Exploring The Possibilities of Generative AI | Max Sinclair

Max Sinclair: Generative AI can create code as well.

So it's not just a case that this is gonna say, here's your, this is what you should do. It's gonna be, and I, this is something I firmly believe will happen in the next two, three years. You already have generative AI code in, Microsoft and Amazon, like Amazon Whisper, if you're a coder, but, um, you're gonna have, literally a free, free text like, build me a website.

This is my product, this, as you said, and it will just build the website because it will be smart enough to do the code as well as understand all the, permutations on back you know, as you said.

Matt Edmundson: Welcome to the e-Commerce podcast with me, your host, Matt Edmundson. The E-Commerce podcast is all about helping you deliver e-commerce. Wow. And to help us do just that today I am chatting with the very amazingly talented Max Sinclair from e-content, Ecomtent, uh, about exploring the possibilities of Generative AI.

But before Max and I jump into that, let me suggest a few other e-commerce podcast episodes that I think you'll enjoy listening to. How to Bring the Magic of Disney to Your Customer Service with Vance Morris. What a legend, uh, Vance was doing that podcast episode. Definitely check it out. And in fact, if you're part of e-commerce cohort, uh, Vance has also got a training in there, which you can look at in one of the sprints, which is brilliant, let me tell you,

uh, also check out how to develop a creative strategy for your brands advertising with Colby Flood, the legend from North Carolina. What a great guy Colby is. Uh, do check that out. You can find these and our entire archive of episodes on our website for free at ecommercepodcast.net.

On our website, you can also sign up for our newsletter, and each week we will email you these links along with the moats and the moats? Uh, the notes and the links. Uh, from today's conversation with Max, uh, they go direct your inbox totally free, which is amazing. Now, uh, Max, I'm sure like you, I love to help people take their e-commerce businesses to the next level.

You know what, I've been working in studying e-commerce for years, since 2002, and during that time, I've developed, well, let's just call it a unique methodology called e-Commerce Cycles. Uh, it's a system that helps businesses grow their revenue by identifying and exploiting opportunities. I use e-commerce cycles on my own e-commerce sites, uh, and I also use it with my clients.

And we have seen incredible growth generating over a hundred million in online revenue. So if you wanna achieve the same kind of success? No, I'm not gonna give you that kind of guarantee. But what I am gonna say is, uh, we have got some free training, uh, that will take you through the e-commerce cycle methodology and show you exactly how we use it in our business.

So if you are in e-commerce, uh, in , I'm a bit tongue-tied today. If you're in e-commerce, check it out. Head to ecommercecycles.com for more information. Do check that out. It is a free training. Hopefully you'll enjoy it. So let's talk about Max now. Max is the CEO and founder of Ecomtent who are revolutionizing how e-commerce sellers create content with generative AI.

Sounds all very posh, doesn't it? Now, prior to founding Ecomtent, uh, which is the, it's a word you get when you merge ecommerce and content together. Uh, Max has spent six years at Amazon. Yes, he has where he worked on the launch of Amazon here in the uk, uh, Amazon business here in the uk, should I say. Uh, the country launch of amazon.sg, which if you're not in the know, is in Singapore, uh, and the launch of Amazon Grocery across the uk.

The man's done it all it seems with Amazon, uh, and throughout his time he has worked directly with hundreds of sellers of all sizes, across many categories. And he saw the pain. Oh yes. Which we can all identify with. The pain of creating content, uh, for e-commerce firsthand. So hence, I'm guessing Max, the reason you started Ecomtent.

Uh, welcome to the show. Great to have you. Great that you are here. How are we doing?

Max Sinclair: Thank you. No, very excited. Great to be here. As I said, um, fan of the podcast, so it's really cool to be on it and yeah. Excited to get going and talk about generative ai.

Matt Edmundson: Oh, absolutely. Its great to have you. Now you have, uh, a decidedly British accent.

Are you calling from the good old islands of the UK or are you wanting to sort of

Max Sinclair: less exciting than few other guests. I'm here in London.

Matt Edmundson: No, not at all. I, you just want, you, you know what you can't do when you talk to guests, assume from their accent where they're from, because everyone now has a digital nomad visa, don't they?

And they're all working in very posh climates, uh, and um, and very tropical places. So I just wanted to clarify. You are in fact dialing in from the great city of London itself.

Max Sinclair: There we go. Exactly. West Hamstead, to be even more specific..

Matt Edmundson: Brilliant, brilliant. Why, did you, um, when you worked at Amazon, did you work at, uh, Amazon's HQ In, in the center of London then?

Max Sinclair: Yep. I, in Shoreditch, um, also have worked in, as you said in the intro, in the Singaporean office. I've been to the Milan office, I've been to Luxembourg. Uh, Bratislava, uh, I at one point helped manage the sales team in Bratislava. So yeah, being, being global with Amazon.

Matt Edmundson: Um, being global, yeah. Well, there is a chance we may have crossed paths down at Amazon HQ in London as I've been to that building many times,

Um, it's just easily Yeah, yeah, yeah. It's be, it would be ironic if we had, uh, you. So you worked at Amazon. Let's jump into this, right? Yes. So you worked in Amazon, um, and we said in the bio that you saw the sort of the pain of creating e-commerce content sort of firsthand. Mm-hmm. . Was that what led you down this journey of generative ai?

Was there. I'm just curious because that it's a heck of a leap. Sure. You know, from Amazon to, to new content.

Max Sinclair: Yeah. Yeah. So I guess what really was the, um, was the inspiration was this kind of onset of generative AI itself. And I think I, um, I kind of saw the, saw the potential. And, uh, with my co-founder, we kind of launched a business which was, um, within both of our understandings.

So, yes, like I understand the pain that, um, sellers have, uh, when they're creating content, but more, I mean, if I'm being honest, more generally, like I think the power of generative AI is transformative. I think it's gonna transform pretty much every industry that we. You know, every industry. Um, I mean, I can go into a bit more details about what generative AI specifically is, because I think a lot of listeners will be saying, oh, a AI's been around for ages, right?

Like, what do you, you know, everyone's heard of ai we're kind of very used to ai, um, artificial intelligence in Google ads and, you know, and all sorts of stuff, right? But this is really a, a big technological shift that has happened. in the last few, in the last few years and has been commercialized really in the last few months, you know?

