How AI Is Transforming The Communications Industry: Lessons From Inside The Agency

Okay, welcome back to my podcast on the future of business and technology. Today's topic is data and AI in the communications and marketing area. And my guest today is Ephraim Cohen, who is the global head of data and digital at the global communications and PR consultancy, Fleischmann, Hilard, Welcome, Ephraim. Thank you. Good to be here. Good to see you, Evan. Yeah, we met a few weeks ago where you moderated a panel and I was a guest on the panel where we talked about AI in the communications world and after the session we had a really good conversation and I thought actually there's so much to talk about and I was so impressed by some of the things you were doing internally that I thought actually we need to reverse roles here and I need to get you on as a guest. So it's lovely to have you with us. We're joining us from today. Today I'm joining from Connecticut, I'm based out of our New York and Connecticut offices. Very good. Yeah. So you are on the global leadership team at Fleischmann Hilard. Maybe you can give us an overview of your role and what it entails. Sure, I'm part of the global leadership team or Fleischmann Hilard overall. So the role, especially in that capacity, looks at how do we make sure we're using all data and digital technologies, the dominant digital technology, of course, being all flavors of AI, but a particular generative AI. And how do we use it really in three ways? We need to upgrade our populations to something you're always doing as a business. We need to rapidly evolve our capabilities and continue that evolution for the foreseeable future because the technology continues to evolve and the data sets continue to evolve. And then we need to not just upskill but upskill inspire and redirect our people, our communications counselors. So it's overlooking all that. It's a fascinating role. And then I also want to see our media, social and digital capabilities, which was a role that had for a number of years, I think, really all prepared for this moment. Yeah. And it's such an important role in so much change happening in that industry. So I would love to get you take on the current state of play when it comes to AI and data and technology in the communications and marketing industry. What are you seeing right now? Well, certainly a lot of change at a pace that causes a lot of pay us. Now there's an interesting challenge. Anytime we've had a shift in the media and communications industries, I've driven by digital technology change and we've seen a few of those shifts since early 90s. In rapid succession, there were ones before we usually had decades to prepare. Anytime you've seen that, we've had the excitement about change, the nervousness about change, and the chaos that can come with it. This is such a rapid pace that it's causing, I'll just call change in suggestion. Something passed. I would always look at these moments as these were opportunities. Anytime things are changing, there is an opportunity as a professional as a business. You get that much stronger, that much better. But we've never had to deal with that at this pace where it's continuing to evolve. The types of leaps and improvement we've seen in the data we work with and the impact about technology is at such a rapid pace. I do this full time and have a hard time keeping up with all the changes. That's a sort of environmental observation. In terms of the impact of the industry, I think this is the other part of your question, there are two layers to it. We always have to look at a, how is the media ecosystem evolving and then how do we evolve? From a media ecosystem, and by that I mean is, what is that ecosystem of all the news and influencer and influence sources people go to, all the channels they use? How does that entire ecosystem that mix? How is it evolving and how are people's behaviors evolving with it? That is changing a far more rapid pace. It's getting a little more difficult to figure out exactly what it's going to look like in six or 12 months. For example, everyone's very focused on optimizing for generative AI platforms, we'll call GEO, a generative engine optimization, but how those platforms come up with their answers. So how we optimize for those platforms, that's changing in a rapid pace. As these people figuring out what's the mix of tools they're going to use to get a platform? When do they go to chat GPT versus Reddit versus New York Times versus their favorite influencer on TikTok or Instagram and new networks? Or two, and other, you know, could be a new network. It's certainly setting up to date. Another new network, so other new content types that may come up. So we have to keep trapped of that rapid change, look at it from an audience perspective, and help our clients, which were a large firm. Many, not most of our clients are even much, much larger and more complex than we are. They change just as rapidly to keep up with that ecosystem in their audience. Then you have on our side, what are all the different skill sets and capabilities we need to continually evolve. So for example, several years ago, we started down the path as a firm, and we said, "Look, data is really important." That should be obvious, but it's not something where we should just go to someone who's an expert in data, ask for an answer and get at that. We really need everyone to have high data fluency, the kind of fluency that comes with hands on keyboard work. I mean, every single person with the broadest range data types and data tools. So for example, we gave everyone access to Omnichom's Omni platform, which doesn't just have traditional communications data like news and social trends, but audience data, and talk them how to use it. So if you were a media relations person, you could learn how to create an in-depth audience profile. And then again, some people actually were using it for work day to day. Some people simply developed the fluency so they understood how to fit it into and make use of data in their council, which is being theoretical or just copying and pasting an answer from an expert. And we have terrific experts more advanced work. And now we have to do the same thing in AI. We have to get people to really understand what it is in a hands-on keyboard, a very tangible manner in order for them to make better use, to do theories, to do applications, to do parts of the ecosystem of our development. Otherwise, there's a gap between what you really know and what you're dealing with. And so those are rapid kind of layer changes that are going on now in that environment. We're still telling people everything I say to you, for example, is accurate as of October 21st at 922 New York time by 923. It may change. That's an exaggeration, but not as much of an exaggeration as it was 10 years ago, certainly 20 years