EP #340] [ENG] - Power of AI to level up your content strategy - Monte Desai

EP #340] [ENG] - Power of AI to level up your content strategy - Monte Desai Speaker 1: Welcome to The Ecommerce Lab By Ecomcy. This is the place for everything related to Amazon private label and e-commerce. Learn exactly what you need to start or scale your business. Get insights from the top industry experts who will discuss the latest trends and best practices in the world of Amazon. From choosing products and sourcing from a supplier to setting up your Amazon account and marketing your business, you will hear it here. Speaker 2: Let's get started. Speaker 1: Here is your host, Vincenzo Toscano. Speaker 2: Hello guys, welcome to another episode of The Ecommerce Lab By Ecomcy, the place where I relate to Amazon Private Level and Ecommerce. My name is Vincenzo Toscano, Founder and CEO of Ecomcy and today we bring another special guest. His name is Monte Desai, Founder and CEO of Pixii.ai where they specialize on everything that has to do with helping you generate the right content for your brands on Amazon by using AI and that's why I wanted to bring Monte today because that's been something that's been changing very fast in the space. AI is pretty much in everybody's mouth, like what is happening, what are the right tools, how to implement it. And today we're going to be talking about, you know, how solutions such as FIXE is going to allow you to do that much easier and why you shouldn't, you know, be missing out. So Monte, welcome to the show. How are you doing? Speaker 1: Hey, Vincenzo. Thanks for having me. Doing fantastic. Speaker 2: Thank you, man. Pleasure to have you here. Thank you so much for your time. I know we've been working for a while now, so it's great to finally, you know, have you here. So I guess for those that don't know you, maybe let's start with a brief introduction about yourself, what you guys do at Pixie, and then let's start with the topic. Speaker 1: For sure. So my name is Monte Desai, CEO and founder of Pixie.ai. Pixie generates visuals that sell on Amazon. And so we serve Amazon agencies and aggregators. The flow is you drop in a photo, get a bunch of visuals, infographics, main image, lifestyle shots. And unlike any other tool, you know, our designs are fully editable. You can make any change that you wish. And then, you know, we try to aim for a zero prompt experience. That's our goal. Speaker 2: Awesome. Speaker 1: And then in terms of my background, you know, I can just mention that really quickly if that's helpful. I spent five years at Bain & Company. So I was a consultant working with Fortune 500 CEOs on marketing topics and ecommerce topics. And, you know, so that's really my background is more of like trying to bring big brand expertise to any agency and any seller on Amazon. Speaker 2: Cool. Awesome. I appreciate that intro. Thank you for that. So I guess let's dive into the topic, which is, you know, generating images for your listings on Amazon using AI, I guess. The first question that I get every single day when it comes to that is like, you know, why I should be using AI, right? And maybe, you know, keep going the old way of, you know, being very detailed with my own graphic design as my using my own kind of photographer, like, what is the first kind of quick win you get straight away by using a tool such as your photo? Speaker 1: I, you know, when we show designers pixie, and they start using it they're like wow. Right, like, our, our main users are actually designers. And it's like their little secret. Speaker 2: They actually say it's not done by them by was it. Speaker 1: Well, because I mean, it's just so much faster, right? You drop in one photo, you get 20 concepts really, really quickly. You know, a lot of them are actually ready to use on Amazon now. We say that, you know, so let's say in two minutes, you can get 20 visuals compliant with Amazon main images, lifestyle shots, infographics. People are seeing 3x the click through rate using our main images. You know, on average, baseline is like 30% growth and click-through rate and conversion rate, but sometimes people are saying 3, 4, 5x. And so, yeah, why would you be stuck in the past? I mean, it's just like a new world has come. It's like, you know, it's like hand drawing when you can use Photoshop or now the new version is using Photoshop and you can use Pixie. And so, you know, I don't think, like to be clear, like I don't think There's always a role for, you know, Figma and Photoshop and all these tools, but I think you're just in a new world now where AI can do it for you and it can do it much faster. And so, yeah, the expectations are also rising. I would say this too, like now agencies are coming back to us and they're like, like AI is a curse. Some of the clients know now. And so now they're like, you just do it today, you know, instead of you used to get like a day or a week or a couple of weeks, but now it's like, can you just do it now? Like, can you just do 20? Speaker 2: I know, I think that's also, that's I guess one of the negatives of this, that now expectations of people sometimes like crazy, like I know now because AI is a thing, like you should be able to do things within one hour. It's