#324 - How AI, ChatGPT, & Midjourney Will Dramatically Impact Amazon & E-commerce

#324 - How AI, ChatGPT, & Midjourney Will Dramatically Impact Amazon & E-commerce Speaker 1: What's up everybody? Welcome to episode 324 of the AM-PM Podcast. We've got a really good episode for you this time. My guest is Anthony Lee. We talked a little bit about his background in the Amazon space. Now he worked for several different companies including Helium 10 And then we dive deep into AI and how AI is going to make a profound difference in everything we do as an e-commerce, specifically as an Amazon seller. I think you're going to really, really enjoy this episode and learn a lot. Unknown Speaker: Welcome to the AM-PM Podcast. Welcome to the AM-PM Podcast, where we explore opportunities in e-commerce. We dream big and we discover what's working right now. Plus, this is the podcast where money never sleeps. Working around the clock in the AM and the PM. Are you ready for today's episode? I said, are you ready? Let's do this. Let's do this. Here's your host, Kevin King. Speaker 1: Anthony Lee, welcome to the AM PM podcast. It's great to have you on. Speaker 2: It's a pleasure and an honor. Actually, I've never been on this specific podcast and so I'm really excited that this is my first time. Speaker 1: Yeah, there might be a reason for that, that you haven't been on. It's because at one point you worked for the enemy. You worked for another company that was competing with Helium 10. So that might have been, back when Manny was doing it, that might have been the reason that you didn't come on. Speaker 2: You could probably say that a couple of times over. I think I've worked with too many at this point. Speaker 1: You've been doing this Amazon game since what, about 2014 or so? Speaker 2: Yeah, that's when I started 2014. It was crazy too because it all kind of just like rushed all at the same time. So I started my own brand. And then all of a sudden, at the time, Ryan Moran was my mentor and Ryan was like, I have some friends that need help with Amazon. Do you want to be a consultant? And it was like, literally overnight, everything changed. It's like, okay, well, then I guess I'm going to tell these people what I did. And so I took on clients and then another friend was like, hey, I'm starting a service and I need your help. And that's when I went to work with Zonblast and this all happened like within six months, it was nuts. Speaker 1: For those that weren't around back then, and maybe they just started selling in the last few years, what was Zonblast? It was a major player for a while there and like the go-to place, them and Vyralaunch, but you guys, I think were bigger. What was Zonblast and how did that work? Speaker 2: So Zonblast was, as far as I know, the first product launch service specifically for the purpose of helping Amazon sellers rank. And so, at the humble beginnings, before it ever actually became a SaaS, what we were doing, no lie, was individually emailing our list with our clients' product promotions. It was coupon-based back when you could do 100% coupons before Amazon stopped that. And my job, my first job for the company before I helped, you know, templatize it and scale it out and then we became a SaaS was to write those emails. So yeah, fun times. Speaker 1: How did they build that email list? Was it some other products that a list that they had or did they run Facebook ads to build that list? Speaker 2: So, originally it started with the owners. So, he started Amazon as well and he created a list for his brand and he realized, you know, I could blast out coupons and rank my products and then he was like, oh man, these guys are going to get bored if I don't start offering them other stuff because I only have so many products. So, it started with his list and then we ran Facebook ads and then we ran Pinterest ads and next thing you know, it was just a whole like ads pipeline into this other list of buyers. So, then we had two, you know, the seller list and the buyer list. Speaker 1: So back then, for those of you who don't know, 2015, 2016, and Zonblast was basically the pioneer in this method. You actually could, to launch your product, literally you could launch to the top of the page in hours a lot of times. You would go out, like Zonblast, like he said, initially they would email this list of people and say, hey, we've got this new fidget spinner or whatever it is, and you go, you know, it's normally $19.95 for a pack of three. Here's a coupon, get it for free. And all we ask is that you leave us a review. So people would go and buy this product, use the coupon and get it for free. And then they would go on and actually leave a review. And as long as they wrote the words, something to the effect of I received a free product in exchange for this review or something along those lines. Amazon said, no problem, this is totally okay. And that lasted until, I remember the dates like it was Pearl Harbor Day. It was October 3rd, 2016, when Amazon put the hammer down on that and said, you could still do these blasts, but you cannot do the review portion of it anymore. So that changed the game a little bit. And then Amazon then, took it a step further shortly after that where he said, you can't do these 100% off coupons anymore. And then they changed the algorithm to start looking at what kind of discount people were getting and that was affecting your ability to rank and the whole game changed. And so companies like Zonblast then pivoted into like software. And so Zonblast itself wasn't a competitor to Helium 10 back then when I was joking with Anthony earlier, but it was more of a, when the software came out, that became more of a, There's like a keyword research tool and stuff that became more of like a competitor and you guys did that for a while. You were based in Asia at that time, right? Speaker 2: Yeah, I was in Taiwan. Speaker 1: In Taiwan. And then you did that for a while and then coincidentally, then you came back to the States and worked for Helium 10 for a little while as well. Speaker 2: That's right. That's right. I actually moved to the States to take that job, moved to California and everything. I had an initial like Zoom interview with Manny and Guy and then had one with Boyan and that was the one he was like, hey, I want you here. And I was like, all right, I'm buying plane tickets. Speaker 1: Awesome. So were you ready to come back to the States? Are you ready for a change? What was it like living in Taiwan? Speaker 2: I loved it. I loved it and I miss it every day. I can't say that I was ready to move back to the States, but the timing just kind of worked out because I had some family stuff I needed to attend to that I hadn't been able to attend to. And then when it was looking like my career options were going to shift, then it was just like, OK, well, we might as well. And California is new because I never lived there before. So we gave it a shot. Speaker 1: And California, I don't think was really for you because you ended up going back to your roots, back to Alabama. Speaker 2: That's right. That's