ago. Yeah, very interesting. I see this across all industries right now. This rapid pace of change that we almost can't keep up. You have implemented AI very well inside your organization. I was very impressed when you told me what the way you've done this. I would love to for you to share that and give us an overview of how you're making AI and generative AI available to employees within your organization and how that is already changing how you work and how you are serving your clients and how you're running campaigns for them. Sure. So, first of all, we're getting everyone up to speed and then how it's changing how we run clients. So the people part builds off of what we call what I was just referring to, that democratization of data fluency and data expertise. So we already have a base of telling everyone, don't rely on central experts. You need to develop things yourself. Whether or not you're going to be doing the work, you will get a better understanding to make use of solutions built off data and now AI or AI and data and or to know how to actually develop ones for clients. So it built off that and that's a that was a five year campaign that was strongly supported by our then CEO, JJ Carter, who's now our CEO, who has made getting everyone highly fluent in both data and AI, a central tenant of, you know, a central goal, strategic goal of the organization. It is the foundation for the future. So what we're doing now is we're building off the Omnicom strategy. Omnicom built a platform that basically allows any agency and to get to every professional and they're rolling it out to everything with an Omnicom. Access to all the major, I shouldn't say all, but most of the major frontier models. So most of the major models in text, chat, GPT, Gemini, Nova, Lama, etc. video, you know, two, you have three and so on, an image, an audio of stack people have all the tools they need to create the right solution. So you take that and they have this massive data stack, which from what we can tell, they probably have the most comprehensive data stack that everyone can access, in terms of audience and communications and media insights. When you combine those things two things together, you have an incredibly powerful data stack. The first thing we're telling everyone is you have two sets of tools here. You have data, you have generative AI, you need to pick up on the experts in those. Now there's a, not just in what they are, but in using them. Again, I will repeat this several times in our conversation. I think you heard me say this, we met in person. Hands on keyboard learning. If you want people to understand something, they need to do it. You know, everything we do as experts is because we grow up actually doing the work. This is all new, so we all need to take a step back and actually do the work. Develop prompt skills. Learn how to develop an AI agent, develop an AI agent, develop several and stream them into a proper AI and an agente AI solution and so forth. So that's where we are right now. It's going incredibly well and what we're seeing is, and this is something that I think is typical of this type of change. When people hear about in theory, they're excited but nervous. They're excited because that's really interesting and that's fascinating. But certainly on a professional side of thinking, "But what does this do to my work and my job and my skills becoming outdated?" Once you get your hands on my keyboard and start learning how to create, you start thinking to yourself, "Oh, I can do this for my plants. I can do this for my work." I can create that, the creativity tipping point happens. And then you get excited about the possibilities. And other organically, they're naturally creating new skills. They're creating their job in the future. They may not know exactly what it is, but they're steering in that direction. And then we start structuring that solutions, which I'll come back to. So that data fluency, that ability to work with data and that ability to understand AI are two parts of what's really a four-part strategy. The third part are the people themselves. So now you have communications people. They have subject matter expertise. They understand the healthcare system, the farm industry, food and agriculture and the technology enterprise, whatever crisis and reputation, whatever topic of industry they work in, they have that expertise. And now they have data and AI technology to turn that expertise into a real solution. And in fact, some of our most powerful solutions today, things like our crisis simulators, our replication management agents and agents of solutions, they were actually developed by more senior subject matter experts. Not people had a technical background, but who took the time to learn, gotten fired, the creativity tipping point happens, and said, oh, I know what I want to create and I will create that. And when you have a subject matter expert acting as the engineer, you're getting better solution, both in terms of its usability and in terms of the output. And by the way, I don't want to say they are working in a vacuum that do work with our more technical experts. But it is the SME to subject matter expert, the counselor leading a solution. So they have the subject matter expert data and technology and is a fourth ingredient that's going rapidly. And that's knowledge basis. And those are particularly, you know, potentially one of the most important aspects of this four ingredient people data technology kind of mixture that we're providing to everyone. And for those, I guess we're listening to having heard about this, think of a knowledge based as a carefully curated library of information for a specific topic. So when you go to a general, a general AI application, it's like hiring an incredibly smart student who, when they were studying at the university, their library of information to learn from was a lot of just random stuff. And frankly, a lot of it was junk. So they learned to read, they learned to process information, but the actual content itself, is not a library you want to go to. And so what you're doing at the knowledge library is we're digitizing carefully curated libraries of information. So for example, if we're doing a crisis simulation, we are building it off of a library, proven case studies and best practices, not whatever happens to live in chat GPT. If we're doing insights into the food and agricultural industry, it is a library of qualified quality checked information built for this. So you get a much better, more accurate and a much, much better output. And you're going to see more and more of this as time goes on. So everyone has access to, you know, and everyone is getting access to, I have my expertise now. I know how to build an AI solution. And I know how to build it using data, using digitized knowledge libraries, and using AI technology