Welcome to the Marketing AI SparkCast, where AI meets modern marketing. Hosted by Aby Varma, a global leader in marketing, AI, and innovation, each episode brings you real conversations with leaders putting AI to work in strategy and execution. This episode is brought to you by Spark Novus and the Marketing AI Pulse community. Here's your host, Aby Varma. Welcome to another episode of the Marketing AI SparkCast. Super excited to talk about AI adoption in enterprise organizations today. And we have an amazing guest joining us today, David Kuoch, who is the vice president and AI evangelist at the Citigroup joining us. David, welcome to the show. Thanks for having me, Aby. Appreciate it. Yeah. You're welcome. I think this is such a fascinating topic, the wild ride of AI adoption that we've all been in. Think I it all comes with different sets of challenges for different size of organizations, but most of which for enterprise organizations where everybody wants to do it, but things are locked down. And there is a lot of considerations, right or wrong, when it comes to AI adoption. So we're gonna unpack all of that, David, with you. But for our listeners, would love for you to give a little background about yourself and how you are in this current role. Yeah. So I I sit at at a very interesting point where, obviously, with our organization, we're we're not a small shop. There's about 200 twin... 220,000 people over here at Citi, and by all means, they're not a small small shop. Where I sit is is kind of unique where my background has been in project and program management, and I'm working with the engineering team to actually solve complex problems with AI. That's one thing. So embedded in what's happening behind the scenes. And then the address... The other perspective is where I sit is also the data governance. Right? And and looking at sort of the data, how it operates within Citi and also as an AI evangelist, teaching the rest of the organization how to fish, so to speak. Right? You know? When I say myself as the AI evangelist is is basically I work with all the accelerators within the organization. And within my organization, we have about 15,000 people just in in enterprise data operations. So it's a big group, and I sit within that space to educate not only enterprise data operations, but the rest of Citi as well. So, but it definitely is an interesting ride, and it's a lot more complicated when you have a big organization. And I'm kinda jealous of the smaller organizations that can move fast and and break stuff. We, unfortunately, can't break anything. We have to live within the sort of the governance framework within, you know, a financial institution. Perfect. How long have you been with Citi, and did you apply for this role? Did you craft it yourself? Just talk to us a little bit about how you laid it out. The reason I joined Citi is before the the big sort of AI, I knew that it was kinda coming. There was people kinda talking about what they were doing behind the scenes. You look at financial institutions. It's like they're kind of the first ones to get it as far as, like... Because there's a lot of data that's involved with the financial institutions. I joined knowing that they're gonna be opening up the sort of ramp, so to speak. So I got in an interesting time, started working with the, the data team on the governance side, and then when they opened up the tools, it just happened. It just evolved. We opened up the tools and let people kinda go at it, and I created a brand for myself within Citi. I presented to over about 5,000 people within Citi. I've done 50 plus demos within just the last year, across the organization. And it's been an interesting ride, and not only just for our team, but the rest of the organization as well because we're trying to move the meet... Needle in terms of, AI adoption and also behind the scenes of all the work that's been done. You know, hundreds of teams, kinda doing exciting things that I can't speak speak to, but, it's happening behind the scenes. Awesome. How do you evangelize AI in an enterprise organization, especially that size as Yeah. It's interesting. When you talk about AI evangelism, the way that I look at it is is there's different levels of skills that are involved within the organization. Everyone from the very beginning to people that are on the engineering side that are doing really complicated things. So the way I look at it is raise the floor and then also raise the ceiling. Right? That's how I kind of sort of look at the way that that I'm doing. The AI enabled network is kind of raising the floor, kind of bringing people to the level where they can understand what's kind of going on. You know? And then also the other part is, I think that you called it up, is is what's the business value? Those... The two different things. They kind of coexist in the same space, but, obviously, when you're looking at AI, it's one focus that is is a lot of, like, experimentation and getting people to to to have a certain level of skills that brings them up to the level. And then there's the business value. How do we deliver business value to the organization? Right? Moving from sort of POCs to something that's scalable over long term. So that's kind of the lens I look at. And whenever I can, I try to, like I said, lift my? Lift those people that are getting involved with it, and I'm also looking at the business value that's being delivered on the back end for our company as well. Love it. So one of the things that we often see when we are working with enterprise organizations that what sticks is how can you make AI practical. So it's not like pie in the sky stuff, obviously, attached to business value, but how do you make it practical? So what are you talking about, just the first part? The... It's it's a journey. It's one of those things that doesn't happen overnight. It takes a long time. Even in a large organization, we have our struggles as well, and it's still a constant journey. AI adoption. We look at sort of AI enablement transformation