Our CLO – Beata Mosór, was interviewed by Bartosz Ziółko on AI in marketing topic. The interview discusses the potential impact of AI on the advertising market, specifically questioning whether it will disappear. Beata Mosór, a marketing professional with experience in AI and technology, argues that AI will not eliminate the advertising market but will change it fundamentally.
Here’s a breakdown of the key points:
- AI is changing, not eliminating: Mosór emphasizes that AI is transforming the way advertising works, rather than making it obsolete. She highlights how AI can personalize ad targeting, optimize campaigns, and automate tasks previously handled by humans. This shift will require advertisers and agencies to adapt and learn new skills.
- AI-powered personalization: A significant theme is AI’s ability to personalize ad experiences. Mosór notes that AI can tailor ads to individual users based on their past behavior and preferences, leading to more effective and relevant campaigns.
- Automation and efficiency: AI’s potential to automate tasks in advertising is also discussed. This includes content creation, campaign management, and data analysis, which could potentially reduce costs and increase efficiency.
- Challenges for smaller players: The discussion touches on the potential challenges for smaller advertising agencies and businesses. AI-powered tools and services, developed by large companies, could make it difficult for smaller players to compete. Mosór suggests that smaller companies may need to focus on niche markets or leverage AI in creative ways to maintain their relevance.
- The role of human creativity: While AI can automate aspects of advertising, Mosór highlights the ongoing importance of human creativity and strategy. She emphasizes that AI can enhance human skills and not entirely replace them.
- Adapting to the changing landscape: A crucial takeaway from the interview is the necessity of adapting to the evolving advertising landscape. Advertisers and agencies that embrace AI and learn to utilize its capabilities will likely be best positioned for success.
In summary, the interview suggests that AI will not cause the advertising market to disappear, but will fundamentally change it. Advertisers must adapt their strategies, embrace AI, and understand its capabilities to thrive in this new environment.
Here you can watch the video >>
Transcript:
Bartosz Ziółko: Hello everyone! Today I’ll be speaking with Beata Mosór, whom I’ll introduce shortly. But briefly, we’ll be discussing the intersection of AI and marketing. Now, let’s move on to your resume. I see about 30 entries, but perhaps we can skip a few. I see you started at nazwa.pl.
Beata Mosór: Yes, that’s correct. I worked there for four years, and it was my first major company, I began as a marketing specialist, eventually managing a team of 30 people and five sections. And I was quite young then, so I think it was a great achievement, and I primarily worked with server technologies—hosting, servers, and cloud computing, which wasn’t as prominent back then. The focus was more on shared hosting and hosting technologies. I also worked with domains, and that’s where my technological career really began.
Bartosz Ziółko: Okay, and later you were a PR specialist at various IT companies, if I’m not mistaken.
Beata Mosór: VSoft is a fintech company. I worked on projects there, primarily low-code solutions for fintech, like itBuilder and archITect. These solutions supported the development of large-scale solutions for banks and insurance companies, and I focused on promoting and supporting their sales through PR efforts. Lunar Logic, based in Krakow, is well-known, largely because of people like Paweł Klipp and Paweł Brodziński. I handled international marketing and PR, as well as product development. They even created code for Twitter.
Bartosz Ziółko: At Google, it sounds like you had a collaborative relationship, not an employee one?
Beata Mosór: No, I was never a Google employee. I always describe it as collaborative work. My collaboration with Google started in 2012. I began by co-organizing various events through the Google Developers Groups, and I also co-organized Google DevFest. Later, I helped to create a university course with Google as a partner. I worked with Google Analytics Wednesday, and co-organized Google IO Extended here in Krakow. My collaboration took many forms, until I left Lunar Logic, and joined forces with Dawid Ostrowski. I was fortunate to assist him in organizing the first Google Launchpad in Warsaw, a startup acceleration program. I also had the pleasure of mentoring at Google Launchpad, both in the Krakow and Warsaw editions. That continued until I had my second child, and could no longer manage multiple businesses.
Bartosz Ziółko: Right, and that’s when you began to start your own businesses. I haven’t heard of most of them.