Mm-hmm. . And, um, that's, um, that's kind of what, um, what really excites me and yeah, I'm, I'm, I'm looking to bring this to an industry that I know well, and that's kind of what motivated me. Okay.

Matt Edmundson: So, um, Well let, let's deal with that straight away then. You, so we used this phrase, generative AI in the bio. You just kinda mentioned it there.

And like you say, AI has been around for a while, but generative AI is quite a new thing. So what do you mean? Let's, let's differentiate terms. What do you mean by this?

Max Sinclair: Definitely. So I'm gonna, I'm gonna label kind of AI as everybody knows it as deterministic ai, uh, and this deterministic AI was kind of conceived in, in the 1950s and kind of commercialized in the 1980s.

And what you have here is a computer, uh, able to mimic the problem solving capabilities of a human to complete analysis on big data sets and broadly either do data classifications or regressions. Mm-hmm. . So what you do is you, you feed in a large dataset and the machine will tell you, this is type A or type B, or this is a regression and how these numbers relate to each other.

And it's used, um, you know, in, um, in medicine it's used, uh, you know, we use it at Amazon to kind of, uh, identify like, are these ASINs parent and child asins attributes? Should we merge them? Um, this kind of question, uh, what's happened very recently, um, really commercializing kind of the back end of 2022. Is a second wave of ai, which is called generative ai.

And this is where a computer is able to produce entirely new content. So new images, new video, new texts, new code. And what happens here is you train a, um, train a a machine on the patterns and characteristics of the input data, and then you use it to kind of generate new and similar content. And, um, this is, this is a, a big step because in deterministic ai, the answer was really in the data.

So you are kind of the, you know, you're looking at large pieces of the data and saying, this is, I, you know, your, your answer for what your question be it, how is these relate? Or who, how should I target this person with this ad? Your answer's in there. Yeah. With generative ai, you are kind of, you, you are creating new things.

So you're kind of saying, this is a data now make it look like this, or do this or do that. And you're creating new content. Um, and this is gonna, I mean, I can go, I can go a level more deeper if we want to, but like this is kind of the, um, this is a step change. Um, and I think it's gonna have really, really broad consequences everywhere.

And, um, it's very exciting. So, yeah, I think, um, when I, when I, what kind of being saying on top of this and bringing it to e-commerce and tailoring it to, um, sellers to help them is, uh, yeah. Is what is kind of what, what is driving me in, you know, the reason why I started ecomtent.

Matt Edmundson: Yeah, yeah, yeah. So the. So generative.

I mean, it all sounds very exciting, but it all sounds a bit Terminator. Uh, Do you know what I mean that, um, Skynet's gonna take over the world and destroy us all. Um, it was interesting, uh, the time of recording this. Uh, as you may or may not know, dear listener, when we record the e-Commerce podcast and when it goes live, there's usually a quite a significant time difference between the two.

Or at least there is at the moment. We're, we're making steps to change that. But, um, at the time of recording, uh, last week I did a LinkedIn live, um, on, I do this thing called on a Monday lunchtime, the LinkedIn live chat show type thing. So if you're on LinkedIn on a Monday, come join in. Um, connect with me on LinkedIn.

You'll see its there. And last week, um, I was on chat, GPT, which is probably the most, um, in the news Generative AI system that I can think of at the moment. Right. And I just, I just put into that chat GPT thing. What are the top five trends that I need to be aware of in e-commerce for 2023? And it was just really interesting, some of the answers that it mm-hmm.

came up with. So what you are saying is the answers that, um, open Chat came up with were all completely, um, , uh, generated. It's all new fresh content. Yes. Wasn't like it just went through a blog post, copied and pasted mm-hmm. , how did it, how did it come up with those five answers then?

Max Sinclair: Okay. So in, I'll go very quickly into the technical, technical explanation and apologies, but I think it's quite interesting just to understand like what's happening.

Yeah. Um, generative AI uses, uh, something called Generative Advisoral Network, and basically you have two networks in here. you have the first network is a generator and you give it a load of input data in chat GPT's case. This is all the text on the internet up to 2021, which is important. And I'll, I'll come back to why that's important.

Yeah. But you give it all the data up to 2021, the machine and the generator learns the patterns and characteristics of that content and then tries to create new content similar to that. And then you have a second one called the discriminator which is trained to distinguish real content from generator content.

So it's presented with the original content, it sees the content from the generator, and it basically goes yes and no. Yes, no, you know, in, in microseconds. Mm-hmm. of does this look like the original content or not? And this will happen, you know, trillions of times a second. And then once he, um, You know, once a generator fools the discriminator enough, then it publishes the content.

So it looks like the original content. So this is kind of what's happening. Um, you know, when we talk about generative ai, which is, which as I say, it's, it's a new, it's a big technological shift on how we think about like machine intelligence. Um, oh, just, just to return to the point that you said on chat, GPT, I. Just explain a point here, which is chat GPT and, and I, I think we'll go into, you know, some, some ways that I, I would recommend using it because mm-hmm.

At the moment it's free. I dunno, when this been published, it might cost $42 a month. I dunno. But at the moment it's free tool to use as we are recording. Yeah. Um, But what it is is a language model, and it is the point of a language model is to gen, as I said, it's to generate patterns and text similar to the data that it's trained on.

This is not accurate data. This is. This is, is trained to sound like a human talking and it's trained to, you know, give you an answer that sounds correct, but it's not, it is not necessarily accurate. So we should always, like, there's, there's some very useful ways of using chat GPT, which we'll probably go into later, later in the podcast.

Mm-hmm. . But, um, we should, we should always have that in mind that. What this is doing is creating content. That sounds correct. It's not create, it's not giving me accurate content. And I can give you a, um, an interesting example of this. Uh, my, my, um, girlfriend's brother is a doctor, and when he puts in chat gpt, you know, some medical questions, it will then give to him like a medical.

Uh, like a medical, um, kind of article as an answer. Mm-hmm. As it is trained to, right? As you, if you ask it to write in this spirit of Shakespeare, we're doing it in Shakespeare, blah, blah, blah. Uh, but then what we'll do is we'll completely make up the references cause it's seen online that, you know, these come from references and you should make, you know, you should reference random, like mm-hmm.

professional sounding names. So it kind of gives you this medical answer with references which don't exist. So, and that's because that's what it's trained to do, is to, is to kind of replicate the data that it's trained on. it is super useful and there's many ways that we should use it, but we should always be, um, we should always know like fundamentally what the technology is and, and what like, and what, what it's supposed to be doing.