right. California didn't work out the way we had hoped. Partially maybe because of California, especially if you've ever had to commute anywhere in California, you know, like it's, everything that they say about it is true. It's a nightmare. But also COVID hit like that year, like in the middle of our lease. And California went on full lockdown. And so there was just like a whole bunch of stuff that was like, oh man, it's probably going to be easier if we just go back with family. Speaker 1: And then you pivoted over to Howard's company, Signalytics? Speaker 2: Yeah, actually, at the time, it was so funny the way it worked out. At the time, you know, my prospects with Helium 10 were starting to look up, but the problem was It was an office job, right? Like at the time, obviously everything's changed. But at this exact moment in history, it was an office job. And I was doing an hour and a half one-way commute. It was just, it was brutal. And they were very kind of traditional in that corporate structure. Like, hey, we really want you in the office. We do too much collaboration with the team and we really need everybody in the office. So that was kind of a deal breaker. It was like, Like I love this company, but it's literally sucking the soul out of me to have to drive to this office every day. And then Howard came up and was like, I want to do something. I want to do something and you're the guy that can do it. And so he kind of came to me with an offer I couldn't refuse. I mean, it was a significantly higher salary and it was remote. And I was like, this is kind of a no-brainer for my family right now. So that was when I left and then Probably, but maybe a month later, everybody at Helium 10 was remote anyway. Speaker 1: Hey, all things happen for a reason. And now you're in Alabama and I just saw recently you posted you're about to move overseas again. Speaker 2: Yeah, we're going to Portugal. Just waiting on this visa process. We've already got an apartment there because that's a requirement for you to get a residency permit. So we're definitely moving to Portugal. I just don't exactly know when. Speaker 1: A lot of people don't know but Portugal really encourages like they're big in the crypto space, big in the remote working space and they actually are one of the countries that's actually encouraging people to come there and do exactly what you're doing and the cost of living there is significantly lower and a lot of people say the quality of life can be much better. What is it that you're looking forward to most about Portugal? Speaker 2: It's beautiful. It's super safe, too. That was the big thing. Doing tons of research, like, where can we go that's incredibly safe? Because I have a family, and they rank No. 6 on the World Peace Index. They score really high for healthcare. But we're moving to an island, like so Portugal, the autonomous region of Madeira. And so it's, you know, excellent weather, beautiful scenery, ocean all the way around. So I mean, what's not, you know, what's not to like? And as an added bonus, Currently, and I've heard this is gonna change eventually, but currently, they don't tax crypto at all. Speaker 1: Yeah, there's a magazine called International Living that I used to subscribe to, because at one point, I was fed up with the US. This is about 2008, 2007, 2008, and I went all over looking for like, where can I live that I'd be willing to live that's gonna be better than US, and I almost moved to Panama. Panama was rated number one for expats from the U.S. because it's close enough, it's a two hour flight to Miami if you need some emergency medical or something like that or need to get something and the quality of life there and the benefits they have coming in. You could live on $2,000 a month there with a full house made, a gardener and a driver and then movies were like half price and grocery stores were discounts and all this crazy stuff. As long as you could show that you had I forget what the number was, you know, $30,000 a year in income coming in from wherever. You got all these perks and benefits and I remember Portugal at that time was like number three or four, but I think it might have actually, I haven't seen the most recent ratings, it might have overtaken Panama from what I hear. Speaker 2: Panama's really, actually, we looked at Panama too. Panama's really good for American expats, primarily because of ease. And the reason is because A, a ton of Americans live there. B, almost everybody there speaks English. And C, they actually take USD. That's like an acceptable currency. Speaker 1: Yeah, they call it the Balboa. Speaker 2: So, that's why like, because, you know, the culture shock can be really hard for people moving from any country to a completely different country. I have to adjust to everything, new currency, new language, blah, blah, blah, blah. But Panama ranks really high and it probably always will because of that. I mean, they have huge gated communities that literally everybody that lives in them are expats. Speaker 1: Like up in David and there's a lot. Now, have you been over to Portugal to check this out with your own eyes or have you just done everything? Okay, so you've actually been on the ground there because before you went to Taiwan, you hadn't been there, right? Speaker 2: No, we actually visited Taiwan once before. So when we decided to move, I was like, all right, let's go on an Asian tour. And we actually went to Hong Kong, Korea, Taiwan. And we walked around to see like the school that we're planning on sending our daughter to and talk to some people. So yeah, I mean, at least visited once before Taiwan. But Portugal, we yes, we visited, we actually Hired an immigration lawyer, talked to a real estate agent, like we did all the stuff that you should probably do before making that kind of commitment. Speaker 1: Now your background before this whole e-commerce is actually in writing. I mean you're known as a writer. Were you a journalist or have you just always been a naturally gifted writer or where does that come from? Speaker 2: Actually, so I was working in restaurants forever and ever and I desperately wanted to get out of restaurants. Speaker 1: Like management or like waitstaff or what type of kitchen? Speaker 2: I've done it. I've done it all. I've worked in the kitchen. I've done waitstaff, bartending. I've managed my family. It's a family thing. My mom worked in restaurants all my life. It was just something that I didn't see the future prospects for. But as a high school dropout, some people don't know that about me, but as a high school dropout, my prospects were kind of slim. And I was like, I have to learn something. So, there was actually a point in my life where I go, In high school, I liked to write poetry. Maybe I can take some of that skill and turn it into copywriting. So I actually made a conscious effort to learn how and teach myself copywriting. So I read all the books like from all the greats, all the Mad Men era guys that wrote books about advertising and marketing and copywriting. And then after consuming that, I just