like, yeah, it's true that you can do things faster than ever, but at the same time, there's still the nature of the time will tell me to. I find a product, I create a prom, understanding this is aligned with the brand guidelines and all of that. So I guess that's also creating some mismatch right now. Speaker 1: Well, I mean, it's going to be a really interesting, you know, next few years, right? Because, because exactly what you say, it's like, wait, is that going to save us time? Or is it just going to make us do more work? Because now we can, yeah, instead of turning out like three clients in a day, we can do 10. And like, wait, whoa, you know, a 30. Yeah, it's going to be a very interesting time for the next five years for sure. Speaker 2: Yeah, that's actually a good point. Like I was discussing this with a person the other day, which is we're going to see companies that maybe have only 10 employees, but each employee has like 100 accounts. It's like at the same time, it can be used to some extent to, you know, create more exploitation because now you have the excuse, ah, now within eight hours you should be able to do X because you have 50 agents under you. But yeah, at the same time, it's the bandwidth that we as human beings have in terms of process so much, so many tasks in a given day, you know? Speaker 1: I mean, I'll give you one example from my old world, like, you know, at this consulting firm, we would hire a ton of these like MBAs from all these fancy schools like Harvard, MIT, whatever. And, you know, there's tasks now that I would I would be like, hey, like as a manager, I'd be like, hey, go research this market and figure out who the big players are and like who's going to win and what would it shine as market share. And that would take like a day, right? And they're very well trained and experienced and all that, really smart. It would take them about a day. And now I wouldn't even ask a person, I would just go to ChatGPT and I would get the answer in about five minutes with like 20 citations and it would make me a presentation. Like it's A whole different level. And for junior people, I think for them, they have the biggest adaptation to make, right? Because the most low-level task that we would use to train junior people, that's disappearing now, right? So a lot of agencies are using Pixie for the first draft, right? Immediately, they get a client brief, they immediately get 20 concepts and they show it to the client. And sometimes you would add a junior person to that. And so what is their role now? Speaker 2: Yeah, very, very, very, very tough. You know, like, this is something that I think brings me to my next question, which is, you know, with everything you guys are doing at Pixie with You know, the technology of allowing now founders to be able, or even agencies to be able to do the images of the listings, you know, within a very short period of time compared to what it used to take. Like, at the end of the day, what do you think this is going to lead us in terms of differentiator, right? Because if everybody ends up using AI, I think this is actually another thing that is kind of interesting, kind of, maybe we can call it paradox. It's like if everybody ends up using AI, everybody has access to the best images. Why do you see this space going? Everybody's going to have 10 out of 10 images and then who actually wins? I mean, taste. Speaker 1: Machines don't have taste the way humans do. And the thing about humans and taste is that it's always changing. People say the algorithms are always changing and they are, but our preferences are changing. Now we want loose genes. And 10 years ago, we wanted skinny jeans. All right. And so that's and that's happening on a super fast cycle. So taste is always going to be there. But, you know, I think this idea of everybody using AI, like. Speaker 2: I think we think we think that everybody's using, but actually it's very little people. Speaker 1: Yes, for you and me, probably. Right. But we're going like I go to conferences and they'll be like. You know, how do you use AI? Like, you know, an AI masterclass, right? I'll go and I'll be like, all right, I'm so excited to learn something. And they're like, use ChatGPT. I'm like, whoa, come on. Or like use Trust and GPTs. And I mean, we're so advanced now. Like we're we're doing Creative Breeze. We're doing entire designs, entire listings. Like I'm doing a lot of my personal life, finances, all kinds of investments, all kinds of stuff using AI now. And so I think for you and me, we're just in this world where we think everybody's using it and it's not really the case. You still have time, by the way. If you're not using AI now and you want to differentiate, there's a lot of opportunity. It's still early. Speaker 2: Yeah, yeah. Now, when it comes to your technology, I want to bring this to the table. When it comes to AI, there are so many algorithms out there, right? And many ways that you can essentially engage with AI to come up with images. And I think something that I'm sure people are going to find very interesting to learn out of this episode is like, you know, what are some of the things that make Pixie unique when it comes to that? Like, what are some of the things that you