what's called OpenClaw. You've probably seen OpenClaw all across Twitter. OpenClaw is basically an open source, almost think of it, I guess, like an open source operating system that lives on top of your Mac computer. So everybody on Twitter recommended Mac Mini, so Mac Mini is what I got.

11:27I went and bought three from B &H, two of the base model, and then one is just a little bit more souped up. That's my personal Claude bot or OpenClaw bot. I think Claude bot got a message from Claude to change their name to OpenClaw or MaltBot is, I think, the new name. So I set up OpenClaw or MaltBot on these three computers plus this laptop that's sitting here.

11:49Nik:One for me is just a full-time second brain. Now, the way I'm thinking about this second brain is I'm feeding it everything. It has read access to emails, to calendar, to WhatsApp, to Telegram, Zoom, Fireflies call recordings, Slack messages, Now, keep in mind, I said read-only access. So that's the difference between everybody freaking out about giving it full access to everything, which would include the ability to write, send emails, delete emails, write messages out publicly, versus it has its own email address, right? Which kind of state that's like its own system OS email to set up the Mac and whatnot.

12:33Outside of that, it doesn't have any ability to send emails. But it has every ability to read and receive. So the benefit of that is every day, it sends me a brief every three, four hours with basically a full update of,

12:46Nik:hey, by the way, on Slack, you missed this. On email, you missed this. Here, you missed this. These things are still pending. This thing came into your email. It looks urgent. And then the more that I teach it back, like, hey, by the way, this isn't urgent. This stuff needs to go here. When you see this next time, send it there. The more you can do that, the more it learns and the faster it's going to just work for you as a better agent. And the other two went to my offshore team members, Ramiz and Manuel. So one is for Ramiz, one is for Manuel, and they're going to basically 10x their output as a result.

13:21Now, you're probably wondering,

13:23Nik:all right, this all sounds maybe somewhat interesting. I'm not sure. And what kind of stuff can I build with it? Well, there was a few things that I wrote in my newsletter that I wanted to share here. So one is called Zykes. One of my friends that I grew up with, his name is Nick Poogee. He's the most decorated or the ADA lawsuit defending attorney who's defended the most ADA cases in the country. Maybe the world, but definitely the country. He's the youngest partner at Denton's. He's a beast. And I was just thinking, these lawyers, a lot of them get brought in to just defend brands versus proactively protect.

14:07But what if we could build a tool where we could almost use the same exact methods

14:12Nik:and tactics and tools that these scummy ambulance-chasing attorneys use so we don't even have to get to a defensive attorney and get to the settlement piece. So we built Zikes.com, Z-I-K-E-S.com, me and a couple friends. And it works. You put your domain in, we run a ton of tests. And within five minutes, we tell you how much liability you have on your website. Any website might have a million to 10 million in liabilities. And then it tells you exactly how to fix it. And hey, if you don't want to fix it, you'd rather just pay somebody else to do it. You could do that too. But when it tells you how to fix it, what would normally take months to develop?

14:56to be able to build a database of rules,

14:59Nik:constantly be updating that database of rules, then to create an app with a front-end, a back-end, a database, all talking to each other, all secure, all efficient. And then be able to create these front-facing front-end reports and UI and all that stuff. That would have easily taken weeks, if not at least a couple months. We built this in probably three, four days and then continued to refine it. and it's an amazing product. Another example, if you check out DTC101.com, this is just a site I built when I was traveling last week on the plane. I took three, four years of newsletter data, extracted that into about 6 ,000 plus tactical, highly offensive, like playing on the offense, not offending you, but highly offensive style tips and turned it into d2c101.com.

15:54Nik:You could go there, you can browse it by category. If you have a specific question, you can ask. It'll find you the tips that are immediately relevant. So if you're like, advertorials, how do I do those? It'll show you all the tips related to advertorials. Another one I built last night because I was getting requests of, hey, can you help me build an advertorial? Can you help me write an advertorial? Show me how to create a good advertorial. Well, I've written many advertorials. I'm talking hundreds of advertorials myself over the last nine years. Well, why don't I just take all the learnings of the advertorials, distill them down into a huge Markdown file, which Markdown file or.md file is basically the equivalent of a text edit file, but just for AI bots or OpenClaw, Claude, OpenAI.

16:41Nik:They're basically text files with some level of unique formatting inside. So when the AI reads it, it can understand, oh, this is supposed to be a headline. This is supposed to be italicized. This is bolded. This is linked. Plain text, you don't have any of that because it's, again, plain text. So markdown files are great for that. They're super efficient size-wise and file-wise, but you can send them around and they've got some level of formatting. It's almost like the equivalent of adding tone of voice to a plain text article. Shopify brands, this is the one. Roku Ads Manager lets you connect directly to your Shopify store and run streaming TV campaigns that can drive action, not just awareness.

17:19Nik:Roku's positioned this as a big differentiator, shoppable TV through action ads and their Shopify integration. If you want top of funnel on the big screen with a proper path to purchase, go to advertising.roku.com slash limited supply.

17:36Nik:You know, what I did is I basically took all my learnings of advertorials over the last nine years. I probably spent a few hours doing this, consolidating all of them, creating a huge markdown file with all the learnings, with kind of how I think. I even spent 20 to 30 minutes using an app called Whisperflow, W-I-S-P-R-F-L-O-W, where with Whisperflow, you can just talk to a mic. And you can do this on your laptop. You can also do it on your iPhone, which I highly recommend. I pretty much exclusively message with Whisperflow now. But you just talk to it. And the same way that I'm having inflection in my voice, with tone of voice, with texture, is the same way that Whisperflow then documents what you said into a text box.

18:21Nik:So what I'll do is I will speak for 35 minutes of advertorials straight, almost like a podcast. And then I, and literally I can take my podcast that I have done of, you know, I probably have four, five hours worth of podcast recordings just on advertorials, right? And I can put all of that into this mini brain and create my own advertorial product. So right now, if you go to write.nick.co, you'll see I actually created my own advertorial platform where anybody can go and create an advertorial. And it's, well, right now it's free to use. I don't know if I'll put something on there. Probably not for a little bit, but play around with it.

19:05Nik:You can see how intense or relaxed you can be with this AI stuff. You can use it from helping you write a copy all the way to building an app that helps you write advertorials. Another problem I came across yesterday, a friend of mine said, I feel like our positioning on our website for our SaaS product is just like, it's meh. It's not amazing. Can you help me write some better copy? because I feel like we're not nailing how we're talking about it to other people in the B2B world when we're trying to sell to them. And I thought, well, okay, on the consumer side, what I like to do is I just like to go scour the internet and read a ton of reviews.

19:52Nik:I like to read what people are saying on Reddit, what people are saying on Twitter, basically how are people talking about this app? And most likely that's different. Well, in this case, it was definitely different than how they were talking about it on their website. Because on the website, you have product managers and product marketers pushing that, right? But in the real world, you have how customers are responding and what they love and the benefits that they feel. So I basically used Claude to go and find a ton of information on the internet. Plus, I used Grok because Grok is the only AI model that allows you to get into Twitter without paying a separate developer fee or an API fee for that.