Hi everybody. My name is Udi and I'll be talking about. The rise of generative AI in marketing which has given rise to the term called vibe marketing. So a quick intro about me. I have over 16 years of experience in marketing, MarTech, and also go to market. So I'll be talking about the current state of use of generative ai where it's going to go, and what challenges you're likely to face. So with that we'll start off with a presentation. I think the first thing we need to keep in mind is what does the word vibe actually mean? Because and honestly I don't really like the term vibe because it tends to give people the impression that you're trying to match the vibe of your audience and, that's what vibe marketing is. That's I would say, 20% of what the truth is, but actually. The definition is much closer to what the original definition of Vibe was, which is Vibe Coding and Andre Carti, which whom you would know as one of the most prominent figures in artificial intelligence. I think he put it very succinctly, is that. If I'm building a project or a web app, and it's not really coding, I just see stuff, stay stuff, run stuff, and copy paste stuff, and it mostly works. What he's basically talking about is prompting these large language models to help you with the coding and then putting down the code in your wherever you wanna post it, and then getting the output you need. Vibe Marketing does share some of these characteristics, but as you will see, it's actually much more than that. Just to give an example, so here's my vibe coded project, which I built over eight hours. It's just a fun little website where you can go and click on all the places that you've been to. The reason I'm putting it up here is to show that how accessible coding has become to people. I do have a background in computer science, but I haven't done any programming for the last 15 years. And despite that gap. I was able to come in speak to chat, GPT, cursor wa windsurfer, et cetera, and get a working product out in less than eight hours. Now applying this to marketing, I think what vibe marketing is and there's no clear industry definition, obviously, but what I think it is giving. Everybody and not just marketers. The ability to interact with these chat these large language models, whether it's chat, GPT or Gemini or, whichever you want to take. And then cutting down the time to market from. Right from campaign planning all the way to execution, iteration reporting, and then re-op optimization. So in that aspect it helps you, first of all these tools help you to understand what the what personalization levels you will need. For individual cohorts. And that's basically what the vibe guidance is all about. And, and then that basically allows you to create hyper-personalization so you can personalize content down to the individual and not just groups. The second thing is it helps you to rapidly scale up. Creating content, copies, images, videos, rapidly, create post, start the campaign, observe the results, iterate, and then optimize based on that. And also because you have a lot of social media listening tools nowadays listening tools were a little bit difficult to use in the past because the systems that used to take all the text, the kind of feedback and review texts those were not fast enough to help you react fast. But of course, because of large language models now you can literally throw. All your reviews or comments from your customers and then get the necessary instructions to move to the next step of your campaign. So yeah, yes, vibe marketing does touch upon the vibe aspect, but I think it's a lot more than that as you can see. I think before we move on to what exactly is happening, I think we need to understand that. Marketing as a whole, or mar marketing technology as a whole has gone through a lot of evolution, starting from moving to cloud-based services. Whether it's using Google ads or Facebook, et cetera the advent of mobile which allowed you additional creative space, additional formats, it's additional targeting options to go after to the increase of social media where your PII data comes into the picture, where you create that whole unified profile about your social preferences to, to help marketeers do better targeting. And all of these gave rise to big data. And that led to the advent of artificial intelligence. Now generative AI is only a type of artificial intelligence, okay? Because we generally just talk about AI as ai. And that is the only kind of. Artificial sorry. Machine learning is the only kind of artificial intelligence that we humans know of. But where generative AI comes in is basically it's a deep learning form of artificial intelligence, which means that it is able to process, I think analogous to how the human brain, you can do processing of multiple layers of information to come to your output. And that's basically. An evolution from what marketeers have been doing for the last 20 years. For example, I used to head a product called Adobe Media Optimizer which was from 2014 to 2019. And that was a very powerful machine learning system, but it was a predictive AI system. There was no generative ai. What generative AI helps you do is it helps you to create, so whether it's art new forms of creatives, and of course, as we saw it uses deep learning as its foundational form of delivery. Again, the other thing we need to understand is what exactly are these data models that people keep talking about, LLMs and models and et cetera. But the easiest