The AI Marketer's Playbook
The AI Marketer's Playbook is an actionable podcast focusing on AI and marketing. Each episode covers AI strategies, tools, and trends that are changing marketing. Listen to interviews with industry experts, analyze case studies, and get practical tips. This podcast is for anyone looking to leverage AI in marketing to improve results.
Tune in to stay updated on AI marketing.
The AI Marketer's Playbook
22 | Revolutionizing Automation with AI Agents with Valeriya Pilkevich
In this episode if AI Marketer's playbook, Audrey Chia hosts Valeriya Pilkevich, a marketing analytics expert and CMO of mixNmatch. Discover how mixNmatch is redefining online shopping with AI-powered search and personalized outfit recommendations. Valeriya breaks down the AI workflows that monitor fashion micro-trends, enrich databases, and provide real-time styling insights. She also discusses the rise of autonomous AI agents and shares tips on leveraging AI for marketing productivity, ROI, and scalability. Whether you’re a startup, business owner, or marketer eager to integrate AI into your processes, this episode offers valuable advice on staying ahead in an AI-driven world.
Join my weekly Newsletter: https://lp.closewithcopy.co/welcome
Hello, and welcome back to the AI Marketers Playbook, where we cover actionable frameworks to help you leverage AI and marketing strategies in your business. I'm Audrey Chia, your host. And today I have with me a good friend, Valeria Pilkovic, a marketing analytics powerhouse, who is truly revolutionizing how businesses Implement AI solutions. Now as the CMO of Mix Match, she is transforming the online shopping experience through AI powered styling. And her track record includes delivering massive results for Fortune 500 companies with a huge return on investment. Now, Valeria brings with us a really practical perspective to AI implementation, and I'm super excited to talk all about AI with you, Valeria. Welcome to the show.
Valeriya Pilkevich:Thank you, Audrey. I'm super excited to be here as well and to chat with you.
Audrey Chia:Super cool. So for those of people who don't really know what you do, can you share more about, you know, how you got started, how you stepped into the world of AI and what are you up to right now?
Valeriya Pilkevich:Yeah, of course. So my background is actually, as you mentioned, in marketing analytics. So I would say it was always a consulting. So I was always the one person who was working on big data projects or big marketing data, let's say, projects and sales data. So I used to have Fortune 500 clients who had a A lot of data for marketing sales on and we used to do like marketing mix modeling for them, which means it's more like on the traditional AI, let's say machine learning side, you know, just getting all of the all of the insights, which we can get from the data. And then I was the one actually the consultant who was helping. To translate this insights from the data to the actionable, actionable business strategies and like how to improve the ROI, which channels to invest in, what to focus on, how to improve customer, acquisition strategies and so on. So this has been the background. So I was already exposed to AI, but rather this traditional side of AI, but I was always on the business side. And, yeah, basically the past two years after chat GPT came out surprise, surprise, I was digging deeper into it and, you know, just starting with prompts and with visuals because it was very hard to say, I think at the beginning I did not recognize the power of like chat GPT, you know, to, and what it can do, but I started with me journey rather with visuals because it's something that you can jump on quickly and easily. And it's, I think it's this well effect first. Because you see that you can, like, without being a designer, you can produce some nice visuals. And this, has hooked me. And then like gradually I started learning into prompts and how to use chat GPT, how to use different chatbots. And then gradually like going to the automation workflows and creating more customized AI solutions. So it's been very, let's say it's been always. It's been a great journey, but it's been always also very, it's, it's never boring. And then everything you discover along the way, it's always, well, you know, because you're like, oh, wow, I can automate the entire thing. And, so it's, yeah, it was very exciting. And recently, around half a year ago, maybe eight months, I also I'm now cm o at a startup mix and match. So it's basically AI powered search. you can imagine it this way, you can just chat to search, for example, you can say, I want a business schedule outfit for a networking event, or I want a rather chic outfit, with a touch of elegance. And, basically everything that you think of. And it works exactly. So it basically searches for the complete looks. So you can both, search in products or searching looks. So this is what we are working on. And we want to offer both a B2C platform. it's actually going live now. So I'm, I believe by the time this podcast is live, we are, we'll be already live and we offer it as well as a B2B integration. So also for every shop, it's a completely API for a slow, solution. So it's invisible basically for, if you go next time to your favorite, favorite shopping outlet or eShop and. You notice that it's AI powered search, you can chat with it, you know, just, just search for anything you want. it might be the Twitter behind it.