would say, you know, we're very good at this or that? Because I think some of the feedback I get A lot of times from people when it comes to AI, they might say, oh, when I do AI, people come with seven fingers. Or the text might be overlaid here with grammatical mistakes and things like that. So what is your feedback on all of that? Speaker 1: Yeah, so I would say three things that make a stand out. So one is you drop in a photo and you get 20 and they're beautiful and they're tailored to you and they work and you can show them to clients and you don't have to write new props, right? So that's one zero problem. The other is that it's editable, right? And so our designs and images, you can take full control in a canvas editor and any small thing that the client wants, If you want, you can just change that one thing. And the last thing is, you know, it's optimized for Amazon, right? We also do work on other platforms as well. But, you know, Amazon has a lot of rules. And then beyond the rules, there's a lot of tricks to try to, you know, stand out and convince people to buy. And so, you know, we scraped 100,000 Amazon listings to try to figure out what works. And we've baked that into the product. We've baked that into the templates. Because ultimately, I think, like, From a technical perspective, what we found is that people just don't care what tool, what model, what technology. They just want it to work, and they want it to work fast, and that's what we do. Speaker 2: Cool. Now, when it comes to this, how are we essentially navigating the whole thing of the nature of the products when it comes to that? Because also something I've been hearing, and I want to hear your feedback on this, is for certain products Depending on the complexity, AI is still better at some products than others, depending on, you know, sometimes the lifestyle it has to show or the environments and things like that. Do you see some categories where you see AI getting very, very accurate when it comes to images? And maybe sometimes a product that you see, okay, it's taking a bit longer because of the type of product, the elements that the image needs that where AI is still struggling. Is there something there or do you think it's become pretty baseline across the board? Speaker 1: You know, so anything that I say now is going to be out of date in about a month because the models are moving so quickly. But today, CPG crushes like anything that's like a can or a bottle or, you know, food, beauty, health, supplements. Those categories, AI has figured out because they're all about the same size and there's like a million images of them. And so those are good. I think the categories were, and those are a lot of volume, by the way. So most agencies that we work with, I mean, every one of those CPG brands needs like a seasonal refresh or a new flavor or, you know. The stuff that's like still a little bit challenging, I think is You know, home goods, sporting goods, because there's just a big variety of what you could have. So furniture does well because, you know, AI understands what a couch is and what a lamp is. But if you have like a really odd shaped rack or, you know, we had someone who was doing, I think someone on your team was actually doing like a mandala shaped I don't even know what to call it. And so I can't even describe it. I don't know what I was going to do. So yeah, CPG crushes it. Clothing really, really crushes it. And then things beyond that, it starts to get kind of choppy. And the issues that you'll see is If you look at the size wrong, or that's the main one probably, look at the size wrong, or it might distort the product a little bit. But I think, you know, if you give it a few months, I think that'll also be gone. Speaker 2: Cool, cool. Now when it comes to AI implementation as well, something that's been a very hot topic, I would say, is how we also make sure these are optimized for Rufus, right? In terms of how Rufus is analyzing images, in terms of the copy they put on the images, the lifestyle, the elements, How are we actually reverse engineering that? If so, are tools such as Pixia also taken into consideration? And how do you think that's going to play a key role moving forward when it comes to optimizing images as well? Speaker 1: So yeah, totally. We do optimize for Rufus. You know, I think the whole conversation about getting AI to recommend your product and all that, I mean, if you write for humans, that works pretty well for AI. Amazon, Google, every major platform has moved a long time away from keywords. It used to be, look for the word grapefruit and it's just like a find and replace or a control F for grapefruit. And that's not been the case for a long time. Semantic search, which is like searching the way humans search and reading the way humans read, that's been happening for a while. And so Rufus does that. So as long as you write your title and your bullets in a way that's You know, digestible, clear, obvious to any human, then I think it works pretty well. You know, the images, Roof is going to change a lot in the next few years. Like at this point, they went back and forth on whether or not they're using alt text. And so if they say that, yes, we are actually going to go ahead