with them. I can choose the right model for a given solution. So once you give that to people, if you really want to enact a larger change, they will start figuring out the solutions. They know their business. They know their clients, they know our agency and how we work. And so we're doing is driving a very bottoms up innovation wave, if you will. And then we're looking at it and taking out of it. What are the things people are developing that should be those hero solutions for the agency? But keeping that innovation wave going, that will continue to devolve those capabilities. That's the first part. Then there's the capabilities themselves. And that has a, that is the three part chance, really, to the three part challenge before. So we have to get our people working this way. They know how to talk about these newer capabilities to our clients. Then we're developing the capabilities. We're doing that by taking this innovation force and then, for example, to our media relations people, now that you have this expertise, let's put together, what is that toolbox? Or what does that solution flow of AI agents and data and knowledge, you know, the whole package together that makes your media relations function as one example, more powerful, more accurate, set your time spent on building relationships, building those relationships with deeper insights into the media and understanding how the media ecosystem works and developing story ideas with far more precision, both about what the audience is looking for and what the media is looking for. And by media, I mean that most it nasty quite whether it's a podcaster or a reporter at the Wall Street Journal or Boge or an influencer on Instagram or TikTok. So now they're able to develop that solution. We can make that more formal. And then the final part is, you know, clients have a big challenge. They have to help support their own company changing and they have to support changing their profession. We have the advantage that we can focus on our profession. You know, our work in agency, that is what you do. If you work in a bank, you have to change yourself professionally and support your bank's transformation. So we're trying to figure out work with our clients. Our teams are building these solutions and they know these solutions. They have the subject matter expertise. How can they work with you client to help you evolve as quickly as possible and make use of these more powerful solutions. And I'd say right now all this is moving incredibly fast even though we can feel midstream response. So how do you make sure that people gain the right skills? You said you wanted them to have hands on keyboard experience, building data literacy, building AI literacy, starting to learn how to use these tools, accessing the amazing data sets that you have. What sort of support did you give them? How did you make sure that those skills were being developed? Sure. I mean, a lot of it comes back to the basics of teaching. There's a lot of in-person and online workshops to walk people through. For example, how do you use data? What are the different types of data and then how do you specific tools? Same thing on AI. The basics of what is AI? What is prompting? And then we ship to hands-on to actually learning how to do that. And we want to do is directed. And the balance is every company wants employees to upskill. There's not enough time in the day to do all the teaching we really want to do. So our focus is on give them enough in terms of an understanding of data and generative AI technology, as well as other types of AI. Give them enough understanding of what it is. And then give them enough of the hands-on keyboard training. And try, it's a classroom. I make it fun, inspiring, get them thinking creatively. So they can go on and continue learning on their own. We do have continuing training. We're rolling out more advanced classes. But the idea is ultimately, it's the rule of, you know, hopefully you all have a teacher of where you've learned in the classroom. But the teacher gave you the tools and the inspiration that you wanted to keep building on that learning afterwards. So here you might learn how to build an agent in a workshop that we conduct after learning with this. So first you learn the basics of AI, then we do an hands-on training workshop on how do you build an agent. And then out of that workshop, you don't have five ideas for five other agents you want to build. And so you go ahead and do that. And when you're doing that, you're learning. It's like you're practicing, you're learning. It's like anything. You're developing the mind muscle for it. So that, you know, teach them what it is and then get hands on, hopefully as inspiring the way as possible are the two most important things we can do. And then continuing to direct them. So while people are continuing to. Do these trainings with us, but also learn on their own, they're understanding what they should learn. The reason I've mentioned agent training so far, you know, a lot of people talk about comp engineering skills and they are critical. When you learn how to develop an agent, you are learning how to apply your subject and how to expertise, how to choose the right data sets, if and when they apply, and usually do in our profession, how to connect and use the right knowledge bases and make sure their setup, say your agent is going to a quality library versus just using their library from University to CSU, that's not very good. And then how to pick the right model. So when you're doing the workshops, you're actually learning the value of all these components and how to put them together. And then as you continue to create agents and you're thinking about, I wanna do an agent that does this type of writing for this audience. And so they're following the choose this data set and this writing knowledge library. And I'm gonna use the cloud model this time. It's great, you're reinforcing those learnings. So that's been our learning approach and then continue to provide guidance and say to people, okay, now that let's say 30% of you are at the point, we have pretty good expertise in building solutions. We're gonna put you in this direction because here solutions you can focus on building that have real demand for our clients. So the idea isn't just to create a pure learning environment, but one that, and people should continue learning on their own for sure, but then help direct people to where it's gonna have the biggest impact, not just for us as a business, but for them as individual professionals. - So we're a pretty strong center of excellence that's directing obviously, incredibly strong. - I would love you to take me through a typical workflow then of someone working at Fleischmann here that creating a campaign for a customer. How is AI changing that? What is different today using AI agents