started writing. I wrote like for my own little blog and I was just writing every day all the time as much as I could. And I actually spent an entire year applying for jobs to write, a whole year, literally every weekend. I would work during the week at the restaurant and then every weekend I would sit down and submit five or six or ten, you know, my resume to different applications and was rejected for an entire year until one company was like, the first one that didn't care that I didn't have a degree. And they were like, oh, you have experience with WordPress and you're a writer. Boom, you're in. And so that was what I was doing for a living when this whole Amazon thing found me. Speaker 1: And now that's one of the things you do for Signalytics, right? I mean, I've seen some blogs that are like super well-researched and like super long that you've written. And that's one of the things that you're doing for them, like their SEO and their blogs and the results of different studies where you dive deep into different things. Speaker 2: Yeah, so I do that for Signalytics and that's really been the crux of what I've done my entire career. Most of it's involved around marketing, but it's always been content, right? Content, just because it fits so well with this industry too, right? Like you need somebody out there testing, testing the software or testing the algorithm, you know, sell these units over here, see what happens. And so that's what I would do. And then I would just write about whatever I figured out. Speaker 1: And then I know recently you kind of started dabbling into crypto and then actually went full deep dive into the crypto space. What's your opinion? And now, you know, there's a lot of negativity out there in the press about that. A lot of people say that that ship has sailed and, you know, sorry if you lost money and everybody's going to lose money. And I disagree with a lot of that. I think it's a natural shakeout, just like happens in any business, just like the Internet's happened, you know, and all that rise, Google and Netflix and a whole bunch of others. What's your opinion on the crypto space right now? Speaker 2: Well, for starters, I agree with you wholeheartedly. That's exactly right. This is the natural shakeout. The technology isn't something that is just, hey, this is cool. We could do it or we could not. It really does have use cases. And for that reason, there will be a reason for people to use various types of crypto. You kind of got to dive deep and explain stuff that a lot of people might not understand when you talk about what is the use case for the cryptocurrency itself. But essentially, if you're familiar at all with Ethereum, you know that you have to pay a gas fee to make it. So there's a reason why the cryptocurrency would be needed. And there's tons of reasons why you would want to use blockchain technology. And I know that you've dived into this and you talk about the utility and function of NFTs and that is precisely one of the many avenues for why blockchain is so important. All these functionalities, they come from the technology that's built on blockchain and as long as the blockchain exists, There's a really good chance that the cryptocurrency is going to exist and be necessary for the blockchain to work so that the NFTs and all of the use cases are out there serving people. So I think it has a future, it's just right now everybody pretty much looks at it as a speculative device for investment. Speaker 1: The problem with the industry, there's been a lot of fraud, a lot of people trying to get rich quick, a lot of unsound businesses set up in it, but it'll shake out and some will lose money. I have a buddy that lost $15,000 in the FTX. It's just down the drain. That happens with early adopters. It happens with early things. Not everything makes it, but at the end of the day, it's here for good, especially Bitcoin. I know some of the other oddball cryptos and stuff, a lot of those aren't going to make it. Just like when the internet came out, Pets.com, Didn't make it and everybody was saying nobody's gonna buy dog food over the internet and there's a furniture calm or something Nobody's gonna buy, you know, any kind of thing over the internet now look at Wayfair any kind of furniture over there now? No, look at Wayfair. It's it's gonna it's just in a transition period but like you said that the key is the Ethernet I mean is the blockchain that's underneath it and And that's where NFTs come in. That's not really crypto. They're two different things. I mean, it employs some of the same technology, but they're two different things. And people think of NFTs as just a bunch of JPEGs. Why the heck am I going to pay a bunch of money for stupid JPEG? I could just screenshot. It goes way deeper than that. Yes, some of these projects that have come out are done by 16-year-old kids that know how to program. Don't know how to run a business. They don't know how to do the back end. They don't know how to deliver on what they promise. They take the money and run and it just becomes a shit show. And there's a ton of that going on, but that will get weeded out. It's always going to be a little bit there. Always going to be the dark alleys, but that'll get weeded out. But you look at a project like I don't know if you're involved in Manny and Gee's project, Bulls and Apes. I don't know if you're involved in that NFT project or not, but I am. I bought one of the bulls that I bought back in May, May 31st, when they did the initial mint, which they sold $4 million worth in a matter of six days with a money back guarantee, but they're real business people behind it with a real utility. One of those, at the time, Ethereum was a little bit higher than it is now. So in converting to US dollars, I paid roughly $370 or $400 or so. I can't remember the exact amount, but if you convert it from Ethereum, it was 0.17 Ethereum. And I think back then it would have been equal to about 400 bucks. For a bull, I just recently sold that for the equivalent of $6,500. Just last week. And because the project is solid, because of what they're doing, they're about to do some additional stuff with Apes. And I think this project is going to go, it's one of the few in the space, because it's on sound principles and by people that are running it as a business. Not as a get-rich-quick thing, you know, from the bedroom, it's solid. And I'm doing, like you said, an entire e-commerce business with an NFT angle to it. I've talked about that in other episodes, so I won't go into it here, but I believe strongly that this is, if we can make this work, we may fail, but I think if we can make this work, it's gonna be game-changing. But speaking of game-changing in this space, Something that you're really serious about right now and I'm just starting to dabble in it and that's one of the reasons I wanted to get you on the podcast. is AI, the power of AI. We hear AI a lot, people use it as a marketing term and you'll see some PPC software tools say, hey, let us manage your PPC and we'll use our AI. It's bullshit, there's no AI, it's