Matt Edmundson: That's really interesting. And you, the other thing that you mentioned about Chap GPT is the data that feeds it only goes up to 2021.

Max Sinclair: Yes. Yes. So this is, um, so I mean, this is another point, right? That, um, the data is, they say officially it's a 2021. I think there have been, I mean, I follow this stuff as I'm sure many listeners do on, on LinkedIn.

So there's been like, it's a live model. They're training it, right? And, um, I think. It they are, might be updating parts of it just to kind of experiment, but officially it's only been trained until 2021. Um, so it is not the most up to date, but, um, I, I'm not sure if you're aware, but they're kind of working on GPT four right now.

Um, and this is going to increase, um, the number of parameters. So the number of data sets that, uh, you know, the machine is trained on from 175 billion to a hundred trillion. So what this is to kind of con, these numbers are really hard to contextualize, but to contextualize it, this is like going from earning 500 K a year to 20, uh, 285 million.

uh, pounds a year. Yeah. So it's a, it's an, it's an enormous upgrade. Like it's an upgrade kind of, it's forget like os iOS updates, what, what you're gonna be able to do with GPT four. The, the rumors are, is you're gonna be able to interact with both text and speech. So you could talk to it and get your answer and, and kind of at the moment you use it and it's amazing.

But like with GPT four, you could use it and write a 60,000 page book from a single prompt. So It's, it, the, um, the speed at which this technology is, is moving, is incredible. Um, and yeah, it's just a super, super exciting space, um, in, in my opinion.

Matt Edmundson: It's a really interesting one, isn't it? Because one of the things that came up in my little LinkedIn conversation last week, uh, Max, and I'm kind of curious to see where, cuz you are involved in ai, um, is where you sit.

Yes. Sit on this because if I can go to chat GPT and tell it with a single prompt to write a book for 60,000 words. Yes, that book's not really come from me, has it? And it's. There's a sort of this, this fine line, isn't it? I could go and market. I mean, there are people on YouTube now. I've seen the videos, they pop up in my feed.

I've not watched them, I have to be honest with you. But the, the use chat, GPT to write your first book, um, and publish on Amazon, you know, turn yourself into a millionaire kind of thing. And it's, it's a really interesting conundrum, isn't it? Because all of a sudden, um, creation. in some way stops being from the creator.

It starts being from the machine. And I'm just really curious, I dunno if I have an answer, Max, uh, Max, but I, I, I feel like there's this sort of very big gray area with ai. Um, and I'm, I'm not quite sure. I, I get excited by AI and some of the things it can do. Mm-hmm. , but some of the other things like being able to write a book.

With a single prompt. I just, I dunno how I feel about that, if I'm honest with you.

Max Sinclair: It's, well, I think what I would say is it's terrifying in some level, but also it's very exciting. And what I would also say, is that like this is coming and it's coming for every single industry. and like I was Out to dinner last night with friends of mine, one of them is a lawyer.

And I was explaining to them like what you're gonna have in the future is you're gonna have an a generative AI model trained on all the law in the uk. You're gonna give it a prompt like, my neighbor has built a fence over blah, blah, blah, and it's kind of blocking my sunlight and what you know, and it will just give you the answer because it's like,

Chat GPT can't do this because, and I think this is a difference between like an open source model, which is like chat GPT and kind of a private model, which is like what we have at ecomtent and like what you would have where you say, okay, I'm gonna build like a chat GPT for law and I'm actually gonna make it correct for law, which is not, you know, it's not chat GPT, you couldn't do this with the law, but, and you, and I mean, it's gonna come, it's, it's gonna come for every industry.

Um, so I think the genie is out the bottle, right? Like you. There's, there's kind of no, there's no going back from this. Um, but I think it's very exciting. Like, I, I, you know, I'm a, um, I'm a firm believer that kind of technological advances are good for humanity. Um, you know, if you take us all, uh, a thousand years ago, we are all, you know, 80, 90% of us were working in agriculture as, uh, as pretty miserable.

Um, you know, Terrible. No education, terrible diet, terrible life expectancy. And now we are kind of. You know, every, probably everyone listening to this podcast has got a job, which is far more exciting than a, um, than working in a field. And that's based on technological advances and the population, um, in the UK has grown from one and a half million to, you know, 70 odd million.

And we have, uh, 4% unemployment. And, you know, pretty much every, you know, they're obviously still some farmers, but they're doing it at much better scale thanks to technology. Mm-hmm. and everyone else is doing jobs that would be completely impossible to imagine if you a surf in, you know, the middle, the middle ages.

So I don't think that technology is something to be scared of. And I also don't buy the argument that we are now at the forefront of some revolution, which is so much better than any other revolution we've seen because like that's exactly what the Victorians thought when they're looking at like the steam engine.

And that's exactly what everyone thought that you know, when you are looking at your period of, your life. You're like, okay, this is much bigger than ever before. But if you look at human history, technology has definitely been a benefit. And there's kind of, you know, there's a lot of progress that we could make, with it.

So, you know, it is, it is scary in some sense and the world is gonna change, but, uh, you know, I think it's very exciting.

Matt Edmundson: Yes. Uh, and like I, I, I totally agree. I think it's a really interesting turning point, isn't it? It's, um mm-hmm. , I'm kind of curious to see what happens to the lawyer when, uh, you, I, I mean, I actually have a, a barrister friend of mine who is actively working on the, the legal version of this kind of AI type thing.

there we go. There we go. It's really fascinating what he's getting into. Yeah. And the amount he has to spend on computers just to do the processing is just unbelievable. Anyway, eyewatering amount of money, but, um, . So I, I, I, I, I'm just kind of curious to see what happens to, yeah. That lawyer and the transition that has to happen to take place.

Um, I can see, you know, for me, one of the things that I would love, I would pay money for actually, is you, is if I didn't have to worry about social media, you know, how do I build an Instagram following, there's this AI system over there that goes in that mm-hmm. , you know, sort of examines your niche and it'll create all the content for you.

It'll build your following. It'll do it all in an organic way. It'll be totally natural brilliant. There you go. Um, and it will,

Max Sinclair: that's what we're trying to build. That's what we're trying to do right now.