Beata Mosór: My first projects were Project: People and Project: Values. I always create multiple brands at once. Project: People and Project: Values were launched around the same time to see what would happen. Project: People was much more successful than Project: Values. You might know PP as a lean and strategic agency that mainly worked with large technology companies and startups, corporations. We collaborated with ING, Nationale Nederlanden, T-Mobile, Google, Ringier, and Axel Springer. We really worked with major players, but also many startups. In total, over 180 companies from 18 countries over 5 years. It’s been seven years officially, and the company still exists. I’m still listed as the CEO in the company register, waiting for it to be removed by gov entities, but we haven’t operated since August 2022. That’s when, with Asia Oswtafin, they decided to shut down Project: People.
Bartosz Ziółko: Okay, and now we’re transitioning to the topic that’s central to our conversation: the new wave of AI in marketing.
CIRCmodel.com as AI-based Business Operational System
Beata Mosór: I worked in marketing a long time ago, really at the beginning of my career. Later, I moved into management, leadership, organizational building, and product development—building technology products. CIRCmodel.com is my latest project, which I’ve been working on since Project: People closed. I’m thinking about what to do next, and for some time now, I’ve been developing CIRCmodel.com, a model to support building organizations, managing organizations, and their processes using AI agents. It’s a completely new approach to organizational building, and also organizational transformation using AI. This means potentially replacing human workers, but not in the traditional sense of using an agent to automate tasks. Instead, the entire organizational management model will change, roles will change, and it will require training AI models for specific roles.
Bartosz Ziółko: AI in the marketing department? Or the entire company?
Beata Mosór: The entire company—the whole organizational management model, including the executive board. For example, in the past, organizations were often founded by someone in marketing or sales, along with a technologist (CTO) and a financial officer (CFO). We had CEO, CFO, and CMO, but with AI, these roles in the executive board will evolve. The responsibilities of the board will change, and they will basically work only with AI models or AI operators. That’s my concept, and it’s supported by the AI development trends and the experiments I’m conducting.
Bartosz Ziółko: But are you talking about a model for the marketing department, or a model for the entire organization, like a factory?
Beata Mosór: You could start a company that immediately has an AI-powered operating system. So, starting a company now (Therannos, if I remember correctly), you could use the model I’m developing to set it up entirely without employees. Essentially, as a solopreneur, using AI models based on my frameworks.
AI vs ML in Marketing
Bartosz Ziółko: I see. Now, what are some less drastic, broader trends you see? What’s the impact of AI on marketing? I’ll offer a little guidance and context: everyone’s talking about this new wave, and a particularly significant development was ChatGPT a few years ago. That definitely changed things. Ten years ago, there was a similarly major development: deep learning networks. Those also had a significant effect on marketing, leading to the development of various kinds of profiling and so on.
Beata Mosór: Well, there was a rise in the discussion of behavioral economics as a field, and profiling based on behavioral patterns and user profiles. I think Google’s paper on Transformers is a landmark publication regarding AI, which fundamentally shifted perspectives on technology, and how to use it. Marketing trends have evolved in various ways, and these developments were prominently used at the Krakow Tech Summit to tailor marketing content to user needs, or in the case of ADTech, to adapt ads to user needs. Looking back now, everyone’s worried that the advertising market will disappear entirely.
AI & Marketing Labor Market
Bartosz Ziółko: So, you mean the labor market?
Beata Mosór: No, not the labor market, but the advertising market itself. For example, with Perplexity, launching their model for recommending products, specific products were suggested. You could say a similar concern emerged with Microsoft’s investment in OpenAI two years ago. We might end up with a completely monopolized market, with no advertising as we know it, eliminating smaller players, microbusinesses, and small companies. These entities won’t be able to afford paying as much as the big players to appear in search results using chatbots. Imagine asking ChatGPT, ‘Recommend running shoes,’ and it only suggests three brands because it doesn’t have room for more. You could imagine those three brands paying the most.
Bartosz Ziółko: Three is still optimistic and a lot…
Beata Mosór: However, I look at it from this perspective: using AI allows for a perfect match between supply and demand. These recommendation systems allow for a personalized experience, fitting my purchase history, my needs, my habits, and my style. A simple example is AI in Netflix: when we think about it, Netflix tailors content to our lifestyles, when we watch movies, what kinds of movies we watch, and what games we play. It suggests content perfectly tailored to our needs. And in the advertising market, we can strive for the same model—we’ll see only content and ads tailored to our purchasing capabilities, what we usually buy, and even what we’ll likely need in the near future, like a new face cream or new shoes, because we recently bought a cream whose effectiveness is limited to a certain time frame.