just an algorithm, it's not true AI. But there's been some stuff recently and I've mentioned it in some of the stuff like in Helium 10 Elite and some of my talks, but I know you've, Dove in way deeper than me into this, but with Dally and Midjourney and ChatGPT and stuff. So I would like to, if you don't mind, talk about that for a little bit. What's your take on where this AI stuff is going and how it's going to impact e-commerce? Speaker 2: Yeah. Speaker 1: Especially as a writer. Speaker 2: Yes. So, quick side note, I just realized that we've talked and in us talking, we have pointed out my varied interests and it went from Amazon's algorithm to NFTs and now to AI. And I'm like, I don't know, it kind of makes me feel like a polymath. This is cool. Anyway, okay. Yeah, 100% artificial intelligence. is amazing. You're right. There's a lot of people that use that term. And the thing is, is I feel like it's a thin line, right? So people use machine learning algorithms, and immediately they're like, well, this is an artificial intelligence. And yeah, you're not exactly wrong. This is a form of artificial intelligence. But I think that the line that most people are drawing Which has become abundantly clear ever since OpenAI and ChatGPT is, does it use natural language processing? Because at that point, it acts like what we interpret as an actual AI, and most of them don't. really advanced sets of machine learning algorithms. But we're moving into an era where the actual like what we perceive as AI because they utilize natural language processors, which is amazing. Why is that so important? Because it allows the AI to actually communicate with humans in the way that they communicate, right? We don't have to know how to code. We don't have to write the algorithm. We don't have to set up the rules for the algorithm to do this or that. Like literally, because the AI knows how to communicate with us, We can just tell it what we want and it can communicate back with us with either vast knowledge. I mean, you can assume that up until 2021, programs like ChatGPT had access to the World Wide Web. So vast amounts of knowledge, but also the ability to formulate it in such a way that it's easy for us to understand the message that they're giving across. They can make some unique connections. I mean, it's still growing and obviously humans are going to be better at some things. Anyway, so nonetheless, the technology is incredible. It's definitely not going anywhere. 100% it is here, not only to stay, but it's one of those things that I believe is going to become a staple, right? People are, it's the next step up, right? Everybody got involved with the web. There's a lot of people who thought that like web three would be the next step up. The problem there is the barrier of entry is too high. Speaker 1: It is. Speaker 2: Web3 is not easy for people to grasp. Speaker 1: It's not easy to get even crypto or NFTs. There's a process. You've got to be technically savvy. Your grandmother, your mother can't do this. It's, yeah, there's a big barrier there that has to be broken down. Speaker 2: Absolutely, and I do believe that one day it'll happen, but if we're talking about the next step up, I think AI is it. And the reason why is because if grandma can't use it now, in a couple months when it gets better, she will. It's constantly becoming, that's one of the coolest advances. It's not just that it can perform your actions better, it's that it can understand The instructions you give it better and interfaces are being built now to make it easier and easier to communicate. So instead of going into open AI and having to know to use their playground and then set up a prompt example constraint, Now we have ChatGPT and soon enough we're going to have an app. Guaranteed, soon enough it'll be an app and you can just literally, like Siri, go to your phone and be like, do this for me and it'll have access to all of your phone's apps and it can do stuff like that for you. You'll be able to explain in plain English automations that you want The AI to control over your phone and that's somebody that has zero technical capabilities. That's something that they can pull off and that's the reason why AI is going to stick around and it's going to flourish. Speaker 1: I mean it's amazing what you can do like the big one that just came in the news about a month ago or so it's called ChatGPT and it's been around for a while but they it's actually at a website called openchat.ai is that right? Speaker 2: OpenAI.com actually I think yeah and then you get an account and then that gives you access to DALI and to chat.openai.com. Speaker 1: Yeah, and Dolly's for images, and there's Midjourney, and there's several others that are pretty impressive on what you can do on images. I mean, on the image side, and there's ones coming for video as well, where you can actually, they'll actually create videos. You just type in text, you type in a description, I want a video of a cheetah. Running across the road with a Lamborghini chasing it, but the cheetah is winning. And I want this to be in an African safari type of situation. And it'll create a video of that exactly for you that looks totally real. Or it could be a cartoon. You could say, I want to be an animation. It'll do that. You can do all kinds of cool stuff now with images. And it takes a little bit of playing with. I was talking to Guillermo. We went to dinner in December. And he's been playing a lot with Dally and Midjourney from the image side and he said, you know, at first it was just spitting back some some images that weren't so great. But then once you really figure out how to talk to it and how to actually structure and put the commas, you know, separating what you want. You can get some amazing stuff. And for E-com, I mean, you could create images with this, you know, backgrounds or settings, or maybe you're hiring a photographer to actually shoot stuff. You can use this for an FPO, for position only. You know, in the old days, you'd have to draw something and say, I want to kind of look like this, or you have to find a tear from a magazine and say, I want the shoot to look like this. But you can almost create it now. With that and you could even do product testing, you know, it may not be the final images you use but you can create 10 different images using the AI by typing in words of what you want. It'll come back with 10 different images and then you can run that on a on a PICFU or something say which one do you like and then you can take the one they like then go take that maybe it's not good enough quality off of Midjourney or Dolly. and actually shoot it or create it. But with the tech stuff right now, like you said, they just opened it up and there's been more than a million people actually sign up for a free account. The reason they opened it up, I was told, is that they need a human use case. They need to know, you know, they've been playing with it in the lab for a while and they need to see how they can fine tune this with all these different humans interacting. It costs money. I mean, I saw somewhere