Matt Edmundson: Well, I'm your first subscriber man. Let me, here's my Instagram login. Go for it.. Um, cuz I can see. But a again, I mean mm-hmm.

as soon as it starts doing that, in a few years time, people will just stop watching social media because they're just gonna, well, maybe they will, maybe they won't. I don't know. It'll be an interesting one. So ecomtent then. So we we're in the midst of this sort of technological revolution. And it is, it's both exciting and it's scary.

And yes, there are inevitable opportunities for the savvy e-commerce entrepreneur, right? Yes. And so for you, I'm guessing you've looked at that and gone, right? This is where, this is where we fit in with ecomtent. We are gonna create some stuff over here which e-commerce guys can use. Exactly. That is generative ai.

We're gonna build on that technology. We're gonna help you guys rock and roll with it. Um, so what sort of things then are you guys doing and playing with right now?

Max Sinclair: So I can, I can say what we're doing right now, which is we are creating product images and lifestyle images. So what we do is, um, similar to how I explain how the technology work, we train, uh, uh, a private AI model.

on, um, you know, three to five photos of a product. Mm-hmm. , and then we use generative AI to put that product, um, in any scenario, uh, with someone of any ethnicity, any gender, any, um, you know, any background, uh, based on the prompt, so mm-hmm. , you could say. Um, I mean, on our website for example, we've got the kind of demo video of this elephant toy and you could say, um, you know, from the website we go like.

put it with a child on a beach and you see this elephant toy being held by a child on the beach. Mm-hmm. . And, um, we, you know, this is kind of step one and we have like a very exciting roadmap. Um, You know, the ambition is to, you know, create all content which is more optimized using generative ai. So that's descriptions, bullet points, uh, a plus content, everything we wanna do.

Yeah. But right now we are kind of, we focus on the hardest bit, which is a lifestyle imagery. Um, because we know like that's very painful. It, you know, to kind of organize these shoots is, um, is a pain and it's costly and it takes ages. Um, so we, we do that with generative ai and you can kind of create Thousands of images instantly with our, with our self-service tool.

Um, so yeah, that's, that's what we're doing. Um, that's what we're focused on right now.

Matt Edmundson: So if I, I'm just kind of curious here. Yeah. Max, if I'm honest with you, cuz we, we, our product shots are 3D renders on our eCommerce websites because, um, they're so realistic now. You get much better lighting, you get exactly how you want it, right?

Mm-hmm. . Um, and it's easy to do for our products. I've gotta admit, um, like with our supplement brand, for example, it's a cylindrical bottle with a label on the front. I mean, he can't get really that much easier, uh, you know, in terms of 3D generation. Mm-hmm. . , we still use a photographer to do the lifestyle shots.

Now obviously they are restricted to, um, I don't pay them to go all over the world and generate these shots. They, they sort of do them here. So I'm restricted to what is available to me here in mm-hmm. Liverpool in those photo shoots. Um, so I'm intrigued by what you do. So I'm thinking of the elephant on the beach.

Yeah. Is your system then taking a picture of a beach and the elephant and just being really creative with photoshop. Or is this beach, which it's generating a beach which has never ever been seen before.

Max Sinclair: Exactly. Correct. Exactly. So, um, it's tra you know, we, we have private data sets, we train it on, but basically it's hundreds of millions of images.

Um, and then it will create a net new image based on your prompt. Um, so if you say a beach, if you say a beach of sunset, it's gonna kind of understand the, our model will understand what a person's product is, uh, the customer's product is, and then it will kind of build that product into a scene. Um, amalgamating all of the, um, you know, photos on the internet of whatever the background is and whatever the whoever the person is interacting.

And I wanna make an important point, like, , this is not kind of the, um, kind of background removal, kind of swapping a background, which you kind of see everywhere. This is kind of creating a new photo so your, you know, your elephant with a child in, in natural setting held, you know, held as, and it's kind of understood like this is the elephant toy, you know, is similar to a teddy bear.

The teddy bears are held by children. We've got like millions of images of how they hold them and how they interact with them and the smart, and it kind of will recreate an image similar to that. Fundamentally a new image that's never been seen before and also would like, would never be generated again exactly the same if you gave it the same prompt a million times over.

Matt Edmundson: Wow. So, you've got some really clever computers. Uh, At, uh, gchq.

Max Sinclair: Yeah, I have a fantastic, uh, fantastic, uh, CTO who is a PhD in ai and Is, is, is super smart and is kind of building these very complex, uh, very complex models.

Matt Edmundson: Yeah, no kidding.

No kidding. So that's, um, , that's the lifestyle shots then. And I, I, yes. I can see how that can work. Um, I'm just, I'm kind of, I'm gonna go check out the site and, and play is what I'm gonna do. I wanna see the elephant thing. Um, because yesterday I was on, uh, we use a system called, I dunno if you've heard of it, called Jasper. jasper.ai helps

Max Sinclair: us write content.

Yes. Jasper ai. Yeah. Yeah.

Matt Edmundson: Um, it's a mixed reviews from me really. Some of it's good, some of it's bad. They've got this image generation feature on there. Um, and yesterday, uh, I was like, can you generate for me an image of a trillion dollar bill? Um, Because e-commerce for the first time has, has gone through the trillion dollar mark or whatever it is.

So I'm like, I'd, I'd like to talk about this, uh, on, on our, one of our LinkedIn lunchtime events. Yeah, I'd like a, I'd like this image, could not do it. Um, and it may be the prompts that I were putting in or what, it was just really interesting. So there are times when it's good and bad and, um mm-hmm. , I, I. , I get how this stuff is helpful.

Max Sinclair: I think it's, I think it's a different, Jasper kind of pretty much use open source models, which means they're like, they're, it's a bit like chat GPT. They're generic. They can basically do anything but not everything well, if that makes sense. Yeah, yeah. Whereas if you had, in that case, like a train model on creating like, you know, artwork on, on, um, you know, bank notes, which is very niche, but like you could imagine someone, like if there's a huge demand for this particular request, then you'd, then you'd be able to do it brilliantly.

So it's just a question of like, this model is kind of a generic open source model. Um, and therefore, It can do, you know, it can get pretty good at most things. Um, first is like a private model, which is going to then like be very specific at like under understanding British law or like understanding artwork on this, on this. Whatever.