Audrey Chia:Firstly, congratulations. I am super excited for the launch. I think mix and match is something that a lot of people will need. Like even for myself, sometimes I have to go for a keynote and I'm like, What do I wear? And then you start looking for options, but you also don't really know what goes with what, right? and I think that is a very beautiful way of using AI to then, you know, tailor very personalized recommendations in a functional yet creative way. Right. and I would love to know for Mix and Match, where did the idea come about and how has the experience been?
Valeriya Pilkevich:Yeah, thank you. It's a great question. It's actually evolved. It's a microphone. She was thinking about, yeah, just as you mentioned, like, typical problem that everybody has, like, you know, having a huge amount of clothes in their wardrobe and thinking every morning, I have nothing to wear. You know, because you always have to make a mental effort. Okay, how do I combine it? Does it, does it fit together? Does it fit to the occasion? And then we constantly buy and more and more and more things without actually thinking, does it, this new thing suits to what I already have? Or maybe it doesn't because obviously not everybody's a stylist. not everybody of us learned how to combine this stuff, how to combine different clothings, but it's, you know, it's more than just clothes. It's, it's like, it's, It's a lasting impression that you make. It's also, yeah, it might be like the business, business contact that you make and he will remember you by your clothes. So it's kind of very important, both in professional setting, but also personal one, of course. but we also, well, so first we were thinking to go fully in B2C direction. So to the end consumers and offer it as an app where you actually can combine new things with what you already own. So. So there would be many ways to digitize your existing wardrobe. So exactly with this idea, not over consuming, but rather buying what you need. But then we noticed a huge, actually need in the market now for, from B2B sites or from the brand side to innovate because they're losing margins. The brands, there's also a lot of returns. many times consumers scroll like almost two hours, you know, pages and filters before they find something that they actually. And maybe by the time they found the size is not available. So it's the whole experience is completely not working. Let's say, and we talked to a few brands representative and we realized, Oh, that's, there is a huge demand. So we decided to go first into the actually talking to brands, to fashion brands, to e platforms, to be, to be, and then when we have a case and then going also to, to be, to see market.
Audrey Chia:It's super interesting that you are taking that approach, right? There is like a, a platform that I saw before in the makeup and cosmetic space. So they too have been trying to use AI, right, to help people to find that perfect, you know, blusher or that perfect lipstick for your skin tone. And I think it's very interesting to see how Brands are evolving and adapting and even if you are a offline brand or you know, traditionally an offline brand who is trying to have an online presence now with this new wave of AI as a brand, you need to figure out how do I leverage this, to my brand's advantage or miss out. Right. And one interesting point that you mentioned was how there's a lot of maybe waste, and also. Lost customers, who land on the site, but can't make up their mind. And then they just disappear forever. I think that is a very strong use case for AI, where you are personalizing the experience and delivering even more value to them. And I would love to know, like in that process of creating mix and match, right? How did you guys build the data sets for it? As much as you can share on your end.