adoption. That's the big thing that that people are talking about. When I look at it, it's it's not just the access, but access to it. People can have the tools, but it doesn't necessarily mean it's gonna be adopted. Like, we we think of, like, you're gonna give people these tools, and they're gonna adopt it right away. Not necessarily. That's almost the exact opposite because as you know, there is this inherent fear where it's like, hey. Is AI gonna replace people? And I'm I'm in the minority, and I say, hey. AI is gonna replace the people that don't know how to use AI. So what I tell people is like, would you rather be on the person that's on the front lines learning and and getting yourself... Getting hands... Or, you know, getting hands dirty in terms of, like, learning all the different tools that are associated with it. So that's one aspect of it is. The other aspect of it is is you look at when you release these tools out, it's it's... You gotta give people some structure. Where do you look at your work at day to day and where there's opportunity to kind of improve it? Right? The manual stuff, the the low hanging fruit, the... All the things that you're doing, manually day to day. That's the opportunity where you can actually bring AI. You know? And it's it's in many different ways from different parts of the organization. Maybe you're marketing. Maybe you're in compliance. Maybe you're in finance. But look at your day to day things that you're doing and where you see opportunities, and that's where you start. So in large organizations, when you go about solving it, so what is the approach that you take? Do you have a grand plan? Do you break it down into smaller pieces? There's obviously a huge organization. But in general, your approach, is it tied to specific use cases that each department or each of the teams want? I'm just wondering. It is a it is a tall ask to evangelize a new piece of technology with heavy governance footprint in a regulated industry for a large team. So just wondering that it's a lot of... That's a kind of a complex multilayer sandwich right there. How do you how do you bite into that? Yeah. So going back to what you're saying, I think the first thing is when you release the tool, whatever tool it is, depending on what you get as an organization. Well, you know, we have multiple ones, but maybe you just have one. Maybe you have multiple ones. You start off with the with the education. I think the first part is, like, the big core thing. Like, everyone starts at this sort of, like, low point where it's like, hey. You're experimenting. You're kinda testing things out. That's where you start off with. And there's 10 sort of principles I kind of go look at as far as, like, AI enablement approach where I think education is really important. That's, like, get everyone on board. Let's focus on that right now, and then I'll talk about the other thing. But the first thing is, like, basically, you know, as I mentioned, it's it's it's it's... You give people access access to it. Doesn't necessarily mean people are gonna adopt it right away. But what you do is you basically sort of say, hey. This is what I'm doing. This is what I'm experimenting you. And for me, what's always been really... It grabs people is you show how it works. Like, not just kind of like... I'm not a big fan of PowerPoint slides, which, you know, big organization that really do. What I like to do is when I sit on these demos, I'm gonna show them exactly the work that needs to be done. I'll take an example of, like, what someone's actually doing in a day to work. Let's go ahead and do it right near... Right right here in the spot. I'm gonna have you do it as well, and we can show people like, hey. What is it? It's it's one thing to do it, but it's it's it's... Means a lot more when it's when it's applicable to your work, and it's resonates a little bit more when it's like, oh, I'm I'm just not doing someone else's work. I'm actually applying it to my own work. I think that's the real value of it. You know, doing what counts and what's valuable to you and your organization. You know, some of the principles I talk about within, sort of, like, the AI enablement framework and and... Is is basically these big principles. Like, I think it's a bit missed point where, we talk about, like, AI. Think about sort of, like, what the desired outcomes is. That's number one. It's, like, focus on what your your your desired outcome is, and I like to call that is around the problem framing. It's a really important thing. It's one of the critical things. I think we start off with something that we we have these ideas, and we we we wanna kind of POC it and stuff like that, but focus on what your desired outcome. Maybe it's a report. Maybe it's some kind of metric that you're working on. Start off with that one. The other thing I think is really important is around prop engineering. It's one of those things where it's like, you know, think about AI. AI is not difficult from any kind of coding. You base... You basically have to teach AI how to how to design. Right? So prop engineering is really important. And one of my favorites is that I use CO-STAR, which is the, you know, context role, objective style, tone, and and response. If you frame that, you start teaching people how to do that, then they get better response. Because when I look at some of the things that's happening within organization is junk that comes in, you get junk out. So a lot of times people don't know how to prompt engineering, and so that's an important part of kind of designing designing the way that you're you're teaching the organization. Process decomposition, basically turning something that's messy into something that's structured, making sure that it's... It can be scaled over a long period of time. Data interpretation, that's a big part of it. Looking at the data that you actually have. And AI does a great job of looking at the data. But what if you wanna basically give some intelligence to it? It's more important when I'm looking at something that's