Matt Edmundson: Yeah, I get the power of that as well because like you say, you know, these systems work as well as the data you feed it, so if you are creating an AI system, which generates a plus content and all you are doing is analyze an a plus content from really successful listings, which you're gonna know from your days of Amazon, then I can, I, I look at that and go, that's gonna create much better resources for me than say Jasper.

And so therefore I'm, I'm, I like that that's there because yeah, that's where it gets really, really, clever.

Max Sinclair: Exactly. That's the, that's the, that's the goal of what we are, we're doing. Like I think, um, we are gonna move to a world where, like you have a lot of open source models and then a lot of people building like private models for specific industries or whatever, kind of using the latest technological advances.

But then like using that in a sense, like, okay, let's apply this to whatever we're doing. And kind of building that and using specific data sets like as you say, all the best performing a plus content, um, to kind of generate the best outcome for our customers.

Matt Edmundson: Really interesting. Really interesting. I do wonder Max, whether at some point mm-hmm.

um, you're gonna get clever enough where I don't even have to have a, you know, that you go to a web design agency, like Right. Design for me. An e-commerce website. Yeah. And they'll charge you whatever, 30 grand, 40 grand, 50 grand, whatever the price is these days. Yeah. It's a hundred bucks. If you go to Shopify, it's probably 120, 150 grand if you go to a big magento agency.

And if you, you know, Somewhere in between or whatever. So I'm gonna go and spend, say, 30 grand on a website design. Is it gonna get to the place where actually I go to people like you and I say, listen, this is my product. Mm-hmm. , um, I think these are my target customers who will buy this product. Um, based on the research that I've done.

I feed those parameters into an engine and it goes right. We think based on that data, this is gonna be the best, yeah. The website for you, both in, in your colors, in layout, in branding. Here's some content we think you should put on there. Mm-hmm. . Oh, and by the way, if, um, a, a man in his thirties comes on, we're actually, and we know that we are gonna tweak it around, so the colors are slightly different.

That, and, and the website almost morphs and changes. Depends who, who comes on it.

Max Sinclair: I think I think, you're completely right. And that's kind of like our, uh, our like vision statement we have here, which is like unimaginable creativity. Like we think this unlocks unimaginable creativity for all of the creative people we work with, all the social media managers, all the brand owners, et cetera, et cetera.

And also limitless personalization because now, You can just create an infinite amount of content, right? Mm-hmm. , you can personalize it to an inf uh, to an infinite degree, as you said. So you could have, like the man in his thirties, he sees this, the woman in her sixties, she sees that. And I wanted, I wanna add one more thing, which is like, Um, generative AI can create code as well.

So it's not just a case that this is gonna say, here's your, this is what you should do. It's gonna be, and i, this is something I firmly believe will happen in the next two, three years. You already have generative AI code in, Microsoft and Amazon, like Amazon Whisper, if you're a coder, but, um, you're gonna have, literally a free, free text like, build me a website.

This is my product, this, as you said, and it will just build the website because it will be smart enough to do the code as well as understand all the, permutations on back you know, as you said. So I think, that's definitely coming.

It's not something that we've, like, we're not focused on that particular thing, but we are gonna see, you know, I would, I'd bet good amount of money that we're gonna have kind of generative ai, you know where site creations and app app creations in, in, you know, in the not too far off future.

Matt Edmundson: So everything that you've talked about, um, you know, code is very two-dimensional. Uh, text is very two-dimensional. Images are very two-dimensional. Um, are, we getting to the place, I mean, I, I see it a little bit with deep fake videos, but are we getting to the place where actually I'm gonna be able to create TV shows for want of a better expression using generative ai.

Max Sinclair: I think like, I, I think we're probably further off from that, but like we are looking at videos like, we know that videos are more, um, like kind of get higher engagement on social media and also a like best practice to have like a mini video on your Amazon listings.

You. , you probably know well. Um, so we definitely, like video is on horizon. I don't think the tech is there, um, now, but, um, I think that's, um, that's not beyond the realms of reality. Um, I think this is gonna really shift. , you know, this is really gonna shift most, most things. Um, I think it'd be hard to, hard to number the things that generative AI won't touch, rather than talk about like what, what it will do.

Because like if you are taking back to basics, like you're just creating new stuff, right? Which is what most people do in their jobs day to day, like, and if that can be done more optimized, quicker, faster, better with more data sets, like it's, it's hard to see where it doesn't touch.

Matt Edmundson: This can be really interesting and I'm, I'm very tempted to go off in this conversation, which says, will Netflix not actually have any shows?

You're just gonna, it's gonna understand who you are as a person and show you stuff. Yeah. It's gonna make stuff up on the fly that only you see. Yeah. Um, but let's, let's, let's bring it back to a bit more, uh, stuff that we can do today, right? Yeah. So, um, if I'm, if I'm sat here right, and I'm listening to you going, you know, Max.

This all sounds very intriguing, but if you were starting an Amazon business today, or if you were starting, maybe not Amazon, maybe your own online e-commerce website today, um, or even you've been around for a while, but you've not really hit the AI thing yet. Um, I mean, you've heard a few things on it and you've maybe played on chat GPT, what are some of the things that I should be thinking about, um, as that kind of person, what, what, what things are gonna really help me in e-commerce?

Max Sinclair: Sure. So I'll, I'll start with the chat GPT, um, because at the moment, at time of recording, as I say, it's free. And by the way, I think they've done the most incredible marketing campaign ever.

Matt Edmundson: That's been unbelievable

Max Sinclair: to, to release it for free, get everyone on all these podcasts and LinkedIn to be like, these are the ways you can use it. And then just charge two months later.

Matt Edmundson: I've already had the email saying it's gonna be 42 or whatever it is. Yeah, yeah.

Max Sinclair: Which I haven't, I mean, I use it almost daily and I haven't, I haven't done it yet. I don't. Anyway, , I think to, to answer your question, so you are a new e-commerce person. Number one thing I would do, is to, um, you know, let's say you know that you want to go into a specific category like toys, cuz we talked about toys.

Mm-hmm. . So I think the first thing you could do is to use chat GTP to help you analyze the reviews, um, on competitors and understand what customers are looking for in terms of. Um, you know, what they like and what they dislike, and you know, you can, you can flip it either way, but you can literally say, as long as this, by the way, again, as long as the ASIN is before 2021, or you kind of copy and paste manually.