Valeriya Pilkevich:Yeah. Yeah. it's, it's a good question. So, you would be surprised to know, but even if you ask, if you go to charge a PT now and upload your photo and ask it, Hey, can you give me a recommendation for my, okay. Maybe the front, but the prompt could be more elaborate. Can you give me a recommendation for what colors I should wear or maybe what forms suits me more? It will give you a pretty solid replies. So everybody's asking us, Oh, are you a stylist? Do you have this huge knowledge in your head? And we were like, well, we do have a data set, that we, for example, just part of it also by scraping like Instagram, for example, right. Or, or, all the available sources. But on the other hand, there is a lot that, junior AI models already know, and they could, they could advise you and, or if you upload there, I don't know. 20 images of 20 different clothes and ask it to more or less, tell the ones that would suit together. It was also come up with a pretty good reply. So this was, this is, yeah, this is actually, it was very surprising, but yeah. yeah. And as I mentioned, the workflow that I want to share with you today, it's also a part how we've been using, or less, yeah, this automation workflow to enrich our database as well. For example, at some point we thought, okay. We kind of trained AI to more or less recognize what suits together and, how to adjust it also to a personal profile. For example, if the person already has a sort of profile or the person went through, let's say a styling quiz, or they uploaded their own photo, so it's all integrated, but how do we make sure the looks are relevant? So what we provide, so it's relevant. And then we said, okay, let's, let's look at the. the most relevant looks are social media. And then we said, okay, let's, let's try to monitor the trends. And this is how we, for example, also like monitor a real life or real time, the trends from social media and integrate it as well in our database. So we know the looks are, you know, reflecting the latest trends as well. And, I was actually surprised because I was. It was me who built this workflow without any like being, having technical knowledge and, you know, whatsoever being, being able to code, but then currently it's possible for everybody, you know, and this is, I think we live in such a, in such a time where everybody could, could build a product, a SaaS, for example, without having, without ever having the, this deep technical Knowledge and bring it to the market fast.
Audrey Chia:Yeah. And it's amazing that a, you know, with AI, that you as a maybe non technical person now have new skills to create new workflows that previously would not have been very possible unless you had that knowledge and then be also like rapidly be able to create like your. Version one of whatever idea you have and then bring it to market and test it. Right. And you also mentioned like you have a specific workflow that you could share with us, that would be amazing to see a behind the scenes peek into how your mind operates.
Valeriya Pilkevich:Yeah. Yeah, of course. So this is more or less what I mentioned with you. it's, nothing. It's not super complicated, but I want to, add a few things, maybe how you're, also your listeners can modify it because it's, I think it's very powerful and there are many use cases. So, as I mentioned, what we do, we actually monitor, we try to monitor all the latest micro trends that are appearing, for example, on social media. And, for this, for example, I'm using Apify. I believe, some of your audience might have heard it about it as well. So basically, it's an external tool and by the way, what I'm currently sharing as a workflow in make, so it's an automation workflow done in make, which is a no code or low code automation platform. And AP fi is another external tool, which basically allows you to say, Hey, I'm interested in this Instagram channels, or I'm interested in this certain YouTube channels, or I'm interested in whatever TikTok, you know, or even certain websites or even Google Trends data. So they have huge amount of different, you know, like external websites they have access to. And initially what they do, they basically scrape the data. For example, for Instagram, they can scrape the visual posts. So this is what we're usually doing. Like the, the images for TikTok, you can even scrape the videos or even for YouTube. So both the descriptions, but also the videos. And, if you, if you think about it, there's. So many like millions of ways how you can use this data, right? Not just for example, the way we are using it, we're using it actually to enrich our database. And as I mentioned, just stay relevant with this micro trends, because later, of course, once there is a data, it goes through, also open AI module. And this is exactly, also probably like those automation, low code automation tools were there for, I don't know. 10 years or maybe more, but everybody just started talking about them now because again, this is AI powered automation is where you can, it's a completely new level, right? Because you can do so many things with it. So later then we go to, to GNI and there is a big prompt as well. It says, Hey, it's actually vision module. Right. So it analyzes the images and then it, again, it extracts all of those, what we see in the images, basically how people are combining clothes now, what colors are trendy, what prints maybe are trendy and so on and so on. And then it actually puts it into, or it sends it into our backend system where we have our, you know, like, database, with all the styles and where we have our, the whole architecture. And, if you think about it. There are so many ways. I mean, this is how we use it to improve the actual product, right? But even if we talk about, I'm sure many of your listeners are interested how to, you know, use AI for marketing purposes. And there, if you think about it, if you are able to also scrape those micro trends or scrape the whole content, the topics from YouTube, from TikTok, whatsoever, you can even. Have the unlimited database of great content examples. You can have unlimited database of the ideas, for example, for your personal content or for the content of your clients. So the topics, what are trending, what people are talking about now, and, you can use it in so many different ways because I believe the social media is there the place of, you know, of this, let's say. Data goldmine, which you can use for so many different purposes.