way to describe that is it's, and it's a system which takes in existing data. Okay. Like text image videos. And it understands what those things are. For example if you ask Chad GPD to create an image of a cat, it looks at a database of what cats look like, and based on your prompt, it generates or it predictively generates what should come next. And that's basically what it throws. So that's basically what data model is. So you take data in, you prompt it, and based on that you can get the output that you're looking for. So that's like a very ELI five version of Gen ai. And the reason I'm talking about that is vibe marketing. The whole term came from the fact that generative AI became so much access, so easily accessible, not just to individuals, but also to enterprises via APIs. What why marketing now? Because gen AI came, I think it's been about two and a half years since open AI can launched the public version of chat GBT. But what's happening now is, and this is true in, in the industry of marketing, that marketing does tend to react a little bit slow to these technological advances, not because that's the nature of it, but because a lot of iteration and testing usually has to be done before taking an idea from a concept to an enterprise ready platform. Okay? But what's happening now is like, based on various studies and surveys, you can see 67% of CMOs are pivoting. They're pivoting towards generative ai, but. By doing that, they are indirectly also getting into the whole vibe marketing scene. Second thing is people are starting to notice that. When you, in the past, you need to have, you had to have 10 marketeers, but now you can get that job done with just one person and use a bunch of these AI tools which replicate, and in many ways, they do a better job than most of the human counterparts. And because of, and a second thing is it's not just a number of people who are involved, but it's also. How much time it's taking for marketing teams to go from conceptualization to execution of the campaigns. And then we also see that there is a huge shift in consumers also wanting vibe match. Honestly, that just means personalization to me. But we are now able to personalize not just at a need level, but also at a level. For example, you could. You could include you could make your emails sound in a certain slang or in a certain dialect, which was impossible earlier because you had to manually instruct the system to do that. But now by putting checks and balances in place, you are able to do that by just sending out a prompt. And most importantly, obviously it's all about the money. And that's where we can see that. Companies using a lot in generative AI are typically seeing a higher return on ad spend which is always the most important metric. Key differences are just to lay it out, speed, it's much faster to launch. Any campaigns if you're using generative ai. Team sizes are smaller, more agile. And you also have a little bit more control because the coordination aspect of working with multiple people disappears. In this model you are also able to run these campaigns at a much lower cost. AI tools don't cause, don't cost like enterprise sauce tools, they, they're not like a million dollar purchase you make, you usually pay on a monthly basis. And if you don't like the tool, you can always move to a different one. Personalization wise in the past, as I said, you could personalize you, you can still personalize, but now it's hyper-personalized. It's very mood based. Testing, it's it's again, faster because, not because it, the, the testing methodologies have changed, which are. They're pretty much the same AB testing multi-arm banded, et cetera. But the speed at which you can do it now is much faster. Also you're able to focus more on the emotional connection, which with the customer, which is becoming a very important factor, especially because brand loyalty has dropped significantly over the last couple of over, I would say over the last couple of decades actually. And finally, workflow. If you know how to use the right tools, if you have a little bit of understanding of automation tools like Zapier, or if you have a little bit of skill in coding in Python, or even like prompting chat, GPT or. Any of these tools to give you the right prompt, you're able to move really fast and adopt really fast. So those are the main differences between traditional vibe marketing. Now, the way vibe marketing moves, and this is, and of course different companies have different flows, but this is my way of doing it now is and it's not different from running a standard marketing campaign, it's just that the tools have changed to some extent. So defining the vibe defining the vibe is basically defining what a creative direction that we are trying to get for this particular campaign. Take that, prompt it into AI in natural language, and then the AI starts to generate content. Okay? And right now most of this content is primarily text-based, but we are reaching a point where. To a certain, at a certain limited level, you are able to create images and even videos which have customization. Although in mainstream, I probably seen only five or six video ads, which were completely AI generated. But that's also because it's really expensive and it's not a great solution for SMBs and startups right now. But again, we'll get to that in a bit to see where the where the remaining remainder of the challenges are. And then after that, testing and launching again there's a