like every query costs two or three cents to the company. And it can't go out to the open web, so it doesn't index the entire internet. I mean, you can put URLs in there and it can recognize it. But it can't really absorb all the information of the internet yet. It's kind of a locked encyclopedia in a way, but it's super powerful. I can actually take it on Amazon. This is great for people who aren't native English speakers, or even if you are, you know, sometimes I just heard someone on another podcast say it takes their employee about two and a half hours to create a listing, to do the bullet points and description, make sure they got all the keywords in there. You know, this is after all the keyword research and everything's done. And with these tools you can do it in about 15 minutes and it may not be perfect when it comes out of there but you can fine-tune it or you can get different variations. You can actually go into this tool right now and for free and you can say create me an Amazon, this is what you type in, you're literally typing this in a little box, create an Amazon listing for a vacuum cleaner. Include the keywords long handle, quiet, strong suction, you know so on in a large capacity bag. Do it in a you know comma in a British tone of speaking, comma with a little bit of humor. And it will write your bullet, five bullet points for you and a product description and a title. And sometimes you get stuff that's, you know, you're like amazed. You're like, damn, a machine just wrote this based on that? You're blown away. And then sometimes you read it and sometimes it'll get it a little bit off. Like I had, I did a test, I was playing with it the other day for a slow dog, a slow feed dog bowl. I said, create an Amazon listing for a slow feed dog bowl. Use these keywords. I want it in this tone. Make sure it's got a call to action, a CTA. And it wrote something really, really cool. But one of the bullet points actually said, this dog bowl is so good, we may use it ourselves. Eat from it ourselves. So it's not perfect. Speaker 2: You told it to be funny. Speaker 1: I did tell it to be funny, but it's amazing what it can do and it can analyze, you know, you can go to like Helium 10 x-ray and analyze the reviews and you can download, let's say all the four and five star reviews, just, you know, download them once for the last six months or something. Copy and paste that into or actually write a command in OpenAI in English that just says, analyze these reviews and write me an Amazon listing emphasizing the five main points discussed below and then paste all those reviews in. It will go through there and it will find the general sentiment of all those reviews and actually write you a listing that will appeal to that It's amazing what it can do. What's your experience with playing around with it? Speaker 2: Every day I like the tools more. It's wonderful. So my experience has been very similar. But just like you described what Guy was doing with the images, like you do the same thing with with ChatGPT-3 and OpenAI. You play around with how you talk to it. It also In OpenAI specifically, you can play around with the settings, specifically the temperature, which kind of dictates how similar or kind of off the rail it goes based on the examples you give it. Speaker 1: What do you mean temperature? Speaker 2: There's a setting called temperature. Speaker 1: Oh, I didn't know that. Speaker 2: And this determines how similar to or how dissimilar creative and off the rail the output is to the example input you put in, right? Because you can do it Zero shot is just where you tell it to do something and then temperature really doesn't matter. But if you do like one or a few shot where you give it examples, then you can set it and it's either going to like mimic the example or if you put it all the way to one, it's a scale of zero to one. Or if you put it all the way to one, then it's like, all right, well, I got the loose structure, but I'm going to do this the way that I think it should be done. And that's just, that's an open AI. So it's just playing around with this stuff. And it's what I call prompt engineering. So essentially what you're doing is you're figuring out how to communicate with the AI. And it's people who are at least decent at writing. I'm not saying copywriter, I'm not saying anybody that makes content, but if you're like decent at structuring sentences, are uniquely qualified to get better at prompt engineering, which is really cool. So, I've kind of, I've been playing around with all the tools and I kind of like, I mean, ChatGPT 3 and OpenAI specifically, and I kind of figured out like the evolution of how people can use it, which has been really, really cool. So, for the vast majority of everybody out there, if you go into ChatGPT 3, don't just tell it, hey, create me a listing. There's actually papers done on this and you can look them up and they will make you a better prompt engineer if you read them. Speaker 1: But- How would I find those papers if I'm listening? Speaker 2: Just type into Google like Googling like prompt engineering, prompt design and then you'll see a bunch of articles written and then sometimes their scholarly actual research paper is written on the advancements in prompt engineering and prompt design. And those are the words that you want to use. Prompt engineering and prompt design is where people are either creating prompts or they are fine-tuning the AI so that it can take the prompts and do what they want with them. But if you go into ChatGPT-3, the best way to use it is to give it small chunks and then iterate. So currently, the OpenAI machine, which is what ChatGPT-3 is connected to, has a memory that spans back to about 8,000 words. So you've got It's almost like a novella worth of conversation you can have before it starts forgetting what you guys were talking about. So in that, what you want to do is you want to break it down. So instead of saying, write me an Amazon list, you go, write me an Amazon title, start with the title, include this keyword, include this benefit. Tell it the tone, like the more specific you can get about like what angle you want it to use. And then it spits something out. And then what you want to do is either use that or you want to tell it, hey, that's good, but, and then you start iterating changes, right? But how about a different tone? Or how about, make it not so long. So you do that until you have the perfect title. And then you take that title, put it in quotation marks and you say, write me five bullet points, X number of characters, or, you know, over X number of characters. If you tell it to do under characters, it tries to make them as short as possible. And sometimes you want that, but if you don't want that, you tell it to do at least this many characters for the product. And then you put that title it gave you, and then it's going to spit those out and you can iterate changes to those and do the same thing for the description. And in that iterative process, That basically forces it to give you its best work and you don't have to