Audrey Chia:This is super powerful. Like even breaking down a workflow like that already streamlines so much information in a short amount of time that would otherwise have been so manual to dig through, you know, pages and pages of information.
Valeriya Pilkevich:Exactly. And just, just think about it. I don't know, like how, I mean, obviously there are so many businesses, just as our business and our startup, it, it was not possible just two years ago, three years ago to, to have this, this level, even when we talk about natural language search or visual search. Right. And, if you think about it, like how would you ever collect or know about all of the micro trends or all of the trends that are currently out there, you would never, I mean, you'd have to hire a team of, I don't know how many fashion analysts to actually monitor them every day, but you're, everybody's able to do it.
Audrey Chia:Wow, and it's just the beginning, right? This is like in the past two years, so many, you know, new solutions, new products, new platforms have popped up. I know beyond Mix Match, you also do AI consulting, right? So what are some other kinds of use cases besides the one you already shared? Like, what else can people think about and experiment with?
Valeriya Pilkevich:Yeah, thank you. It's a good question. So I'm doing mostly I consulting and training and usually companies that come to me are SME so small medium enterprises and typically they would come and saying, Hey, we tried to GPT a few times, but we are, we saw your LinkedIn content and we're interested to learn more about it. To see, to learn how we can leverage it more. And then what we would do, we would first do a lot of like, you know, onboarding and discovery sessions and, trying to understand what they're currently doing, for example. And, you know, those GNI use cases are usually like marketing and sales and maybe productivity use cases, just how to improve, how to streamline some processes. And then we'll try to understand, okay, what, what are the leverages? And then we typically start with low hanging fruits, like, for example, beating, building those custom GPTs or projects and plots where they already can, you know, see the benefits of it. Because even like, you know, for ICP analysis, right. Ideal customer persona, you can just build a bot or custom GPT, and then you can just always, you know, chat with it when it comes to creating again, marketing copy, for example, right. Because it has all the information that you are fitted in or. of course the streamline, streamlining proposal management or streamlining lead qualification. And there are so many use cases exactly around this marketing and sales and content creation. And, even general productivity, you know, just looking at typically, okay, what do I do repeatedly or what do I do every week or maybe every day, which is, you know, which I can simply automate. Put on autopilot and then just thinking, okay, what makes sense to use? Is it a custom GPT or custom AI tool? Can it be completely automated? So you don't even have to have to be to, to, you know, to, to write something to it, or might be even an agent, right? If it's a, if it's something more complex.
Audrey Chia:And I would love for us to dive a bit more into agents, right? So agents have been the new thing. I know you talk about it a lot. but for our listeners who don't know what on earth agents are, how are they different from, you know, custom GPTs or chat GPTs in general? Could you share a little more?