and generative AI and how is this workflow different to what it was a few years ago? - So first of all, I love that question 'cause a lot of our industry we focus on what are your solutions, almost your package solutions? But the real power is just that. How is it upgrading, making more powerful the entire workflow? That's what it really comes down to, that everyone can participate in. Okay, so as an employee, you develop data fluency and you understand how to make use of data, you've developed AI fluency, you understand what it is and how to develop agents. So whether or not you're building or using what's in the workflow is, it doesn't matter because either way, you will know what to do. That's really important 'cause it will be. It's like you've learned to drive a car. I don't have to teach you again if I tell you to drive a grocery store or take an overnight trip somewhere, you know what to do. Let's take a generic campaign. So the first thing you're doing is, you want to understand, you have your clients business objectives, you want to understand your audience. So you might be going through a series of synthetic audiences. When you roll out synthetic audiences, we have been experimenting them for two years, but formally about six months ago and our view as synthetic audiences are, you always want to start with your audience insight. Understand your audience in the context of your objective. So if we want to sell running shoes to Bernard, we need to understand Bernard's interest in running shoes. We're Bernard goes for information. One spires what are related topics. That is a lot of expensive research or it was until two years ago. So first you start with synthetic audiences and saying, "Hey Bernard, here's a topic. Give me your, let me know, how can I best reach you? Which handles what types of content, et cetera." I'm oversimplifying in this interview, but that's essentially what you're doing at a much faster, lower cost higher impact manner. And that's important, it's not about efficiency, it's really about impact. And have a higher impact and to free up time and money to invest in that impact. To invest in audiences, that's one. Now that you have those insights, you might say, "Well, let's take those insights and come up with a creative camp and our creative team will work with different agents to really help them along the creative process." They then take that creative idea and they might, they were saying, "Okay, now it's time to turn it into a proper plan." And so we'll have a series of agents to help you figure out what's the right way to bring that creative idea to life on different channels, different formats for that audience. So we can much more quickly optimize content in a much more granular way. Then you have agents to help you more quickly and more precisely develop the distribution strategy that earned, own, paid, and shared distribution strategy. And then you have agents to execute too. You want people writing copy, you don't wanna create AI Slop. You also wanna take that copy and say, all right, check it to make sure it's optimized for LinkedIn, check it to make sure it's optimized for Instagram, check to make sure, especially in this day and age, it's the right content for the political and social environment and for the audience. And then we'll go execute. So you basically have these assistants helping you every single step of the way. And then of course you have ones that will help you measure and get insights from your measurements so you continue to optimize the campaign. That's a pretty typical kind of general consumer campaign. Then we have the corporate communications and reputation side of our business. So if there's an issue you're handling, again, you have your audience insights for your stakeholder groups, how is this issue as important to them? What matters, why? And then you, we have a series of agents that help you project out, forecast out of different actions by the company. We'll listen to different reactions by the stakeholder group. And we're using this with clients right now. We've had enough experience at this, I shouldn't say enough, but there's about approximately six months of real world experience after a while of developing. So you can say, this is working. This is taking out a lot of guesswork and it's adding a lot of precision for our counselors direction to clients. So a synthetic audience in the corporate side that connects to forecasting media and stakeholder reactions that then connects to once we know what the rate of action is, further optimizing the content is a symbol of a powerful workflow to improve your day-to-day reputation management. And the crisis you'll inevitably have to deal with where how accurately you deal with that crisis can make a great, not just your reputation, but the value of the company. And this is becoming the normal. We'll look back on this in a couple of years and say, well, of course, this is just how we work. But today it's a real push to get people inspired, excited, and then to make it happen. - Yes, so you mentioned the point that you want humans to write the copy and you don't wanna create AI Slope. So how do you balance the data science and AI and technology side with the creative side? And how do you find that sweet spot where you get the best of the humans and the best of the machines working together? - So I'll give you my answer, but this is still an evolving space. But the most important thing is that AI is an assistant to help you get to a better place. But unless, remember, it is basically trained on everything that's already been done. Meaning part of what AI Slope is, it can feel very generic, and people look at it and go like, I can't tell you exactly why AI Slope, but it is. It's 'cause it's almost too perfect, it's generic, and you've probably seen something like it before. It is people that come with something original. But what it can help you do is along the way, help you think of it as that, you hire that assistant from university, you've trained them in what you do, you said to them, now I want you to challenge me and help you better articulate the thinking. So I personally use this myself all the time in terms of, when I occasionally write in terms of strategic thinking, I'll say, well, here's a direction I'm going in. What are some other similar directions or how should I look at it differently? You know, give me ideas or pathways to continue thinking about it. Once I pull in that direction, and then it'll help me come up with clear and better articulations of the strategy or the message. And the same rules apply on the creative side, where one of the more powerful creative assistants was done by one of our ECDs or executive creative directors. And you took the methodology to use and develop an agent on it, and allowed them to direct an agent, come up with ideas and go back and