AI & Data-Driven Marketing
Bartosz Ziółko: So you believe such mechanisms already exist, or are you referring to the future?
Beata Mosór: Well, in the cosmetics market, such mechanisms do exist. Rossmann uses them, and it’s not AI, but simple datasets utilizing purchase data and combining it with data like weather and air quality.
Bartosz Ziółko: It depends on how we define AI.
Beata Mosór: Right, for me, it’s still Machine Learning. Let’s look at Rossmann as an example—a good example because many people are familiar with this. There’s an app that collects data about our purchases. Every purchase requires scanning the app or entering a phone number. In essence, they link your purchase data with your phone number. The specific initial data point isn’t important. They also gather location data; they encourage users to share their location data, showing where they go in the city, where they travel, and their frequent locations—through running competitions, for instance.
This allows the Rossmann app to build a detailed profile of each user, tracking their movements, purchasing habits, and the time of day they shop—in essence, most geo-location data. Add data from weather services or public event listings, and you have a very accurate picture of the user. You can target them with ads in a very effective manner. Not just upselling or cross-selling, but simply reminding them about purchases, at the right moment, when they are likely to buy, or when they walk by the store, or are habitually shopping at that time of day or place, or have already been in that location.
Is remarketing AI?
Bartosz Ziółko: But honestly, is this technology already that advanced? Because not long ago, internet advertising, at least on websites, was largely about seeing ads for things you’d already bought for the next couple of weeks.
Beata Mosór: What you’re describing is a simple remarketing mechanism. You saw those ads because they were programmed by people. People generally think the more ads, the better, and they don’t set any limits. Most advertising systems have something called “capping” – a limit on ad frequency. Your ad won’t be shown to more people than a certain amount, or shown to the same person more than X times a day. This makes sense in terms of building touchpoints and encouraging a return to the shopping cart, or completing a purchase.
However, now, companies like Meta are aiming to (as Mark has publicly said), to make it so these advertising settings, like frequency and audience targeting, won’t have to be manually adjusted. They’ll be determined by AI based on product costs, locations where the product is sold, and the product’s ingredients. This will be a perfect match between what we, as sellers, offer, and what customers want. AI will make that match, and we won’t have this problem of digital advertising clutter anymore, because ads will only appear when you’re most likely to buy, or at the very least, to make a purchase, or when the chance you’ll spend the most is highest; in other words, when the ad mechanism is most effective. Of course, these systems and models aren’t perfect yet, and we might not even be aware of the truly perfect ones, as they are still in development.
We’re also in Europe, so the digital footprint data collected on us has some limitations. Things look quite different in the US or Africa, where there’s much more data available about consumers. This is also true for data sales. The US also enjoys a high degree of freedom in this area.
European regulations, however, don’t adequately protect users. My experience is that data protection laws are somewhat illusory, because even if the law prohibits collecting data from cookies, or requires informing users about collected data, it’s largely ineffective. Browsers like Chrome from Google, and other major tech companies, are able to gather a massive amount of data on our online behaviour, our digital footprint; and this also applies to phone apps that access the same data. Just as I talked about Rossmann and location data, we’re also sharing location data via Google Maps.
Bartosz Ziółko: But it seems like it’s become quite bureaucratic. It really comes down to presenting information, and getting someone to click, because if they don’t, they can’t use the site. I’ve heard stories about Facebook, specifically that if you spend the first $1,000 on advertising, it’s basically wasted because the algorithms need to learn.
Beata Mosór: That means the model needs to train the algorithms. You pay for the training, just like you pay for tokens and data when training your AI model. I haven’t had any experiences where the first thousand dollars is wasted. My experience with advertising agencies is that they often lack the knowledge and experience in how these systems work, and they set things up in ways that are far from optimal.
For example, some charities I work with come to me with questions about their advertising campaigns. I recently had one charity client who asked if their advertising campaigns were set up correctly by their agency. I told them that the agency had completely misconfigured the campaigns, with massive user cannibalization.