bit of manual work involved nowadays, but I think in about five years that's gonna be automated as well. Then it's about analyzing and optimizing. So bringing the data you can automate that data, send it to something like Google Data Studio, Tableau, or you can just do your good old. Excel analysis, it doesn't matter, but what's important is taking that data and then having a chat with these LLM models, understanding where the optimization opportunities are, and then adapting your campaign based on the the feedback that you are able to generate. Okay. These are just a few of the tools that are available right now and the best thing about these tools is that. Whether you are a startup, scale up SMB, or large enterprise, there is something for you out there. And I have been involved in the entire spectrum of company sizes. So for example, as a big enterprise, I would just speak to somebody like Salesforce, Adobe, or HubSpot use their enterprise ready products. Although they can be a bit inflexible sometimes, but for majority of these large size clients I think those tools will be sufficient. And of course you will use those in combination with some of the smaller tools. If you're a smaller company. You have these subscription based options and I've seen tools which are as cheap as $14 a month, and they do a decent job. And I, and we and I'm saying decent because as you will see in a bit why the, why marketing hasn't actually reached its full potential when I do the demo in a bit. But as you can see. If you wanna start testing out, there are plenty of options. A lot of free options as well. So I would say go and test as many of these as you can see which fits your company's requirements the most, and then stick to those. Some of the recent viral campaigns, which have done vibe marketing. They come from Coca-Cola. Which which made this really interesting campaign where they use Dali. So Dali is the image generation model of open AI to basically let fans generate AI r inspired by the Koch branding. It sounds really simple, it caught on really well. So they received over 120,000 submissions. And over 7 million social impressions. So I would say that as a pioneer of using gen AI for advertising, I think Coke's doing a great job. But again, if you move to a more performance based example, which is Heinz catch up they did a similar thing where they basically used image generation models to create visuals of ketchup bottles. Sounds really. Basic, but they did get a 40% higher ROI on these AI generated video ads. So I would say that was a big success. Mango has been doing something like that, so they've been using virtual models for their summer collection. A lot of people are a little bit on the fence, but I think the quality of these image images generated are getting so good. There'll be a point in the future where it'll be, they'll be indistinguishable from real people. One of the most interesting ones I've seen though, was this Virgin Voyages, which basically featured gen ai, which is, Jennifer Lopez, but a deep fake versions of her. So you can create. Personalized video invitations using Deepak versions of her. And of course, it's a bit of a novelty, but it really impressed how Virgin was quick to doing that. Of course, that gave them tremendous amount of viral engagements increasing in bookings. So I think the investments that they made in this definitely paid off for them. All right. With that, I wanna do a quick demo to show you a very simple workflow of what a vibe marketing kind of methodology would look like. Now, obviously, what I'll show you is not an all encompassing. Example, big comp. This is probably more suitable for smaller companies, SMBs, startups enterprise companies will have some of these tools built in house with the right guard list. But anyway, let's move on to the demo. So let me pull up. So what we're trying to do is in this case, is we let's assume we, we are this company called Hoka, right? You've probably seen shoes from Hoka. So let's assume that we have a new campaign coming up. And we need to be able to create ad copies and also add creatives, which which basically, so typically what a digital, an advertising or marketing person would do at this point is they have to manually write down these copies, manually create these images to create variations, okay? And that usually would take depending on how big the camping is, that could take two to three days. But here's how fast you can do it in chat, GBD. All I have to do is, and this is something, this is a best practice, I would say create a separate thread with Chad, GPD, where you give instructions for a specific platform or a channel, like in this case I'm looking at X as an example. So what I'm telling Chad, GPD, is that. I need to create we, we will use this thread to create ad copies, creatives for my company, Hoka, for each camping I will share instructions, guidelines, et cetera, et cetera, to create the best ads possible, right? So it got it. And then here is the the camping instructions that I'm sharing with chat GPT. To be honest, I have used chat GPT to refine this. I just put like the raw information in and this is the refined output. But what's important is obviously to put the important campaign factors, like what your target audience is, what the brand guidelines are, what key phrases to use, color palette, et cetera, et cetera. So you, as a marketeer, you would know typically this is what you get a brief from the client or from your, companies, internal