forth with the agent to first come up with the right ideas that he could think through, or he could get seeds of ideas of what's been done before and then he can iterate off those and then go back and further refine them. So it's a difference between, I don't want to humanize AI too. I really don't want to humanize AI at all because that causes a whole other list of problems when you see headlines like AI, AI and the NIPI label. No, it didn't. But now I'm going to be hypocritical. I'm going to kind of humanize it for one second for one moment. You have to think of it the same way you always had your assistants. A great assistant will help you test ideas, will find sources to inspire new ideas and will help you articulate those ideas. A great AI agent will do the same thing. And what you would never do with the assistant, I'm sure some people have, but they shouldn't, it's just hand over, you're jog to them. Because you'll get something okay, but you'll get it okay. Because they only learn what they learn and they're just out of school and they haven't. They're still developing all the experience come up with new powerful ideas. They might want to know why the best assistants are they are learning with you. They want to learn with you and then they want to help test your thinking. And it's the same time again I don't want to humanize but I'm doing it exceptionally now because it is the same kind of workload and the same kind of relationship. You still have to work with it. And where your human assistants are helping you is they're helping you to still help you to develop those agents and manage them and test these ideas with you. But it's the same it's the same dynamic. I think that's a lot of times people forget they see the output look so perfect. They think oh I'm sorry but I'll just outsource my job to it. Okay if you want you know average work that's a probably the fastest path to producing consistently average work. If you want great work you will use it to really kind of test and find. So it's think of the idea, test it, get different perspectives on it, get articulated, hone it and then you come out of that final polished version by going back and forth. And that's a mix of both having the right agents to help you, hoping we're agents to help you, which you yourself can do and you develop that skill set. And then having the prompting skills to know how to go back and forth versus doing a lot of what's called the one shot prompting just put it in and go no it's good answer. It looks good I'll use it. But if you actually made the answer very often it's like those presentations we've all been in work nothing was really set of any substance. Just a lot of really good words put together in a nice flow. Which is what AI is designed to do. And do you see this continuing across those along those lines because I guess this is reassuring to people in the creative industry that actually they bring the creativity they work together with the AI to make them better to challenge them to to elevate what and whatever they're doing, make them more efficient. Looking 10 years into the future do you think people will still create copy themselves or do you think the AI will become better at the creative side too? I think I'll be both. Now I'm not a creative industry expert even with my own industry I've never been a creative but I can observe what's already happening is what's happened before which is something new comes along. So let's let's take high quality camera phones cameras and cameras and phones. And when they first came along with a lot of nervousness oh now anyone can take a high quality picture and shoot a video it's going to be a lot of junk. There is certainly plenty of junk up and then there was a little bit of hammering what about our production department and what about our professional photographers. I don't know the exact numbers but I do know certainly within our business we have a ton of professional production people both for video, for audio, for image, etc. Because what happened was you saw two markets emerge the overall market for visual content exploded compared to when I started this industry in the way times. And within that you have a very simple division of two types of markets. The type where people want to consume what is not the most polished content in the world but they don't think of it that way. We just call it user generated content or UGC style content from companies. That's totally acceptable and it's most appropriate in some cases. And our teams are going out and shooting picker, when I say our teams are non-production account teams are going out and going to an event and shooting pictures and shooting video and doing on the spot interviews and using those where that format format is right. But then we also have the massive market where we need very polished executive interviews or creative videos with celebrities who are creative animations. And that's the production side where you still have not just still have you have more professionals than before directing all that content. So I do see the same thing starting to happen here in its early. So look what I do exactly what happened 10 years ago. I think talking to you from the C firm or a think tank or you know but I don't think anyone really does to just speak with confidence. What I think will happen is it is starting to follow the pattern where we're seeing creatives. Again, once they embrace the technology they're starting to develop these solutions as I mentioned before. One of our more interesting creative agents was developed by an ECD. And you see this market happening where there could be a market where AI created copies such as for hyper personalized copy is fine. You actually already see that online. I see a lot of hyper personalized copy. Whereas the original copy and the original message or high stakes high impact hero content is mere seen by a person who's going to make sure every single word, every letter, every flow, every beat of what they're trying to do is as it should be. And that takes human creativity combined with what I was talking about before the some of the discipline that AI can help press on but you want it to be led by human otherwise that hero important copy will come across as generic just as if you did a CEO interview that's supposed to be the most important interview of the year but you had a non-production person do the interview it's going to come across as not appropriate UGC style just amateurish. So what those two markets look like exactly don't know but do you see the scenes that that's starting to happen today? 