Valeriya Pilkevich:Yes, of course. so basically an agent is more autonomous than custom GPT. so custom GPT, if you think about it, or a project in cloud or any other AI tool, if you build, for example, relevance AI, they also offer those custom AI tools. And, they are, you always have to trigger it. So, right. You have to be the one who actually chats with it and starts the conversation. And then if there is, there will be a workflow. And if we talk about custom GPT, it usually has, let's say one context window or one, instruction window. And then you'll, it's a bit limited, right? It's, it's great way to start. We're ready with it. With Jenny, I mean, But it's a bit limited. when we talk about automation workflows, it's, they usually follow, a strict structure or a strict workflow. For example, of course you can say, Hey, if this happens, then you have to start, then the workflow starts. Then if this happens, then go to this module. If this happens, go to next module, but it's a very strict still. And even by using this open AI modules and clot and other GNI modules, of course, it makes it a bit more flexible, but still it follows the certain like journey, right? Which you have to predefine. When we talk about agent, it's a completely new level because again, they are you can build agents, for example, also with no code tools, like Relevance AI. That was, I was, I was doing. And of course, there are more complex agents which require code and a lot of big data sets of data. It's usually like used in the corporates. And so on, but I've been focusing also on rather this custom AI agents with, for example, relevance AI. So agencies, it's automated, so you don't have to trigger it. It usually has access to, whatever you need exactly. Maybe your CRM or maybe your calendar or your emails or, or something else like, you know, like lead qualification, steps, for example, for different sales purposes. So it has access to your tools. And, it can decide what is the next step. So again, in automation, you have to tell what the next step is in your workflow and how you want it to look like. Well, agent, you just give it all the information you provided with all the access, with access to different tools, for example, as well, And different capabilities. And then it decides depending on what the situation is, it decides what it should do next. And it's completely autonomous. Usually you can build it this way. So it's a, it's a completely new level. And it's something that me personally, I'm just starting to explore. But it's, yeah, I believe we'll hear a lot about agents this year.
Audrey Chia:Like for example, so from what I understand so far, right? So let's say you build a, custom GPT. It's like, tier one intern where you say, Hey, every day, please do this for me. I want this report. But if you have an agent, then it's almost like a senior in your team who then knows what to do and to just like, fill out with the work, right? What, what is an example of an agent or like how. Can agents interact with each other?
Valeriya Pilkevich:Oh, that's a great question. Myself personally, I haven't built teams of agents yet that interact with each other, but it's definitely even the tool that I mentioned, the relevancy, it's possible to do it there to build it. So basically it's, the way you think about it is you think how the. For example, department works. And here, I think it's important to mention it's, you know, the topic has been everywhere, I believe. Will AI replace the, I don't know, marketers or copywriters or sales people. And at this point it will not, because again, to, to build this agent or to build any workflow or any, Autonomous system, you need to, you need to have a deep understanding or deep expertise in this field. So for agents, the way you think about it is you think how the department works, for example, right? And then you think about, okay, who do I need in my, for example, marketing department? I need someone who does the research. I need someone who writes a copy. I need someone who does I don't know, something else. And then this is the way you start building your, like your ecosystem of agents. You start with one, right? The research. Then you think about, okay, the research agent, what should, should it be able to do? It might be able to research the news. It should be able to research maybe competitors. It should be able to research maybe potential customers. And then you build also certain tools or capabilities for this. specific agent. So I think it's a quite, like you have to think very strategically because you cannot just start with, you know, you cannot just write, Hey, I want to have a team of agents, you know, but you'll have to actually like, think about it also strategically. Like who, who do you need in your, in your team of agents? What every individual agents can be able to do, how will they interact with each other, right? What is the, the logic behind when exactly one agent, researcher agent should work and when exactly the, I don't know, the, the, the copywriter agent should work. So it's a, it's quite a complex system. So, and they are, again, I believe that expertise is important, or again, When the more medium more technical person again works with someone who is really expert in this field or You know to do to build something that makes sense at the end
Audrey Chia:I think that's a very interesting topic, like, even for me, right, to think about, being able to, because even as a copywriter, when I'm using AI, I'm almost like taking a step back from my own role and thinking, how do I think? And then, then I build the custom GPT based on the way I think, for example. So that in itself is me taking a step back from my own role. Now, to build a team, you require every individual agent to have that in depth knowledge, workflows, and expertise. And probably you have to figure out how would I react with, for example, the performance marketer, and then figure out what is the best way to interact. And in a human interaction, for example, a copywriter and a performance marketer, there is a lot of exchange of data, insights, knowledge, back and forth, bandering, and figuring out. What is the best way forward but I do think it is really interesting I wonder on your end valeria since you have worked with so many business owners and you know clients before Has there been any key, you know Hesitations by the clients that you've worked with or people that you know about? AI replacing jobs or taking over departments?