Customers who would have bought something were instead shown ads for the same thing, paying for traffic they would have generated naturally. AI mechanisms will eliminate all of this. Large companies aren’t focused on maximizing profit from each user. They’re focused on ensuring the user experience with their ads is optimal in the long term. You’ve also asked about agencies disappearing as the advertising market changes. I believe those agencies who don’t effectively use their clients’ money will vanish. Users will realize they can manage their ads better using AI and get a higher ROI, which will make the process much easier.
AI in Marketing University Course
Bartosz Ziółko: How is the marketing job market changing? Are copywriters and graphic designers facing significant challenges because they’ve been somewhat replaced?
Beata Mosór: I’m launching an AI-based marketing university course, and luckily Bartek agreed to be a lecturer. He’ll give an introduction to AI models, and I’m really pleased he’s doing this. I believe the role of marketers, especially those creating content for blogs, social media, and performance marketing (which we often refer to as ads, but also SEO and SEM), will change. They’re already changing, because often you can create content for a blog or website much faster yourself than through an agency. The back-and-forth process with an agency or marketer is much longer than simply putting information into an AI and giving it a straightforward prompt. There are even templates for this type of content in Gemini, though you don’t need to use Gemini. Google Drive even has AI tools that quickly generate content for media calendars.
Bartosz Ziółko: But does it track the company internally?
Beata Mosór: No, the tools I’m familiar with work like this: on Google Drive, when you create an Excel file, you can choose a template, and like AirTable, you can choose a template for a media or content calendar, which immediately provides a structure for content publishing. This can be the foundation for AI-generated content.
Autonomous AI agents
Bartosz Ziółko: Are there tools that track your computer files, scan the internet, and suggest things like posting to Facebook today? Is that the next step?
Beata Mosór: I’d say Anthropic started a discussion about whether AI agents can use computers. Their public release of this feature as a function triggered the conversation. I suspect no major player will officially implement this functionality until there’s a corresponding regulation in place.
Bartosz Ziółko: From my perspective, the big players always come in last. That’s obvious.
Beata Mosór: Well, Anthropic isn’t a fully established corporate player yet.
Bartosz Ziółko: I spoke with a colleague who’s developing AI algorithms at one of the world’s largest banks, and she said that ultimately, these AI algorithms need to be rewritten as conventional algorithms to achieve similar results, because the bank is worried.
Beata Mosór: Banks are worried about copyright and the fact that AI doesn’t have a sense of copyright. There are two aspects to this. First, the functionality where AI can operate on your computer as if it were you doesn’t require AI.
I worked on an RPA solution in 2018 and 2019; we built it with RPA, so AI wasn’t strictly needed. But that solution had technological limitations. It only functioned on Microsoft computers, specifically those with Microsoft’s operating system, and needed a physical installation—a prerequisite installation file, for example. There had to be an administrator who initially installed it on the computer (it didn’t come as a .exe file, as is typical on Microsoft systems) and, due to its intended use in production environments for larger companies, needed a clear user action, a confirmation of agreement to carry out specific actions. This restricted it to certain system operations.
Now, Anthropic has shown that these actions can be practically unlimited. And we’re hearing that AI agents have started sending emails on your behalf, because they’ve determined that’s usually what a user does. Just yesterday, I was wondering whether AI agents could be tasked with training other AI models—and the answer is yes, but it would require a huge number of tokens, a high level of trust, and models trained for very specific tasks. That means we’d need an AI model that conducts research, one responsible for ethics, and so on—the equivalent of employees who specialize in particular areas and train other models. Ideally, a hybrid approach would use different models from various technologies; not exclusively Google models or Anthropic models, to avoid a lack of diversity in methodologies and thought processes.
The differences between, say, Anthropic‘s Haiku and Opus models, are in their communication methods and the tasks for which they were trained. But each successive model essentially builds on a similar dataset, just adding to and improving it. This can involve enhanced security measures. This is something Dario is openly discussing. So, I find this a very interesting point to consider and reflect on.
AI in marketing: landing pages & www creation
Bartosz Ziółko: Personally, I’m quite skeptical about completely autonomous AI taking on higher-level design and decision-making roles. However, I’d also like to ask your opinion on the practical benefits, drawbacks, and value of AI-generated content. Does it make sense?