You could, you can, if, if the, if it's an old ASIN, you could just paste in the link. It's chat GPT. If it's an, if it's a newer asin, you could just copy in all of the customer reviews, place that in and then say, analyze reviews on the on this. But you could say, you know, analyze reviews on this ASIN you're gonna give me, tell me what the customers love, or give me product recommendations.

And then it will kind of spit out like the key takeaways. , um, for you to look at, and then that kind of, that will kind of set you off to say, okay, people love this toy, but, uh, they've complained that it's, um, got a few defects and the stuffings uneven, or that it's not the size is, you know, it is too small when they thought it'd be bigger by the photos or, you know, blah, blah blah, whatever.

So that kind of gives you an idea of, this is my product that I want to launch. Um, another good way of,

Matt Edmundson: um, sorry. Before you get into the second one, Max, lemme just clarify something there. So I could go and grab the URL off an Amazon product listing. Yes. Paste that into chat GPT and with a command, which says something along the lines of, analyze the reviews on this page.

Yes. And tell me, you know, the things that people love about this product and the things that people, you know, that, that what are the main complaints about this product? Yeah. Yeah. Mm-hmm. hit the enter button. I'm sorry to be over simplistic. Uh, I, I need it to be simple for me to understand, um, hit the enter button and it's gonna come up with a whole bunch of recommendations for me.

Is it as simple as that or am I

Max Sinclair: It is as simple as that. It is as simple as that.

Matt Edmundson: Okay. So I don't need to be more descriptive. In my commands.

Max Sinclair: Um, I think there's a very interesting point, which is like, prompt engineering is gonna become a, a very important, uh, like future job type. Yeah. Where like, how, how do you interact with the ai?

Like what are the best prompts to say? I mean, I've got a few things that I've noticed, which I can share, which I use when I'm writing blogs on my website or whatever, whatever, which would been interesting. Um, but, um, Yes. Like, it's as simple as that. You say, what are, you know, what are the gimme five takeaways on like, um, what people love or gimme five takeaways for product improvement or whatever.

And it will understand and it'll give it to you. So that's kind of a very good starting point. Um, if you wanted to start with, um, you know, start, start a business as you said on like, with a, like a, you know, D2C business.

Matt Edmundson: That's incredible. Okay. I've got that. That was number one. Uh, what's the number two?

Max Sinclair: Um, so you've got your product, um, you've done all the hard stuff, which is, uh, manufacturing and, and everything else. I guess next, you know, at some point you're gonna have to look at, um, you know, keywords, seo, that kind of stuff. So, um, and I've view this, um, in my, in my business, I say, you know, give me a, you know, literally give me a list of a hundred search phrases for a startup sell self-service tool to create lifestyle images for generative ai.

And it will just give you a hundred keyword words and it will say, and what's brilliant is that I'm obviously British and I like will think, you know, like I phrased, if you read the, the website or whatever, you'll, you'll, it's obviously very come from me. Um, and like I have a certain way of talking about my business.

But like what you can do at chat GPT is think about a hundred ways of like describing the same thing. And like all of, you know, not all of them are gonna be great, but it's helpful to see, okay, these are some things that I've missed. Like these are some keywords which are interesting or whatever, and I can, you know, I can better optimize, um, your, you know, your Amazon bidding or whatever, or your, your Google, um, based on that.

So I guess that would be my number two.

Matt Edmundson: Okay. So. I'm just trying to, so you are now using chat GPT for keyword research. Yeah. Um,

Max Sinclair: again, it's not, it. It, I, I like, I know what chat GPT is, which is a natural language model and therefore like, is very good at doing natural language. So it will rephrase things very well for me. Um, so like, you know, it can be a bit of a mouthful explaining what I do, um, cuz it's a new technology and it's whatever.

But like using that will help me like see 10 different ways I can or 10 different Google potential Google searches, uh, or. Or 10 different keywords. Exactly. Okay. Um, so I mean, it's not, it's not based off any, like, it's not based off any inherent data, but it's just, it's, it's a good reference point to see different phrases and keywords.

Matt Edmundson: So do you. I, I'm just trying to think. I mean, it's great. You can use it for SEO and keywords. I, I think mm-hmm. , um, I extrapolating that out. I'm thinking, can you use it then, or how would you use it? Uh, cuz can is a wrong phrase. How would you use it to generate, say, content idea or content ideas for your blogs, for example?

Max Sinclair: Yes. So this is a, is that, this is a great one. Oh, that's, yeah. So you can, like, again, it's, you know, it's very good at understanding language. So if you say, you know, you gave the example. Your, um, what, what was your product? You said the kind of, uh, the cylinder. The supplement? Yeah. Yeah. The supplement.

Supplement, yeah. So you could say, write a, write a blog or write an Instagram ad targeting this supplement at British men in their, you know, early thirties. And it would write you, um, A, an ad which like is, you know, using British, um, slang using kind of British calls to action, like the gym, the office, like it will understand all that very well.

And then you could say, okay, now do it for, you know, fifties to sixties women, and it'll give you completely different, um, slang and, uh, well phrases and, you know, like, um, how they would use it. And I mean, it, it like, this is something that's brilliant that is, is very good at, so if you kind of put in like, yeah, write an Instagram ad for my supplement, this is my target market.

you can generate, you know, 50 different texts and then you can swap your target market and it can kind of show you different, like it will again. It's a good way to kind of, oh, I didn't think of that because like, I'm not a woman in my sixties, but now like, you know, I've got this, I've got this, and it just g it's good for giving you some ideas for sure.

Matt Edmundson: Wow. And again, you're just literally heading over to chat GPT and typing that in the Yes, in the, in the, this is, I'm, I'm really intrigued the more you are talking and. . Mm-hmm. . The more I understand what you mean by this idea of prompt engineering, the people that will, that know what to type into chat GPT then become Yeah.

The ones which are currently gonna be the ones which are highly sought after, right. Because it's, yeah, definitely. I, the, the data you get out is all dependent on the prompts that you write and the information that you give it. Yes.

So what was some of the Go on,

Max Sinclair: yeah. I'll just say you like, it helps to have an understanding of the model, what's it's trained on, what the, like, what the data sets are because, um, you know, like different, again, chat GPT is a generic model trained on, like, we actually, we dunno what it's trained on, but they say it's trained on data on the internet for 2021.