100%. And it's going to be interesting to track and I think the winners as professionals and as agencies will be the ones who figure out they don't style way but kind of embrace figure now with those two markets. Yeah I agree. I would love to also understand the technology the platform that you have created. You talked about the databases that you have you talked about all the different AI tools people now have access to but how does it all come together do you have a platform that allows them to access video creation, writing, image creation all from the same platform? How did you how did you choose which tools to incorporate into that platform and how does this then link to your data site? Sure so I wish I could show it I'm not sure I can but I can describe and it says I mentioned before it's built up on the comes on the platform. So the really two types of platforms this are data platform you go in you get all your insights you blog into the right tool you collect your data you might then form that into a data set that you use in your AI solution if you did. But on the AI side they've really done a terrific design that makes it easier to put all this together. So a simple walk through is I'm going in I'm going to choose the right agent that step of the workflow I will choose the right knowledge library and data set if I need one or both of those things and you can use multiple ones you might have five different knowledge libraries and five different data sets or five different synthetic audiences. So you're literally a drop-down list here's the agent I want you're all the knowledge libraries and data sets I want you're just checking buttons and then you're choosing your model I'm going to use a cloud model or a lommel model or a chat chief and t model and then there might be special tools like web browsing and powerpoint creation will have you. So you're literally going down and flipping switches on your boxes and so if your first agent was I want to get some insights on our industry I'm going to choose some industry libraries and I do it off chat GPT five and I also want to use web browsing and then I get to work what are some insights for the industry terrible prompt but just to illustrate I do that now you know what I like that fifth insight I want to draft an article on it all right now I'm going to choose my writing agent I'm going to switch to the cloud model I might switch on the library and some past writings I've done because I want to choose that style and then I draft an article again in the same workflow so that's one. Now I want to create a graphic for that article I highlight some copy that represents the graphic I move over to the graphics section of this platform I pick the graphics the image model that I want and I create an image of it you know so you might you know move over pick nano banana the latest one everyone talking about create your image find it now I have my graphic to go with the article and I think that it would be fun I'm going to I'm going to kind of make this internal little little video and have the person in the graphic actually talk to my audience but what's in the article again move over to the video section pick VO3 which does the no form of sort 2 I'm a big fan of VO3 actually big fan of both but which was a good job of creating video with voice and now I'll create a little video I can I mean period all that can be done in a couple minutes I mean if your prompts are perfect they don't need to go back before and this is all one workflow this is how people work day-to-day they're creating they're directly at themselves now I then say you know what I want to turn I have an idea for a better writing agent I can go back just click on create an agent and everyone has access that function and put in the instruction set connected to the databases and knowledge bases etc and then just quickly create my agent and then continue to refine it this is all in a single interface people can log in to to every day. So Amikam is using this across the board. Our job is to take it and make sure people have access to the right training to use it for communications. And then the right knowledge libraries and data sets and features that they would need to use as communications professionals and then just get to work. And of course, they can collaborate together in this way as well. We're creating protected client environments. They can all work as a client team or they can collaborate to develop solutions. But to be able to, you know, what, you know, if you're not, if you don't have a platform like this, your ability to create an agent is limited to, have you, do you have access to your premium platform? And your ability to use all these tools in a single workflow, you're using many different applications for a single workflow. That slows down work and that dilutes the value of that work. So that's how it works. Flip the switches before you want. Take your step in your workflow. Now you're ready for the next step. Choose your new agent knowledge library, models, et cetera. Take the next step and so on and so forth. And the description sounds simple. It's not that much more complicated if that all is more than described. - So how do you choose which tools you make available through this platform? Because they're, and how do you make sure that this data is all safe? That these are all enterprise sandbox, as sandboxed versions of those tools. So no data can be used to train future models. You can't escape somewhere else. Are you making access to, are you allowing people to use tools like deep seek, for example? - No, how do you make those choices? - Yeah, you have to like, people do ask about deep seek a lot and the straight answer is, look, if you want to learn those on your own time, great, when it comes to work, it's a hard now for both security and training purposes. Models do not learn off our work, period end of the story. We are using the models, we are not training them. We'll stop. And the security is the kind of security, and to be very high level without going to detail, but it is all kept within that kind of highly secure environment. We've always had as an agency because we deal with so many sensitive issues. It was an interesting challenge when all this first broke out, and the chat cheap key was launched and everyone got excited. Everyone started using it right away, and what they got was a, you know, talk to the hand. Everyone's stuck. There was a lot of work put in by Omni-Common its agencies to make sure security, governance, legal challenges, all that was addressed first. And then as soon as it was addressed, it was, okay, now everyone, go fast, which was a little bit of a win-winch. So it was a very, and certainly within inflation and hillard, where between the consumer and product campaigns we work with, and then the corporate reputation issues we work on, there's always been a heightened sensitivity. I only have to, and it's really, in all seriousness, that we need to live our life under a non-disclosure agreement, just assume we're working on a sensitive, and you have to have that fear of technology environment