departments. But if you are in a startup trying to do vibe marketing I would say this is a really easy, simple way to make sure that your advertise your ads and your creative stay within your requirements. Alright and while I'm asking it is create three variations of the ad copy. Then three different images. And so it went to work. It created this really impressive ad copies with great call to actions with all the right information in place. And on top of that, it also helped me create this this copy. I think it looks great. I would just take it and put it straight up into an ad. Obviously there are some minor things I need to look after. For example, as you can see, the color's not contrasting with the background. And so then again, this is one of the things I wanted to mention earlier, the reality of why marketing is that you will still need a bit of human intervention. Okay? So it's still gonna make your work about five, six times faster. But there are these, if you are spec, especially if you're a big company with very strict guidelines, you have to ensure that these kind of issues don't make it to the final version. Okay? And this so it went on and on and it made, and. It stopped generating because I ran out of credits at this point. However, as a, if you're, obviously, if you're using a company account, then you can connect directly with Dali's AI and then get these images generated. That's a really powerful feature I would say. Now, as far as I know, there are no other third party solutions which are able to generate. Compelling ads like this at scale? At least in, at least based on my knowledge the alternative to this is, for example, you can actually go into Canva and I'm just gonna, I. You can actually go into Canva and then you can manually create those different versions of the ad as well. But again, that requires a bit of manual work. It does make your work a bit faster. I'll give an example. For example, let's say if you take the image of of the shoe, right? For example, let me just bring it up again. And there, there's a shoe and I want to quickly put it on a background, right? So all I have to do is remove the background. And there you go. That was done by ai. And then I can take a template from here. Let me put something like adventure, right? So obviously not all of these are gonna match what you're looking out, looking for, but I'm sure there'll be something out there. And just for the sake of, for example, we just put somewhere like here, oh, that's actually a video. Let me just gimme an example. Gimme a second. I'm just gonna try to find something useful. Oh, these are all videos for some reason. Just one second. I think you have to spend some time finding the right background, but once you do, of course you can upload your own background as well. But once you do that, it's very easy to just put this on that and it, it's not gonna be a perfect ad copy, but if you're looking for speed over perfection, then this is a great option. And then obviously what you have to look at is you still have to go in and and. You still have to go and analyze. You still have to upload those ad copies. Now, there are tools out there there are tools out there which help you, which connect automatically to Facebook and Google and, various ad platforms systems via API, where you can directly push the ad copies and the images. The only caveat there is that. Those systems don't have a generative AI system to generate those copies. So as you can see, there's already, this is one of the other issues of wire marketing is there, the entire ecosystem is fragmented. And because of which, you're not able to automate it end to end, right? So again, that's something else that you just need to be aware of. Apart from that, what you could also look at is doing some of these things on your own Now. This is sentiment analysis, which I was doing for a company. And so basically what we, they basically gave me a list of all the various comments that were reviews, et cetera, that were shared by the customers. And what I did was I spoke to Chad, GPD and said, Hey, you know what? I need to analyze these. I don't have a, I don't have a paid version of a software, but I'm sure there are free kind of libraries and tools available on Python. Let's use those. And of course, it gave me the code. All I had to do is put the data in and then after a few iterations and after a few to and fros. I managed to get this, basically to find out that the biggest problem that the company was having is they were ha people were having login issues. So this helped us prioritize the features that need to, needed to be fixed. And the best part is this entire process took me only about 20 minutes to do. And it was completely free, right? 