and use alt text, we'll continue to support alt text. But it's unclear, you know, It's unclear where that's going to go, right? There's a world right now where these bots are basically just reading the text. Are they going to read images eventually as well? Because in that case, you want to have a lot of images that tell your story very, very well. Yeah, so a lot of change, but I think the TLDR, in my opinion, is if you're writing good copy for humans and making good visuals for humans, that should do the job for Rufus. There's like, why would they introduce new tricks, right? You have to think about this from their point of view. Like, why would they make it like a little secret? Speaker 2: Yeah, I agree, I agree. So I guess when it comes to this is creating all these images and this is something that I'm very keen to also learn from you is like, given that they are going to be doing all the editing, heavy lifting, integration elements and so forth, that means you also need to be very good at giving the direction. Like this is my persona, this is what I want the image to express, these are my brand guidelines. And I feel most of the time people don't have even that defined in the first place, right? And if we don't have that defined, AI to some extent is going to create a lot of assumptions, which might not actually convey the right message. So how you see also playing a role here when it comes to... Speaker 1: That's such a key point. I mean, yeah, it's... If you don't know who your customer is and why they're buying your product and how your product stands out from the competition, it'll be really, really hard to design visuals that really stand out. AI can make some inferences. It can say, oh, yeah, let's show off these features. Let's make things that are beautiful. But beautiful is not the same thing as selling, right? Yeah, I think that that's definitely a place that you mentioned, like how do agencies and brands stand out? Having a good understanding of the customer, I think, is a really, really good place. In terms of our process, what we're seeing a lot of people do is they'll go and write a creative brief by doing things like customer interviews, by looking at reviews and competitors, and then using and dropping that creative brief into Pixie, and then Pixie does all the designing. But I think you're right. Focus on the customer, right? It always comes back to the customer. Speaker 2: Yeah, if you don't have the right, that's the thing. If you don't have the right persona like this, you might end up, you know, even when it comes back to the Rufus conversation, you need to really understand the age that you're targeting, the gender, the lifestyle. Otherwise, you end up doing an image that is Amazing when it comes to looks, but you actually not convey the message that you want to convey to the image, which is going to have a direct impact on your conversions. Unknown Speaker: Yeah. Speaker 1: Who is your customer? What do they use it for? Right. And so a big thing with Rufus and ChatGPT and others is like use cases where it's like, you know, I need shoes for trail running. And, you know, ideally, the thing is also Rufus doesn't have at this point, it's not really using your history that much. So for example, if it would know that, Hey, my, I prefer wide shoes. It might even say something about that. Like, Hey, this shoe runs wide. You don't need to buy the wide size because the shoe already runs wide. Um, and so they haven't even integrated with all the other Amazon features that, that do things like that. So, um, I mean, I think that's where it's going to go. Like true contextual learning where, you know, Amazon knows more about the customer and hopefully they expose that more to you, but it's TBD, you know, it's TBD over the next few years. Speaker 2: Yeah. Now, tell me how would I, especially when it comes to split testing, You think that's also a bit changing the game because the reason why I want to bring this topic is like talking about conversion, something that, you know, we recommend and we do internally as well is split testing the main image, secondary images, A plus and all of that. And that's something in the past was so expensive to do that you were doing most of the time once or twice and that's it. But now, like, you know, you can do pretty much on a monthly basis or even more because you have AI solution like yours that can allow you to do within Within hours. So are you seeing also people adapting that a lot in their SAPs? Like how are you seeing also people using AI when it comes to split testing? Speaker 1: Yeah, I mean there's definitely, that's definitely one of the big use case for sure. One, so I can talk about some of the use cases if that's helpful from the agency side and the aggregator side. So one is A-B testing. Generate 20 main images, you know, get some internal feedback, get some client feedback, and then put them Into like PickFu or IntelliV or some of these tools that help you figure out. You know, what people like. And then just run tests on Amazon itself, right? And then see the winner there. And so, yeah, just as you mentioned, instead of doing it once a year, now you're like, well, why not just do it once a month? You know, why not just keep