regenerate, you don't have to start over. Like I said, you have up to 8,000 words to do this. So, it's all in the same conversation and it's constantly going back to what it's already talked about. And so, it understands what you're asking it when you say, hey, can you go back to that? Hey, bring that title back again. Can we use this keyword and that keyword out of it into the bullets? Hey, in the product description, can you tell a story about somebody that might have used this? Let's change that story a little bit. Like, you know, you're going back and forth and back and forth. And that is the collaborative nature of ChatGPT 3. And that is how you get the most out of it as a totally normal user. Speaker 1: One of the things that AI has, there's some limitations though, it doesn't have sentience, really. I mean, it doesn't, for example, like you and I both know that birds can fly, but penguins, they're a bird, but they generally can't. So, if you type in something around that, there's some common sense rules that AI still struggles with. Like, you know, it doesn't know, does it know that newborn baby birds can't fly? Or birds covered in oil can't fly? Or birds who are injured cannot fly? Or a bird in a cage? You know, there's exceptions and the AI is, they're still fine-tuning some of that on it. So, it's not perfect. Speaker 2: It's not, but that's the reason why the iterative process works so well. Because you can tell it, right? If it comes back to you and it gives you a bullet that says, this is so good, we'll eat out of it. You can tell it, hey, humans don't eat out of dog bowls. Can you rewrite that bullet? And it'll go, oh, yeah, of course. And then it literally logs that bit of information, at least within the 8,000 word limit that it's given you. And it says, all right, well, I'm not going to say anything about humans using this bowl anymore, because I was just informed that they don't do that. Speaker 1: Yeah, I didn't know it went back on conversation and used all that to fine tune. I had no idea. Speaker 2: Up to 8,000 within one session, right? So you know how you can do new chat, but now you have a log of all your old chats? Speaker 1: Yeah. Speaker 2: Up to 8,000 words. It can access all of that information. Now, once you hit 8,000, it starts getting dumb on you. Speaker 1: So if I type in humans don't eat out of dog bowls in that session, then reprocess it. It'll eliminate that because it saw it up in those 8,000 characters above. Speaker 2: It should, you know, It's still glitchy. So every now and again, like I have to, um, I've recently run into some issues, some limitations and what I'm like, basically I'm trying to get it to create JSON objects for me. Uh, but I wanted to do it in a code block and I have like instructions that are kind of long and, um, sometimes it forgets that I said, Hey, do it in the code block. So I have to remind it. Speaker 1: Cause it'll write code. Like you're saying that for people to understand, JSON is code. It's, it's a script. Basically that will allow you to modify, it's used a lot with like spreadsheets and other things to actually modify what was originally written into the code. So you can actually type into ChatGPT in basic English what you want it to do, not even knowing how to code at all and it will come back with the code. Speaker 2: It'll come back with the code. Speaker 1: You can tell it. But you can get it to your programmer or just you could copy and paste into your website or whatever and it will work. Speaker 2: Yep. Here's the cool part, right? If you have just the smallest amount of knowledge about how your website works, right? Like you understand the difference between HTML and CSS, you can go in there and make the changes that you don't know how to make. Literally tell it, hey, I have this. I want to do this. Can you give me the HTML markup language for it? Or can you give me the CSS styling to shift everything over to the center or whatever? It's really cool. The other thing that's really cool about ChatGPT is that it is also fully aware of what OpenAI is. So what I use it for, because sometimes it can get frustrating if you're trying to get it to do really complex things and you're like trying to figure out what the right prompts are and it's not understanding. So what I'll do is I'll go to ChatGPT and I'll say, hey, this is what I want to do and I'm trying to create a prompt for OpenAI. Can you give me a prompt? And it will give you the prompt that you can then use to program the same AI that's fueling it. It's crazy. Speaker 1: Something interesting, I think it was Leo Scovio that said this, that he thinks that the Amazon and some of these other Guys are going to actually start using AI in the reverse way. We're using it to create stuff, perhaps, but they're going to use it to analyze your listing. And they're going to say, you know, how sometimes people's keyword stuff, and they'll put something that's really not totally relevant to the rest of the listing, just trying to rank for that. And they're saying the AI is going to be able to identify that by looking at the bullet points and description and coming up with a contextual, this is what this is about. But this keyword that you have here obviously doesn't really fit into that. Therefore we're going to penalize you and your listing because of that. Speaker 2: Yeah, I have no doubt that that's, I mean, I almost feel like they could do that before, but just to give you an example of how, like, you might do that, right? So on the other end, we have OpenAI, which is a little bit different from ChatGPT because you don't have the iterative process, right? It's just instructions, but it takes a lot more instruction from you. Well, here's the thing, you still are limited by the amount of tokens that you can fit into the examples that you give it. Well, You, as a totally normal person that doesn't even know how to code and doesn't have to know how to code, can actually take a huge-ass spreadsheet full of a JSON list, right? A JSON list, by the way, that ChatGPT 3 can write for you. And you can push that through an API on Postman, which is a free service you can get on the internet, and you can fine-tune the AI. So, what you would do in this example, right? This is what potentially Amazon might do, is they might take 100,000 listings in a specific category. These are listings that we approve of. These are listings that have the right character amount, that have blah, blah, blah, the density of the keywords is this and that. They take those 100,000, they turn it into 100,000 lines of JSON and they feed it into the API. And they have a trained unit now in the model, DaVinci text, DaVinci kitchen category specific Approved listing and then they use that model and then they run everybody, you know, all these new listings through and the AI says this fits. This fits within the constraints you gave us, this doesn't fit within the constraints you gave us. And it's because they fine-tuned it with existing listings that they approve of. And you and I can do that too. Speaker 1: How do you see this potentially changing