Valeriya Pilkevich:yeah, I think there are many, I think there are many discussions like this out there. And, I must say, for example, the clients that usually come to me or, you know, that come from, for example, following me on LinkedIn, they are rather eager to learn and they would be like, Hey, we haven't done much, but we are so excited to learn and to see how we can improve our processes. And for many of them, what I hear is the phrase, like, how can we scale our business without hiring additional personnel, which is a bit scary. I mean, I would not say it's scary. It's something that business has to do to stay profitable. But I would say, this is something that businesses are currently looking at. So I would say for Professionals or specialists or experts in, you know, in every area. But if we talk about marketing specific, like they will not disappear, but the roles would slightly shift or change. So I would say everybody has to upskill themselves. Right. And it's a, it's exactly the topic of, Exactly. We are coming back to the topic of who can create the custom GPT for performance marketing, or who can create the team of agents for marketing. It's only the person who has a good understanding about marketing. Processes or a team of people, only they can create this agent and see if the outputs are correct and make sense or not. so I would say it's, it's a both part. So on one hand, businesses really, they want to do it. on the other hand, I see, yeah, many. Are saying, yeah, we kind of want to scale without having this additional personality, more people in our team. that's why we want to use AI and automation. So I think there will be changes and we will see them gradually in the upcoming years. but I think there is an opportunity for, for all of us.
Audrey Chia:Definitely. I think one thing that stood out to me during the kind of conversations that I had is there will be many jobs that will become redundant, but in the same process, there will be many new jobs that will be created. But then during that time, there will be a gap. So it's, how do you, you know, replace the redundant jobs with all redundant skill sets, with new skill sets that then enable you to keep up with the changing world and keep up with the changing needs of businesses. And I think that is really important for everyone. And I know it's already 2025 and Birack AI has been here for quite some time, but I do still feel like, Not everyone is on board yet, and there are still quite a lot of people who are, on the fence or waiting, or just, you know, going about their daily lives without fully leveraging AI or welcoming it into their lives. So to that, my point is, Perspective would be, it's better to start learning how to leverage it now than later. Especially when businesses right now, I think this is the year where businesses are probably going to be investing more time, resources in actually trying to integrate AI into their business. Right.
Valeriya Pilkevich:Yes, absolutely. They will. And, it's a huge topic now as well. This knowledge gap, or I don't know, AI knowledge gap, I believe it's the right term. it's exactly the case what we see, for example, between, bigger comparations and smaller businesses, as you can imagine, smaller businesses are more agile and they don't have that many. process in place yet, or they don't have processes, but it's easy to integrate a new tool, for example, right? They will not have a huge, like, let's say a big, approval. I don't know, journey to approving your tool, whereas the bigger corporation has it. And I still, I still see I'm based in Germany, but in Germany, many also bigger corporations, they, Many of them haven't introduced anything yet. Some of them just introduced, like, Microsoft Copilot, which is also not the, the, you know, AI, which, and everything you can do with it. It's, of course, it already simplifies many things. In this Microsoft ecosystem. So there is already a kind of AI knowledge gap between, for example, people who are like solo partners and business owners and smaller businesses, because they know to survive, I have to use it and I have to optimize my processes and I don't have a team of a hundred people or a thousand people or a hundred thousand people. So we need to do it fast. whereas bigger corporations or the employees there, they're not quite, They are yet with knowledge on one hand. On the other hand, then there is also the risks because those employees who actually want to be on the, on the same level that would, they would then go to, I don't know, chat GPT and upload maybe some company data or client data, which is also poses risks to, again, to the organizations. So I believe there is a huge need also for AI training now for the proper training also in AI ethics and. Literacy and how to use these tools and what not to use. Like, also maybe the training for everybody, like the, do not trust this outputs per default, you know, because people, we tend to rather trust if we, if we read something and, there will have to be also more like having critical mindset, I believe. And this is something we'll have to learn as well. Yeah,
Audrey Chia:I think that's something that I have observed, especially from chat GPT, although I love it, it still loves to hallucinate on me and give me information that is not within the parameters. And the funny thing is, it can, for example, if I'm asking it to analyze something, it can do great for the first five paragraphs. Suddenly you introduce something new in the sixth, and you wouldn't know if you read the first five and you assume that, hey, you know, chat GPT is. Doing great at this analysis and unless you read the sixth paragraph and you really figure out like I don't I don't think this was mentioned anywhere in the document Yes, I highly recommend people to really double check the work, right?