Um, yeah. You know, they, they could have trained it on, they could have, I imagine they did ignore specific websites and, you know, otherwise it'd be saying horrible things about certain people if they had kind of trained it on the, or, you know, the entire internet. So I'm like, so we didn't really know what, but it's trained on, but like for, you know, for our private models, we know what it's trained on, we know what, we know what works well.

We can like tailor it to, you know, specific categories, which we do. Um, and, and that kind of, yeah. So I guess that's a big thing that we are thinking about is like, how can we help customers Is prompt engineering. You know, for example, we do a lot of work in furniture and we like train it on different rooms, aesthetics, rustic rooms, candid room, modern room.

But it's about like, how do we communicate to the customer, um, like, you know, this is what it, this is what the model knows and understands it's been trained on. And like, it understands different furniture types. You can put your, you know, your armchair in like a classic, uh, you know, rustic room by a fireplace, and that would be brilliant.

And then you could put it in a, like a modern. Whatever. And like it understands different, um, like a French star room and like a whate like different cult like device kind of suite. And it's all golden. So it un like how, how, how do, how do you Yeah, so that's a big thing as well. The, the, like, we, we think about every day it's kind of, it's um, you know, you can really do anything, but like, how do we help people to like, understand the technology and understand like, What works well, um, uh, in your, you know, what we, what we've designed it to work, what we've designed it to work well, right?

Mm-hmm. , um, which is like category specific for us. Um,

Matt Edmundson: so what are some of the, um, what are some of the prompt discoveries that you've made using chat GPT then, which is just the one,

Max Sinclair: I'll give you my, my, my personal top tip, which is, and, and I will, you know, I will not make a, we are both British. I will not make a comment on Boris Johnson's politics because, uh, it's not what I'm gonna do.

But the man is undeniably an extremely talented writer. He was a journalist for. Um, you know, 40 years or however long he was a journalist for. And he has a very distinctive, kind of upbeat, charismatic writing style. Yeah. Um, that is, that's, you know, quite unique and actually quite engaging to read again, like I'm, I'm not gonna, you know, I'm not commenting on, on the, the, him as a political figure.

I'm talking about him as a, as a journalist. Yeah. And I have noticed, or I know that, you know, It is the chat GPT seems to have been trained on a lot of articles, and many of those articles have been written by Boris Johnson, who's pumped out thousands, you know, nearly as many articles as children over his, uh, time as, uh, as a journalist,

And, um, and it is, you know, if you write a blog, it's, it can be quite boring. And like this is the thing. And then I will say, rewrite this in the style of Boris Johnson. Make it jovial yet informative, and you'll have this kind of brilliantly upbeat, um, You know, fun kind of style of, of, of blog writing. So, um, if you head over to the, the blog on my website, hopefully you'll see the blogs are quite entertaining.

I mean, they're not like hilarious, but they're kind of They're, they're, they're more entertaining. And that's because I've kind of copied, um, you know, I, I've copied as he, you know, as he probably has done many times, copied his, um, his writing style for

Matt Edmundson: brilliant, uh, brilliant. Brilliant.

Max Sinclair: So that's one that I use.

Matt Edmundson: Okay. So you are using chat GPT to write blogs in the style of Boris Johnson. Um, yes. You are using it to find keywords, content, ideas, um, understand products. Is there anything else that we need to, to think about? Uh,

Max Sinclair: Um, the last one, the last one I'll go into is natural language translation. Again, like remembering what chat GPT is, it's natural language.

So if you do Google trans, like no good seller should ever be doing Google Translate for their license. Like, no, no, no, no, no. But if you say, rewrite this ASIN in German, use German. You know, phrases, um, and it will kind of re, you know, I'd like, I don't speak German, but I've tested this in Spanish with my Spanish friends.

You know, rewrite, rewrite this in Spanish, using Spanish phrases and it'll write it in, you know, invoke here in Spanish using kind of like phrases, you know, you like different languages. So we have kind of many metaphors and synonyms, which make, translated, make zero sense in other languages. Yeah. , like, you know, making a mountain out of a mole hill or something like this, which like if you translate it into Spanish, like would just make zero sense.

Yeah. And that's kind of one of the challenges of translation is that like a deterministic AI will translate verbatim word for word and it will sound weird. Whereas a generative AI understands the key themes and the key concepts of the listing, and then it'll translate that into the other language and it'll create like net new

phrases and synonyms and metaphors to replace ones you had. So that's a, um, you know, that would be kind of, My fifth tip. Um, or like I'd say, uh, with chat GTP is very good in natural language translation. That's literally kind of like what the research, that's how this was discovered. Like the, the initial research into gans and, you know, uh, everything I talked about at the beginning was, was, was initially.

Thinking about translation and how do you solve for that translation problem? Yeah. Yeah. So it's, it's excellent at doing that. And yeah, I, I'd recommend using it.

Matt Edmundson: Fantastic. So there you go. Top tips for using chat GPT's tool, which is technically still free, although. I think at some point, like you say, they're,

Max Sinclair: you've gotta get this podcast out quickly, Matt

Matt Edmundson: Otherwise just have to pay for it.

To be fair, it's still worth, if you're gonna do all of that, it's still worth the 40 bucks a month, right? So, um, just subscribe and have a go and see how you going. Now what I need is an affiliate link for chat GPT .

Max Sinclair: I can, I can give you one for ecomtent. Unfortunately I'm not, uh, you know, I'm not Sam Altman.

Matt Edmundson: No, no, it's fine. But um, so well let's talk about ecomtent then. So I'm really curious in the last few minutes of the show, bearing in mind that we've talked about, you know, there's free stuff. Obviously ecomtent will be a paid system, but it's gonna be much more niche. Right. And I get that cuz you've not got Bill Gates or whoever funding,

Max Sinclair: yeah. Yeah. . No, we don't. Um,

Matt Edmundson: so, so you, where do you see the future going for you guys? What sort of things do you, do you think will be starting to happen?

Max Sinclair: Um, for us specifically, I'm super excited. As I say, like the goal of ecomtent is to bring this generative ai, uh, revolution to the e-commerce industry.

Like I've worked with hundreds of sellers, I've got many friends who are sellers, um, you know, many professional relationships I've built over the years. Um, I really kind of admire you. I've always admire their kind of entrepreneurial spirit and their, um, kind of savviness, like super savvy, the savviest people.