e-commerce and what we do as sellers? Or is it just a bunch of hype and a bunch of, yeah, yeah, that sounds cool and fun. Go play around, enjoy it. But I'm going to stay here and build my listings the right way. Speaker 2: So, my predictions on how this is going to impact e-commerce and specifically Amazon. I think for the people who use it right, it's going to be good for them. Specifically, I think the biggest advantage is for people who struggle to get the nuance of language, right? Because there's a lot in the nuance, right? The little cultural quirks in wording based on that language. Obviously, you can do something like go to a service and pay them to get a native speaker to translate your listing, but that could be expensive. So, what this does is it opens up a much more economic way for people to get their listing right. Even the people who aren't translating it, they can definitely use this to reword and to kind of... If you're not a writer, do your keyword research, write your listing, but then go to ChatGPT 3 and say, hey, can you make this... Can you insert some emotion? Can you evoke some emotion with this language or can you make it more persuasive? Speaker 1: That's literally what you type in the ChatGPT, exactly what he just said. He's not using that as a paraphrase. You would type, make this more emotional and then the next line would be paste your listing and it'll spit it out within seconds, a more emotional version. Speaker 2: So, yeah, it's going to open up and hopefully make the people who use it right, make their listings and their E-commerce presence a little bit more appealing. What we're also going to see though are people who are trying to shortcut it and they're going to probably churn out a bunch of crap and then try to put that on Amazon. And I think at some point Amazon's going to get pissed about that. And then we're going to probably see some crackdown, but hopefully we have some years before that happens. Speaker 1: I know like Brandon Young was speaking the other day where he's got a team in China that writes a lot of their listings because he's spitting out products left and right. They're not native English speakers and he's got a person that's in the U.S. that's a native English speaker that has to take every listing and go in and clean them up. So it's not broken English or whatever. And that person could be eliminated almost. Speaker 2: Maybe, or if I were that person, I'd be using AI to make my job a hundred times easier. Speaker 1: Some people say this technology is a Google killer. Like ChatGP3, that's one of the little catchphrases that a lot of the media was using. This is a Google killer. But Google has their own version of this. They just haven't made it public. They've been working on it for a while as well. It's just not public. What are your thoughts around that? Speaker 2: Yeah, I think it'd be hard to kill Google because they haven't been in the dark on this. They're building stuff. I think that today, if somebody would just connect ChatGPT 3 to the internet, everybody would use it instead of just regular Google for sure. But it's too expensive for them to do that. It's not ready for that right now. And then by the time they are ready to do that, Google will probably have its own AI assist, I have no doubt. Because they can see right now, they can see that everybody wants it. And the reason that AI assist in search is so cool and so desirable is because anybody that's used Google can see the limitations, right? All it does, it just gives you what it thinks the Appropriate website might be, but an AI assist would also try to interpret what you're trying to figure out, like your actual intention. Google can't figure out your intention. It can only figure out what it thinks is relevant. But in AI, it's just to figure out what your intention is. Like, what are you trying to do? I'm trying to build a chicken coop. Okay, well, here's the list of websites that are going to help and they're not all just like randomly saying, you know, indexing for the keywords how to build a chicken coop. And that will happen and Google has to make that happen now because if they don't, OpenAI will do it for them. Speaker 1: I think it's going to also be, at some point, be part of an Amazon search. So instead of typing, if I'm going to the beach with my family, instead of typing, you know, and I'm looking for, man, what do I need? I'm going to the beach. I don't have everything. I think I need an umbrella. I need a blanket. And I type in beach blanket and look at the results. And then I go type in Beach umbrella, look at the results. Now go type in cooler, look at the results. I think you're going to be able to go into the Amazon search bar and type beach trip, three kids, suntan easy, picnic. And it's going to spit back, it's going to go analyze all the listings and it's going to find the most relevant. Based on whatever factors that is, probably a combination of reviews and comments and questions and the way the listing's written. And it's going to spit back a page of everything, like almost a perfect match of everything that you need for your beach trip. You know, it's going to have an umbrella there. It's going to have suntan oil. It's going to, whatever it is, it's going to have everything on one page tied exactly to what you do. What do you think? You think that may be coming? Speaker 2: I absolutely could see that. Here's the interesting thing though, Amazon already had the data to do that before. The reason they didn't was because it's been sufficient to just give you the people also bought section, you know, taking advantage of that cognitive bias, the bad wagon effect and all that stuff, right? But what's going to happen as mass adoption of AI occurs is we're actually going to be trained over the next decade or so to utilize search differently. And when AI assist becomes a normal part of our search utilization, then Amazon's going to have no choice but to say, well, this makes sense now. See, two years ago, it wouldn't have made sense because nobody searches like that. They would have been like, You want me to what? And that would have been confusing. I believe they haven't done it, even though they absolutely could with the data that they have. But now, A, the tools make it easier, and B, they have the data forever. Once people get used to that type of searching, yeah, I could totally see that. It would make sense. It would get people to probably buy more, and I think Amazon's paying attention to stuff like that, for sure. Speaker 1: Do you think there's any dangers? People always, when they hear about AI, they're like, ooh, these robots, this AI is going to take over our life and screw everything up. Yeah, it's great, but it's also, what about the dark side of this? What are your thoughts on that? Speaker 2: Well, I think that sci-fi movies have made people a little bit more paranoid than they should be. The AI that we're talking about right now, like AI is like this overarching term. The AI that we're talking about right now, these are tools. Just like Microsoft Word is a tool, this is