that's static data. I want dynamic data. So you use AI to interpret that for you. Look at look at at areas of opportunity where you can ingest AI into your workflow from end to end, especially going back to what I was saying is the the the low hanging fruit, the the areas where it's very time consuming. It's it's like it takes a lot of time. Maybe that's where you need to to put the AI into into the process. Orchestra So all those different things. And there's there's the governance. There's sort of the human in the loop part, which which drives me kinda crazy because if I hear that again, I'm gonna I'm gonna strangle that... Strangle some people because it's only one of things where it's it's it's understood. Like, it... It's already embedded in it, and it's important to just gotta tell people, you know, AI is an advisory tool. Right? You know, you're never gonna rely on automation to kinda do it because that's not what AI is meant to be. It's meant to be your assistant, your copilot to sort of get you to do 60 to to 80% of the work. And then it's you as a person that... That subject matter expert to come in and sort of say, yep. I'm gonna validate that that information is correct. This is correct. That's the that's the beauty of it because it does 60 to 80% of the work, And you as the human that's coming in there to validate the work, you're never gonna be... I mean, that's an important aspect of it. Even what we do here within the organization, it's never automated. There's always someone that's actually looking and validating the information. Everything that we put in in our processes, there's there's checks and balances in in in the organization to make sure that that we don't rely solely on AI to make decisions for it. So, anyways, lots of kinda to unpack, but that's what I'm talking about to to people about AI and sort of how we move it. That... That's sort of the framework on on the education that we actually have within the organization. Cool. Now one of the things that I've seen that enterprise organizations struggle with, and you alluded to it a little bit, is scaling. Right? And when you're scaling AI, there has to be some sort of an element of standardization, but the standardization is double edged sword because you want certain things standard, but you also wanna give people latitude on how they wanna use AI in the context of the job that they're doing. Like you said, you know, people are gonna use it. If it's... If I'm using AI in the context of what I'm doing and you are using it in the context of what you're doing, we are more likely to use it in our day to day work as opposed to some sort of a standardized way that is too stringent or too strict and doesn't allow for that flexibility. So if you could, it's sort of a two part question. So one is, how do you achieve standardization for an enterprise org while providing flexibility? And then the the second part of it is the the... With that framework in mind, achieve scaling so it is sort of applied in a standard way across a large population of enterprise workers? So it is... It... It's it's it's those two things kinda have to coexist. Right? And I know that, especially in in organizations, one of the first things that will kind of organically come up is peep... Once people start playing with the tools, they'll be like, hey. I wanna design a prompt library, which is, I think, is a good approach because it get... It sort of sets the sort of, like, the the ground floor. Right? What what type of prompts are being used, how you can use it. You know, maybe Sally or John can kind of share that. So you're building that from the very beginning. Where it gets a bit tricky is when you start to to do that in a large organization is you have people building their own library specific to their groups. So it's kind of like, you know It... They're kinda silo in a certain way, but it's a... If you think about it, it's the same design. It's a modular design regardless of who... Who's doing it. The problem becomes is that you have... When you open up a tool and you have, you know, thousands and thousands of people, those thousands of people are are basically recreating the wheel. Right? So it's kind of like it's great because on one side, you're you're raising the floor and people are getting comfortable with AI. But on the other side, it's hard to get that consistency. And and that's gonna be applicable for any that's in there, whether it's Copilot, Gemini, or Claude. You have this... The whole organization kind of, like, testing it out and playing with the tools and getting those results that they want, which is wonderful because you're like, oh, yeah. People are doing it. You know, when that happens, it's it's kind of like... It's great that it's happening because it's like it's raising the floor, but where it becomes problematic is like, okay. So now we got a bunch of people doing it. How do we get consistency? And that's the the mindset shift when at some point, an organization needs to sort of say, okay. We're mature enough to have these prompt libraries, but we need to kind of, like, shift it to sort of, like, how I like to call it is is sort of operating models. And it's it's it's the idea of building sort of playbook because... Or intelligence level on it. Because if you look at it, you have these prompt libraries that are massive, and people are... You know, different silos are kinda using it. All the same ideas, but they're they're duplicated hundreds and hundreds of times, and you're not gonna get consistency where at some point, you have to say to the organization, we need to we need to shift the mindset and think about playbooks, intelligence level, operating models. And that's when you get sort of, like, the ability to sort of scale things and make it reproduce. Because the way that you approach it now, if you're looking at operating models, you can get consistency, whether it's Sally, John, Tom, whoever's using that tool and they're using the same playbook, now they can run the same report. Now they can get consistency. Now it's