Valeriya Pilkevich:Yes, exactly. Absolutely. And AI knowledge gap, it's also the gap for example, there are Also some population, you know, demographic, population parts, like also elderly, for example. I'm sure there are many who are also super more fluent in all the workflows in the AI powered systems than myself, but there are also a big majority who do not, and they cannot keep up with it. And then, what I'm also sometimes thinking about, yeah, Like if at some point companies start doing like layoffs, they will start probably with this, rather, all the personnel, but they will not be able to upscale or find any new jobs. So I think there are many, many question marks. I think there is a lot of positive things. I think also there are many challenges that are upcoming, but as you mentioned, businesses and individuals have to start today also to. Educating themselves to upskill in themselves, not to wait until the company training, but rather starting to look at the different trainings. But of course, obeying the guidelines as well, not, not to, yeah, not to put something in risk.
Audrey Chia:Yeah, absolutely. Right. And like the balance of risk, but also experimentation, you know, curiosity versus like knowing the boundaries, right? These are things that are much harder to navigate in this world of AI where everyone is still figuring it all out. And I do know, for example, some companies, they are limiting their employees from even leveraging AI because the companies themselves are still figuring it out. So they are holding onto systems and figuring out how do I best integrate it into my company? And because they are not taking the big decisions at this point in time, the employees are also not incentivized, motivated, or given the resources to progress as fast as perhaps anyone else should. But I think as an individual, you shouldn't let that stop you from your learning. There are many avenues to learn, to grow. And there are so many opportunities for you to skill, upskill yourself to prepare for this upcoming and actually current, you know, wave of change. But let's also talk about ROI. So, from the companies you have worked with, whether at mix and match or whether it's with your clients, what kind of ROI have they seen? Like how, how has AI actually helped them in their own processes?
Valeriya Pilkevich:Yeah, it's a great question. And businesses love hearing ROI, right? so from my experience working for bigger companies, and as I mentioned, it was more, rather, it was less generative AI, it was rather, like machine learning, traditional AI and statistics. there we've seen by using those, for example, By analyzing the client's data and being able to predict, you know, the sales with certain marketing investment and media investment, or being able to, you know, making an investment shift store to more profitable channels. We've seen ROI increased by 20%, you know, in terms of return on marketing investment within a year. which is great. while. Like now with generative AI and AI powered automation, the ROI increase is also huge, right? If only if we, only if we talk about the productivity increase and time savings, I think for now it's the most, yeah. The most low hanging fruits, let's say, because everybody could implement those. Maybe not every business can build and a sales agents, which can qualify leads. Although many do of course, but when it comes to productivity, you can easily save, I don't know, 20 hours a week on automating all of those routine tasks that just take, take the energy. And this is basically across. across different industries. It's not an industry specific, because we all have those routine tasks that just, you know, take our time without actually them being very intellectual.
Audrey Chia:Definitely. And I think the great thing about AI is if you're able to optimize your more routine or mundane tasks, right, then it actually frees you up for more strategic thinking, more creative thinking, and even more time for learning, right? Because Then you have the time savings. So in the past, I used to take, at least three weeks to build a landing page, now we can shrink the entire process to three days with the same quality of work, with the same output. But because you're using AI to, you know, streamline so many of the research processes, the copywriting process that actually Usually takes a lot more time than that. and I think it's a beautiful way when you can combine both the human and AI part. So it's never AI by itself, but you know, that human part is as important. Do you feel like that is also important in the work that you have been doing?