You know, I've come across and we are, we're, we are just gonna focus on like bringing generative AI to E-commerce. And one of the things we haven't done so far is text. And the reason is like, I didn't want to charge, I didn't want to even begin to charge people for something which was fundamentally free.

Um, in chat GPT. Now it's not free. Um, well, when it's not free. I think it's very interesting because like that gives us the opportunity to say like, we, like if you want to use, um, generator AI for descriptions and bullet points, like you can pay $42 a month at, um, um, Uh, for at OpenAI, like our prices pretty much similar, but also we'll be doing it specifically for like Amazon descriptions and add bullet point.

So I think that's something that we'll definitely be focused on. As I mentioned, we want go to all content, so that's video, texts I just said. Um, what else? Um, I think we want to look at, um, we want to continue to look at. Optimization and make sure that like, we are, like we're, we are creating like optimized content based on like, um, likes and this kind of stuff.

So it's super exciting space. Um, we're, um, we are, we are like very wary of kind of like darley in these free tools and we are like, we are very focused on number one thing we we're focused on is quality and like, if you do use our tool, you'll see like the images of the products, they never get, they're never weird.

Or like, they're hardly ever, like, only like 1% or 2% of them are kind of weird, like AI kind stuff. Like we, we, we continue always to focus on quality. We wanna be the, the, um, we wanna be the best, um, at this. And, uh, Like as competition comes, which I'm sure it will, we want to just be the, the leader that, you know, you can, you can't trade and you, you know, maybe people will undercut us some price, but like, we'll still be the best in terms of like building images that look super realistic and, you know, you can take your product anywhere.

Matt Edmundson: Yeah. And I think you've got a real in, I'm, I'm curious to see where your journey takes you, mate. Cause I think you've got a real interesting competitive edge in all the time you spent in Amazon um, and I can see why you would double down on that side of this.

Max Sinclair: Yeah, exactly. I managed catalog quality for the launch of Amazon Singapore, so I, um, yeah, I kind of really understand this space. It's quite fortunate and, um, so, uh, so yeah,

Matt Edmundson: it's a really interesting, uh, thing. So Max, listen, it's been great chatting to you. Really has. How do people reach you? How do they connect with you if they want to do that?

Max Sinclair: So, um, we have a 20% off promo code at our website, which is ecomtent.ai/promo.

And that will give people 20% off if they wanna trial the tool. You can also book a demo, um, um, so yeah, like just head over to ecomtent.ai and, uh, you can, you can see everything we've talked about there, and hopefully like, depending on when this is released, you may see more of the stuff we've talked about already there.

Or maybe we'll still be on the lifestyle images, depending, but yeah.

Matt Edmundson: Fantastic guys so do check it out, ecomtent, the merging of e-commerce and content, uh, dot ai. Do look at that. Max, my last question for you, right. Uh, the question I've started to ask everybody, um, yeah. Imagine you're in a, in a room, hotel room.

Yep. You've just delivered your keynote on how generative AI is gonna transform everyone's life. Mm-hmm. , all the lawyers in the room are jumping up going, woo-hoo. We no longer have to do case law . Um, all the marketeers are going, woo-hoo. I no longer have to write product descriptions and do social media.

I'm curious, right. Um, whilst you are stood there and you're standing ovation, you get to thank people who have had a big impact on you. Yes. Um, family mentors, authors, software, podcasts, whatever. Who do you thank and why?

Max Sinclair: Um, God, I mean, like the first person that jumped out to me would be my cto because he has done, he's incredibly hardworking and he, um, you know, has built all, you know, he's built all of this stuff that we've talked about and all of these crazy ideas and, and made it possible.

So I think like he is, uh, the real brains behind the operation. My girlfriend's putting up with all of it. My parents, I don't know, I'd say

Matt Edmundson: I like that. Work, God and girlfriend. Yeah, yeah, yeah.

Max Sinclair: Congratulations to her. Um, so yeah. Oh.

Matt Edmundson: Fantastic. Yeah, yeah, yeah. No, absolutely. Well, it's f it's fantastic.

It's interesting, isn't it? You've, um, how, um, you talk about your CTO and how actually the technology behind this needs very clever people. Yes. And, um, I, I'm smiling in some respects cuz my son is doing, Uh, a master's degree in theoretical physics and is looking a lot into ai. So really fascinating to see where it all goes. Listen,

Max Sinclair: good for him. Good for him. I wish I'd done a, I wish I'd done that.

Matt Edmundson: Glad I didn't. Some of the stuff he talks about , I, me, me, people are much easier to understand. Uh, so listen, uh, Max, thanks for coming on the show, man. Great to connect with. Uh, great to have this conversation about ai. No doubt we'll have you on.

Yes. Uh, look forward to seeing the journey for ecomtent. Um, but, uh, it's been absolutely brilliant. Thank you.

Max Sinclair: Thank you, Matt. Thanks.

Matt Edmundson: Awesome. So another great conversation with Max. We will of course link to Max' info in the show notes, which you can get along for free, along with the transcript at ecommercepodcast.net.

Uh, or if you'll subscribe to the email newsletter it will be winging its way direct to your inbox. So don't forget to check out our free e-commerce training at ecommercecycles.com. Get into that methodology. Uh, let me know what you think. Really curious. We're just putting it out there and sharing it with the world for free, uh, cuz we know it works super well for our business.

And, uh, further. So I'd love to know your thoughts on the whole thing. Be sure to follow the e-commerce podcast wherever you get your podcast from because we've got some more great conversations lined up. And I don't want you to miss any of them, and in case no one has told you yet today, you are awesome. Ah, yes, you are. Created awesome.

It's just a burden you have to bear. It's true of Max, it's true of me. Even in a world with AI we're still awesome. Uh, the e-Commerce podcast is produced by Aurion Media. You can find our entire archive of episodes on your favorite podcast app. The team that makes this show possible is Sadaf Beynon, Josh Catchpole, Estella Robin and Tim Johnson.

Our theme song was written by Josh Edmundson and My Good Self. And as I mentioned, if you would like to read the transcript or show notes, our podcast, uh, website is ecommercepodcast.net. Do check that out. So that's it from me. That's it from Max. Thank you so much for joining us. Have a fantastic week. I'll see you next time. Bye for now.