click-through rate high? That's one use case. Another use case is actually more about speed. So we have some, you know, some folks who are just basically like, hey, let's just rip out a quick listing. Really, really good. We know exactly what we want. We've done a bunch of research. We've thought about the creative brief. You know, we need to make 100 listings. So let's drop them out really quickly. So it's about speed. And then for others, you know, I think it's more about like client retention, and so and outreach, right? And so you're like, send a free image. As part of your outreach or on the first client call when you're pitching them, you could be like, hey, here's some ideas that we have, right? So you can impress them really quickly with things that are beautiful. And you have a lot of freedom in that time to be rough because you can go, hey, look, we just did this really quickly. We don't know much about you and what your goals are, but here's some designs. So those are the three use cases that we're seeing. Speaker 2: Cool, cool. And I would say to start bringing this whole conversation to a closure in terms of maybe we can call it like either predictions or tips or mistakes to avoid, like what are some of the things that you feel like It's something that people should take out of this conversation in terms of AI and when it comes to image implementation, like, is there some things that you feel are coming that are like in the shadows, especially you that are so deep into the communities and technology that you say, oh, I see people are doing this and most people don't know what is coming, or mistakes that people keep doing, or is there something that you'd like to share to close to this episode? Speaker 1: Yeah, I mean, You have an opportunity to stand out by using AI. You can ship designs faster. You can ship designs that make more money. You can ship designs that are more beautiful. I'm talking like 20 times faster, right? Like minutes, you can have amazing designs. And you can still stand out now, right? There's a lot of brands and agencies that are using AI and they are billing out beautiful listings, but the vast majority are not. And so if you hop on now, it's like 20x the speed. Wait, what? Like, let's do it. You know, social media and, you know, Like the world of conferences and stuff like that, you get the feeling that, oh, like people talk about AI, but it's not quite there yet. Like it's there. It's 100% there. People are making money on it. They've reinvented the whole process around it. And so, yeah, now is the time. Don't delay. Speaker 2: Cool, man. I appreciate that. Thank you so much for, you know, all your advice and all your insights. And I'm sure, you know, people might want to, you know, explore more about the tool and everything you guys do. So tell me how people can find you. Speaker 1: Pixie.ai, check us out. You can book time with me or one of our sales reps and just get a sense of what the platform looks like. I think I should mention one thing that we're going to do that I think a lot of folks are interested in is create one master listing and then scale it to a thousand listings. So that's one big thing we're working on now. You know the process, right? You make the listing, you get the client's approval, and you're like, all right, well, we have this, we have a thousand other products to do this for, we need before and after. Speaker 2: We have the blueprint, how we copy paste it to the whole catalog essentially, yeah. Speaker 1: But then make it work, right? Like, you know, if you have a curly hair shampoo that's your master listing, but you have to go for the blonde shampoo and the conditioner and all that stuff, you have to show different images and different visuals. That's what we're working on now, so that'll be out September, so October, yeah, it'll be out. Yeah, I mean, it's an exciting time. I mean, I'm glad to have the conversation with you because it's like, so much has changed so quickly and probably everything that we talked about is going to be out of date next month. Speaker 2: I know, I'm very excited as well. I think what is coming for the remainder of the year next year is super, super, you know, It's going to be mind-blowing for a lot of people. I think most people don't know what is coming and how fast this is moving. And that's why putting a podcast like this out there is key so people can get a glimpse of what solutions are there, what you should be aware of, what things to avoid. And hopefully, that's going to make the whole selling journey much easier. But it's been a pleasure to have you here, Monte. Thank you so much for your time. Looking forward to a partnership. It's been a pleasure. Speaker 1: Thanks, Vincenzo. Have a great day. We'll talk soon. Thanks for listening to the e-commerce lab by ecomC. Be sure to subscribe so you don't miss an episode. While you are at it, we would appreciate it if you could leave an honest rating and review on Apple Podcasts, Spotify, or wherever you listen. That will make it easier for others to find out about the show and benefit from it. Want more? Visit our website at www.ecomc.com where you can get your first consultation for free. Or find us on Instagram, Facebook, and LinkedIn at ecomc.