Valeriya Pilkevich:Yes, absolutely. I wanted to also address what you mentioned with the landing page. I believe it might take you, or it, it comes to every, basically every process. It might take you a bit more time at the beginning because you'll have to figure it out. And as you mentioned, as a copywriter, first, you'll have to think about, okay, how do I approach it? What do I need exactly? Do I need an ICT analysis? For example, for instance, what all of the context that I would need. But once you feel this workflow, maybe the second landing page and the third and the 10th there will be like much, much faster. And I think it goes with everything, right? When you build those AI powered workflows or automation workflows. It's there is a struggle first, because you'll have to figure it out first, then you have to build something first, whether it's a custom GPT or, or 10 custom GPTs, or whether it's an actual automation, you'll have to first put some hours there, put some actual intellectual work there, but then Later, once it's done, it's, yeah, it just takes you, you know, minutes instead of, instead of hours and hours instead of weeks. so this is the beauty of it. And, to your question about the human involvement, it's absolutely very important. For example, at Mix& Match, we always, I also use a lot of, you know, AI part workflows and, custom AI tools for marketing purposes. Also, whether it's a content creation on LinkedIn, I don't know, newsletter, B2B newsletter creation, collecting some topics or, you know, even the updating pitch text, which you have to do every time because there's every time different, you know, Different briefs or different rules, how much you should present, how long, what exactly should be included. So there is a constant like copy flow that you have to keep updating. And, absolutely, as you mentioned, we will never do something without actual human check or human in the loop. So, like, even we have a workflow for B2B newsletter, for fashion news, where we would, you know, synthesize fashion news. happens automatically once there are, there is some news in different, also different media outlets or fashion. It would, you know, segment them by relevance and then it would even pre write the copy. And so basically you could even put it out there without any pretext. But we always do it. And there are things, as you mentioned, there are things where ChagPT would hallucinate or perplexity would hallucinate, and then you would suddenly see something very strange there. So I believe it's it's simplifying it, it's It's giving back a lot of hours in the week and especially for business owners or startup owners who need them. But the quality is still very important or I would say the most important when it comes to actually exposing your, your work, your copy, your marketing efforts to potential clients. So there is always should be human in the loop. I believe for now, if we talk about some background processes, okay. You can automate those, but if it comes to something client facing, there must be human in the loop. You, you cannot go no other way.
Audrey Chia:I 100 percent agree, and I highly recommend all you listeners to always remember to add that human back into the loop for the best results and, to wrap things up, Valeria, what is next for you or for Mix Match or your, you know, your AI consulting work, where do you see yourself going?
Valeriya Pilkevich:Thank you. For Mix Match, we are, we are launching, as I mentioned, and we're super excited to get there. Users feedback and to learn even to it's it's a huge experiment for us to see how they're using the platform, right? Whether they're actually looking for looks or they're just looking for products because it's also a change in the behavior so actually want to shape this behavior and It's very exciting. Also We're looking for b2b piloting partners now So if there is any who are or listening us and are interested to to join this AI powered fashion discovery Maybe then it's, absolutely a huge, huge step for us and something we're going to do is next for me personally, on my consulting projects, I will be, I guess, building agents this year.
Audrey Chia:Super exciting. and I will be here to support your journey as well as look for you for the latest agent update. So that is something I'm going to be looking forward to. And where can our listeners find you if they want to reach out?
Valeriya Pilkevich:Yeah. on my LinkedIn it's the same Vale, which, just send me the, and or connect with me and tell that you know Audrey
Audrey Chia:Awesome. Thank you so much. I'll sure accept everybody. thank you so much Vale, for joining us. It was a pleasure having you for the show. And thank you guys for tuning in. Don't forget to subscribe to the AI market display and hit the bell for more actionable marketing insights. You'll see you next week.