as well. So this is all built in a highly secure environment, and everything we do stays in that environment. It does not go out and train, and that is important, certainly, to our clients and to us. And so when something new comes out, what we do, I certainly encourage people to learn on their own time, these new tools what they're doing, but they also have to remember, if you are not logging in to the Omni system, to one of the Omni components, like Omni AI, to the AI platform, do not do work, that's really important. But that's just habit. For those who remember the launch of Google Images, I remember when grabbing images and putting in presentations, we had to say, like, full stop, do not ever use Google Images for work, only get images from your creative team, or if you have access to a licensed business library, public resources, not good company resources, good. And it's just giving people that they have the sense of responsibility, that I'm not concerned about, just to acknowledge that, okay, if I go in through this door, I'm good for work, and if I go in through this door, that's personal, give the two separate highly secure environment, no training, you can't stress those two things. - Yeah, it's good. And yeah, I think you've approached this really well. You've focused on the training, you've taken time to create the right platform, chose the right tools. What lessons would you share? What are some of the things that you got right, or maybe some of the things that you got wrong? Anyone who is now on that journey might be starting that journey, what advice would you have for them? - Well, I'm looking, you have to be fortunate, being part of the PAM company that has provided these resources. I'm gonna generalize their philosophy, which is they talk about who on all our agencies have the gold standard data in tech. Well, we then took that a step further and said, great, we want all our people to have direct access to that. So you have to first decide, what are the, well, you have to decide what do you want? What is your strategy, for example? Is your strategy to have a couple of heroes solutions? Or is your strategy to lift the entire enterprise and people often confuse the two? Lifting the entire enterprise, which results, our belief is in far better solutions, because you have everyone who really knows the business, figuring out what's right for the market. There are two different things, 'cause you apply resources differently. You apply them, if you're a central hero solution-spotus, you apply them to generally a small R&D type group of people. You get them working, you develop the solution, and you train everyone how to use that solution. So, for people getting started, obviously we advocate for our approach, which is, first of all, you want people to learn how to do this and be inspired. So the first thing you do is you need to take a step back and take a deep breath. Don't get carried away with shining your objects, 'cause they just learn the basics. So once you've learned the basics, that will start coming, that'll help you start coming with ideas of what your future may hold. So when you start coming with ideas, now you're starting to shape those into, what is your product or your service strategy? What are people gonna look like and how do we market them? What are solutions gonna look like, which are the most effective and how do we market those? But having that come out of a bottom-up type innovation approach. What we're seeing is we kind of, maybe we started a little behind in the first solutions being out to the market. But then we're seeing now as a quick catch up and we're gonna have more solutions that are more powerful 'cause they were really built by people at North clients know their challenges and are much more spoke to that. So it's, I'm not sure if I'm answering the question directly. I just still come back to learn the basics, learn the fundamentals first. And then based on that, you can have a much more productive discussion, a far more impactful discussion on what's really gonna matter, 'cause you know what you're talking about. You're not talking in theory. The other thing you need to do, that the other learning curve that's going on is you have to keep track of how the media ecosystem is changing. So you can match the solution that towards how the media ecosystem worked two years ago, but how it's gonna work tomorrow. For example, between models like VO3 and SOAR2, it is very possible that users will do to Netflix, what Netflix did to Hollywood, which is all of a sudden it felt like, oh we're gonna, you know, Hollywood, but now again, there's enough room for everyone. So I'm not, you know, you can figure that out. People can create what seem like Hollywood level production, but truly UGC in a couple minutes. And I know it's limited to, you know, is it eight seconds on VO3 or a minute on SOAR2? But imagine a couple of years from now, you know, Bernard, you can say like, I have an idea for an article. I'm gonna make a short documentary out of it. And that documentary, I'm just gonna load the idea for the article and produce that documentary in 10 minutes on VO3. I'm exaggerating only in that you'll be able to do it. It just might be crap. People take more than 10 minutes to make it quality, but you will be able to do it. That has a lot of implications for the expertise and the capabilities. So I learning both at the same time, people also better prepare for those shifts. But it goes back to like, my vice-versa-one is, get hands on to develop those basic skill sets, then start thinking creatively about what are all the solutions for the challenges today or tomorrow. And then start figuring out how those solutions and so solutions are often full workflows become what are essentially your company's new office. You know, so now you have your, this is our end-to-end media relation solution or our end-to-end consumer peer campaign solution or end-to-end reputation management solution. And it's a series of agents or an agentic AI powered workflow that makes everything more accurate, more powerful and is constantly adjusted around the media. Because it still goes back to got one of the basics first, otherwise it's all just theory. I hope to answer the question I know like that. - It did, yeah. It did because you covered so many things that you did already in this conversation, absolutely. Looking ahead then, maybe five or 10 years, what do you think would define a successful brand in terms of communication and digital presence? So I know we talk a lot about trust and authenticity. This is one of the tougher ones because the idea that people don't trust, you know, that don't trust media, for example. When you see all these studies, people don't trust media. With all due respect to some colleagues who might say that, I think it's just patently fault. It oversimplifies