Beata Mosór: Basically, these solutions, which laid the groundwork for AI when hardware wasn’t yet up to the task, have been around for a long time. I actually used Wix.com in 2007 or 2008 as a low-code platform for building websites. Now, Wix is a dominant player in the US for low-code website and e-commerce solutions. And these platforms, in essence, form the entire structure, the infrastructure, of such solutions. We have platforms like yep.so, for example, where you input a short description of your business, and the AI creates your entire website, complete with structure, visuals (images and graphics), and content. Naturally, this content is editable later on. It’s essentially a no-code platform; you just click, edit, and publish.
Bartosz Ziółko: What do you need to provide for these platforms?
Beata Mosór: It depends on the specific platform. On yep.so, you simply put a short description of your business, and you get a full website with structure, visuals (images and graphics), and content. It’s editable later on, of course. It’s a typical no-code system; you click, edit, and publish. And there you have your website.
Design Systems and AI
Bartosz Ziółko: Do you think these platforms can adapt the site’s style and color scheme to match the business’s theme?
Beata Mosór: There are established design patterns. Design systems like those from Microsoft or Google, which simplify certain models for apps or websites, have ingrained patterns of thinking. When you think about finance, you typically think of a blue-and-white or black-and-white website; you rarely think about other colors. When you think about ecology, you think of green. For most businesses, especially small businesses, this makes sense to leverage these established patterns. There are patterns that these platforms employ.
Of course, everything depends on the prompt you give it. Because when you provide a brief description of your business and ask it to create a visual representation of your value proposition on a website, you need to formulate that description well. It’s essentially a prompt: “Create a website about X, Y, and Z.” The quality of the response you get depends entirely on the specifics of the prompt, the input, and the context provided. A short prompt may yield a simple, generated response, while adding more context, training, or data to the prompt can improve the quality of the outcome.
Platforms like yep.so don’t have access to a huge amount of your data, because people don’t usually build websites every day. But, if we consider a platform like HubSpot with its AI, it can already do this using historical data, especially for larger businesses, and do it very precisely.
Bartosz Ziółko: Okay, we were supposed to discuss one more topic, but we’ve been talking for over half an hour now. I try to keep these episodes to half an hour, so perhaps we can continue another time, in two or three weeks. We could immediately announce that we were planning to talk about AI’s long-term impact on society and what life will be like in the future. But for today, thank you very much for the conversation.
Beata Mosór: Goodbye!
Interviewers:
Bartosz Ziółko
Dr. hab. inż. Bartosz Ziółko, former professor at AGH University of Science and Technology, founder of Tesserans, an AI company specializing in financial monitoring. Co-founder and former head of Techmo, a technology company providing speech recognition and generation solutions. He holds a degree in Electronics and Telecommunications from AGH University of Science and Technology and a PhD in Computer Science from the University of York. Achieved habilitation in 2017.
Author of over 100 academic publications, two patents granted by USPTO and one by EPO. He is also the author of the book “Speech Processing.” He has held a scholarship at Hokkaido University in Japan and participated in the TOP 500 Innovators program at Stanford. Dr. Ziółko has been involved in over 10 national and European research projects. The company he founded developed a speech recognition system processing approximately 100 million conversations. He is currently an investor and AI consultant, actively developing the Tesserans startup.
Beata Mosór
Graduate of Knowledge and Innovation Management from the European University in Krakow. Has delivered guest lectures at, among other institutions, the French EDHEC Business School, WSB Merito Gdańsk, and the European University (International Seminar and International Seminar MBA for German university students). Co-author of The LiGHT Book (alongside authors such as Cialdini and Zimbardo).
Possesses over 18 years of experience in technology products and services, specializing in strategy and leadership. During this time, she built marketing within nazwa.pl (growing from 100,000 to 1 million customers), established the lean strategic agency Project: People (with over 180 clients from 18 countries in 5 years), and collaborated with global brands including Sabre and Google. She has also been a co-owner and/or manager at Project: People, Project: Values, Reversum.io, and the cooperative societies Hermes & Partner. Currently, she is building CIRCmodel.com (an AI-based Circular Operating Model).