'cause all I had to do was run Python. Now obviously, if you want to run this. Another time you can create this as a separate application and all you have to do is upload the data in the right format. So that's another way that Vibe marketing can help you save not just a lot of time, but a lot of money for your company. And also it can basically create these systems which are very specific to your company's requirements. Similarly, I did another analysis. And this was for segmentation to understand which area should we be focusing on to increase the revenue. And this was for a gaming company. So basically we found that getting people to a whale status, which is getting them to spend a bit more like fractional ends up getting them spending a lot more. So we were like, okay, maybe we should build features to help them get to that point. Also maybe, win rates, for example, could we reduce the difficulty levels of the game? And again, the entire process completely guided by chat, GPD. All I had to do was put the data that we had and. It produce a response from me. So that's another way in which I think you can definitely save a lot of time, money, and on resources to get an output like this. Okay. So I think what we've seen so far is that vibe marketing is fantastic. It's going to change the world. It's gonna change the world in many ways in how people do marketing go to market, the speed, execution, time, money, et cetera. But. As you saw in the demo there, there's a bit of an issue there. Is the whole ecosystem is very fragmented. Now, this full stack marketing roadmap, as you can see in the top layer. This is where the AI part comes in, the vibe marketing. Now, I'm not saying that the tools which are below that category don't have any artificial intelligence, but their use of AI is very specific. It goes quite deep and you can do all the AI stuff you want at the top, but you still have layers and layers of integration to go through. And at this point, of course you can chat with chat g and find out, what's the best way to do it. But usually what I've found is that someone, somebody with experience running these various tools is a much faster and more accurate bet because you don't have to do that trial and error to get to the output that you're looking for. So I think that's the first problem. The second and more serious problem is that there is a huge data safety issue. Now, of course, whether you're a small company or a big company ensuring that your in-house data, your ips, et cetera, stay in-house, right? It shouldn't get shared on the free internet where anybody can access it. It has happened, for example, when chat GPD first came out. I think there was a developer from Samsung who wanted to get some ideas about his code and he published the entire code base of his product, Samsung product on chat GPT. And because, these LMS are a bit of a black box, anyone could access that code after that. So those are issues. To avoid that companies like this framework is from my days at Salesforce. Companies do, have different frameworks. This one is called the data security framework. Now, this works really interestingly because when we prompt something like prompt, you prompt an LLM, it goes into the framework. What happens is there is that the first step is there is a secure data retrieval, which is to ensure that only data, which has been sanctioned or approved for sharing with these lms, get approved, right? So companies like, as you can see, products of major enterprises like Salesforce, they already have those guard rates built in place and that's why they charge, hundreds and s of dollars for the products. Second and very important is grounding. Grounding is where the actual personalization happens, right? If you take an example of Ticketmaster, right? And for example, let's say I like pop music. My wife likes jazz. I. You want to send future pop music concert recommendations to me, not my wife. It's a very basic thing, but the important thing is that there's data being shared between you and a potential third party, LLM, and there are ways to prevent that data from leaking. For example, you can do masking, you can do encryption, pretty standard procedures, but you need to make sure that is happening at some level. If you are interacting with an LLM. With your company's private data. And then comes the gateway feature, which is, as the data is sent to these lms, you have two options. One you can work with. You can work with one of these big lms using APIs to make sure that the the kind of right parsing of data right output is coming out. But you also have to make sure contractually or by system design that. The data which has been shared with them is not stored. It's removed. I'll offer your session with them. The other option, obviously is that you can ho you know, have your own locally hosted models, which may not be a financially viable option for smaller companies because they tend to get really expensive as the processing, capacity requirement goes up. And then comes generation. Generation is where the output is. Seen. Usually a strong data security framework will have guardrails in place to make sure that there are no hallucinations or toxicity coming out of generation. Now this process is, is best. If it's done automatically, but it may need some manual intervention as well. And finally, you also need an audit trail to, to understand that if something goes wrong, you need to know at what point of the process what did this happen and who was operating with the system at that time Only once all those. All these different areas are looked after, that's when you can send it to your third party platforms like marketing apps, APIs, data exchanges. And I think this is a part of the process that. A lot of marketeers are underestimating right now if you ask me, especially smaller companies. But if you don't take these precautions now as you scale up, become a larger company these could come back to haunt you. So just be careful about that. So I think that's all the time I have for now and thank you very much. Obviously, if you have any questions, feel free to reach out to me. My email is up there. But once again, thanks very much for coming to my presentation and I hope to hear from you. Thank you.