the complex situation for brand, which results in them following the wrong strategy. Trust has been fragment. So people trust the media ecosystem they choose to trust. There are a lot of societal problems as a result of that. We have to first understand that people do trust media. Trust media trust, but they may not all trust a single source. They go back 50 years and everyone, I assume I haven't seen the stats, they generally trusted the evening news, where they trusted their local newspaper as a whole community. Whereas today within that same community, you could have people, you could break up that community into an activity of 1,000 people, you could break it up into 10 or maybe 1,000 trusted ecosystems. So it's probably fragment. And so for a brand, that means you have to do two things that's really hard for people who grew up in the era where there is much more kind of blanket trust. When you have to look at it from your audience's perspective. So brands are defined by how are they, or brand trust, by how are they relating to their audiences, how are they communicating to their audiences, and other doing it in a way that their audiences can relate to, and through channels their audience is trust, and that their audiences find relevant. And that's a really tough thing to do because it's constantly changing. So I think we'll see in 10 years is, I would love to say, we'll be back at the old days where new central powerhouse media organizations emerge and the central trusts and they're all doing great reporting, and there are no current signs of that. I hope it happens, but I'm not seeing it any great green shoots. I think it'll be continued evolution of what we've seen for the last 10 years, which is, as a brand, you have to figure out, where are the sources my audience is going to, that they're treating as news and information sources. I may not consider them as an individual professional news and information sources, but my audience doesn't care what I think. And then I have to look at them to know which one of those sources lines up with what a brand represents. It's just because an audience trusts a source doesn't mean as a brand you should go working with them. That can have a big backlash because, well, we've seen a lot of this cases in the last and last, and in her people have multiple audiences, multiple stakeholder groups. So I think the only difference now in 10 years is there's probably a little more fragmentation. You could have an audience of a thousand people, which have a thousand different trusted ecosystems. You have to look for the common trusted channels and sources that line up with your brand. And then the change will be even faster because now the sources that are in that ecosystem may change over and, well, we already see that today. You may have found an influencer that communicates news about your company or related issues that your audience trusts and then you feel lines up with your brand values and you work with them. But also, overnight, literally, another influencer pops up and your audience starts turning to them. You're ready for that fast shift. That's only going to the current signs are that will exaggerate. That will continue over the next 10 years. The other part of that is that a lot of who you're going to work with as a brand are not people, but they are always say, you know, the last couple years we said, the algorithm, the media platforms, this is an editor you need to pitch to. But the LLMs are you developing, and this is part of, you know, the geo offerings of today is understanding how are the LLMs, how are the large language models behind chat, GPT, the co-pilot and Gemini, how are they telling your story? Now how do you create information that essentially feeds them? So they tell your story in the most accurate way. And these are all going to shape how people look at brands. So those are kind of two fundamentals. How do you work with the LLMs? How do you work with the robots, so to speak? And then how do you work with a constantly changing evolution of people? Now the other thing we don't know yet, but again, just assume it's going to be radically different as what's the content that will shape your brand. So for example, going back to the SOAR 2 or the O3 model, if anyone in your audience, whether it is a policymakers or someone who's buying your product or a prospective employer or investor, if anyone can create Hollywood level content in terms of quality and is happy to consume that type of content. And how does that change how your brand shows up in the media ecosystem? And there'll be more drastic changes like that in terms of the quality and the types of content are created. For example, people might become used to seeing synthetic or virtual influence. You might have 100 Bernard Mars in the future interacting with different audiences in them. And today people might go, "Oh, I don't want that. I just want to hear from Bernard." They said the same thing about user-generated content 20 years ago. That just looks like crappy content. But that's now a big part of their entertainment ecosystem. To say a thing, 10 years from now, if not sooner, people might find it acceptable. As long as I know this key from Bernard reflects his thinking, "I'm good to talk to a virtual Bernard," he gets some advice. I'm good to have a virtual Bernard or on-cat. You might just break up a whole new persona and have a staff of personas. These are all possibilities that people don't know. People don't like that. People don't like a lot of things they've proceeded. It takes getting used to. That's also going to shape your brands. You have all these different sources. All these different forces impacting how your brand is going to have to show up in 10 years. The challenge for brands is the greed shoots of all that change are happening today. So it might have taken 10 years a while back and they only take a year or two today. That's where that hands-on experimentation comes in. So it does happen much sooner than expected. By next year, Hollywood level UGC is dominating and there are good signs that it will. If by next year people are just fine with virtual influencers, virtual personas, etc. If by next year you have a 90% refresh of all the influencer sources by getting ready today, you can be ready for that change because again, that cycle is happening. So that's it. Fascinating. Thank you so much, if I am that way super, super interesting. For anyone who ever wants to rewatch this, you can simply go to any YouTube channel or any podcast platform where you can watch this one in hundreds of other conversations like it. Thank you very much. Thank you. As always, it's a pleasure and as I said, when we met in person a couple of weeks ago in today, these conversations always get me thinking and I just appreciate the opportunity. Thank you. Yeah, likewise. Thank you.