a far more advanced document creation device that does way more than just spellcheck. But it's still just a tool and exactly just for that. Will people be able to do bad things with it? Of course they will, particularly deepfake. I think deepfake is probably the most easy to manipulate kind of situation right now. Speaker 1: And deepfake is, for those that don't understand, deepfake is basically where you can imitate a person's voice and make it sound like this is what they said or things like that or put a person in a picture that literally wasn't there and looks real, that kind of thing, right? Speaker 2: Yeah, video too. There's a couple of them that got really popular on like Instagram, this one guy. So he already looks a little bit like Tom Cruise but he used a deep fake Tom Cruise mask and like the dude looks like you're looking at a video of Tom Cruise. And that's the kind of thing that you can imagine can be used for nefarious effects but every technology is going to be exploited in some way. We're not at a place where AI is going to Like, try to take over the world. This isn't a Terminator event. This is just a really, really cool piece of tech that, you know, in the right hands, which should be hopefully most of us, can just help make our lives easier. Like, that's precisely what it's there for and that's what it's doing. Speaker 1: I mean, even like in this podcast, I mean, if the podcast editor wanted to take the transcript because it automatically creates a transcript and write a description for the podcast, Yeah, they've got to go through and like, okay, what was this podcast about? Here's the key points. Let me write something. You could take the transcription of this podcast, paste it into ChatGPT and say, write me a 225 word description or whatever you want to be of the following. Summary of the following, and it will spit out an almost perfect summary of the entire podcast. Those are the kinds of things that just save an immense amount of time. That's just one practical use that we'll probably, maybe we'll try it on this episode and see exactly what happens. Speaker 2: Absolutely. Yeah, summary is a really good one. Here's the thing about the AI, this technology. The output is as good as the input, right? So everybody that's trying to use it to replace themselves is probably going to have a tough time because you still have to have good inputs. But if you have good input, then yeah, the output's amazing. Like you have a great podcast transcript, it's gonna be able to summarize that very well for you because the input was good. And that is, you're right, one use case, there are several others, outlines, ideas, brainstorm, treat it like it's the other person in the room. That's what it gives you, right? If you're in isolation, because think about it, right? Why do we go to masterminds? You put yourself in a room with 5 or 10 of the smartest other people in your industry and then you guys go to town and come up with all kinds of actionable ideas. In the absence of that, AI is the smartest friend you're going to have and that's what he's there for. Speaker 1: I guess just to wrap this up, we could keep talking about this I think for hours, we can geek out here for a while. But as an Amazon seller, what are like the top three things I need to be paying attention to or maybe start dabbling with or what would you advise someone that's like listening to this and like, man, I never even heard this stuff or I heard a bit about it, but this sounds interesting, I need to, I don't want to miss out and be behind the curve. What should they do and how should they start to implement this or play with this? Speaker 2: Well, definitely, definitely this does not replace the need for authenticity. Be authentic. Tell a good story from the perspective of your brand. Build an actual brand. But outside of that, how you want to use the tool to make your life easier. It's really good for tweaking copy. It's really good for assessing your copy and giving you pointers on how to target keywords better or make it more SEO optimized. So, assessment for tweaking and then it's really good for ideas, right? Good inputs as far as the ideas that you're thinking. Again, it's the smartest friend you'll have. It'll come back at you. So for example, another way I use it mainly for like listing optimization stuff, but another way you might use it is take your catalog or take a catalog that you really like and you put it in there and you say, you know, this is the type of stuff that my customer base likes to use. What do you like? Give me an idea of a product that I could offer them next and then just have that conversation with the AI because it's got a database of all-worldly knowledge up to 2021. It could surprise you with some of its creativity. Speaker 1: This has been fun, Anthony. I just looked up at the time like, damn, we've been talking that long? Yeah. If people want to learn more about this or follow you or reach out to you, how would they do that? Speaker 2: Yeah, you can actually hit me up directly. My email is anthony at signalytics.ai. Check out the Signalytics blog. Right now we're in the middle of a website redo, but once the website redo is live, I will be publishing probably regularly on this topic. Speaker 1: Awesome. I look forward to reading that and having another chat with you maybe six months or a year from now when this technology is even more advanced than what it is right now. Speaker 2: Sure thing. Absolutely. It was a pleasure. Speaker 1: Thanks, Anthony. Speaker 2: Thank you. Speaker 1: So after listening to this episode, your first order of business needs to be go to openai.com and sign up for a free account and start playing with this. You're gonna be totally impressed and surprised at how this works. Go to openai.com and start playing with this. This is gonna have a major impact on how we operate in the future as e-commerce sellers. So it's something to go ahead and get your feet wet in now. and be on the cutting edge of this new technology. I hope you really enjoyed this episode. We'll get Anthony back maybe later this year as this progresses and as more people are using this and we'll dive deeper into some specific use cases and how it's made a big difference for them. Hope to see you again next week and I have another great guest that's doing almost eight figures on Amazon. He does a little bit of a private label and he also does some resale and he's got some really cool software tools and scripts that he's developed to really help him fine-tune his PPC and a lot of other cool stuff. So I hope you tune in for that episode as well. And don't forget if you missed last week's episode with Bradley Sutton. That's a great episode. You'll learn some insights on Bradley's background and Helium 10. It's a really good episode as well. So make sure you don't miss that one either. That's episode 323. So until next time, I hope you have a great week and don't forget today's little nugget. I'm gonna end with that. It is the secret of living is giving. Simple as that. The secret of living is giving. We'll see you next week.