like, oh, we we can actually automate it, you know, where you're you're moving from that that floor part, which I was talking about. Now you're getting a lot more business value, and you've kinda shifted the mindset and say, oh, okay. We can actually incorporate this into our work, and now it's delivering value. We're getting consistency. So it is sort of like you start off with one, and then when you start getting more mature, then the... I think the organization needs to take a step back and say, hey. We need to look at this as sort of operating models and playbooks, and let's design for that one. And that's when you have people sort of gathering together to sort of like, hey. Let's look at all the... Where there is sort of, like, replication across across different different libraries or not. And you get that consistent where you're like, hey. Rather than having 10, we all agree upon that we're gonna actually have one model that we're gonna use for everyone. We can have... We can tweak it and do it for specific teams, but now it's, like, one thing that can be applied to... For across the organization. That's what I mean about about delivering the business value, and I think that's an important part of it when a when a company matures. Yep. Love it. I think the... With playbooks, 100%, when we work with CMOs and our marketing clients, we see that a lot where, you know, when people are in a stage of experimentation of a sort of the wild wild west, everybody's doing their own thing when they're experimenting it. But then the moment you wanna start using it in real life sort of production scenarios, there has to be some some methodology, aka playbook or operating system that is used. And I feel that strikes the balance between, you know, having that framework from an org standpoint within the org so everybody is not doing things completely differently, but also provides the freedom and the flexibility of people to adopt that in the context of their job, their department, their role, and that sort of thing. And so as a follow-up to that, do you see as these playbooks and operating models that you guys sort of deploy internally for evangelizing an adoption of AI within various teams, Is there some sort of a feedback loop that teams are giving you feedback and saying, hey. This is not working, or maybe there's... We need to tweak some of these models. These are not flexible. I'm just wondering, are those models and playbooks living, breathing creatures that get updated based on feedback that you're hearing from your teams or employees? You know what? It's always hard to sort of design for a large organization. It's kind of like the the idea which is, like, you can't make everyone happy. You know? It's like it... It's it's such a difficult thing because when you're designing... How I look at it is, like, I wanna get as many people as I can. Right? And that... That's the funnel I'm looking at is because if you look at it is... Like I said, it's very... It's all the same. However way you wanna wanna do it, it's all the same. So you look at the largest and sort of say, how can I scale this across the organization where it's it's appable and it's just modular so that compliance and risk is taking the same things, and they're just tweaking a little bit? Maybe they're they're adding their tone. Maybe they're... They have different policies. Maybe they're having different procedures and stuff like that. But it's all modularized where it's like, okay. This is specific for that team. So when I approach that, I'd look at the biggest funnel, see how many people that I can get within that funnel. And then at that point, the feedback loop is like, I wanna tweak it. But then you're gonna go one layer under that and saying, this is just for this this part of this organization, so it requires a bit more customization. And you work with those team members to get feedback, but you don't... You know, having version control around it is really important because you're trying to satisfy, like, this bigger group. And then from there, you're working to kinda create sort of iterations of it so they're using the same sort of framework. That way, the next time someone looks at it, it's like, oh, I'm very familiar with that. You know, we start out with this. We know what this iteration's gonna be, and this iteration can now be modularized for a different group because it's a different type of, things that they want, but very similar. Does that make sense? It sure does. I mean, to me, I think that's amazing because then people don't feel hamstrung in doing it in some sort of a a very specific rigid way, and then there's definitely latitude and flexibility. And I think that sticks. Right? Because I've seen that too many times when you sort of force teams. It's a square peg round hole type situation. You force teams in it. Those things are never sustainable. They'll do it, but then if it doesn't allow for the flexibility that their org or their workflow or their department needs, it sort of dies early death in a lot of ways. It it it does. And and you'd be amazed how how... Especially when the large organization... How much is that duplication's happening where, you know, I... I've sat in meetings where it's like, oh, Joe, something is working. I was like, yeah. We did that six months ago. Did you know that? And it'd be like, oh my gosh. No one's talking to each other. And it's... They're... We have all the same ideas, and you realize six months into it, like, oh, yeah. Someone's already six months ahead of you. So why don't you just work on their designs? Oh, yeah. It's also hard when you're... When a big organization because the onus is we have a lot of these people that are sort of saying, we wanna do this specifically for our org. And whereas, like, I'm always thinking we have to be agnostic to it. We should help everyone. Sometimes politics plays into it where it's like, you gotta help the team that you belong in, but whenever you can, you throw out a bone to the other teams and say, hey. I'm gonna give you a head start. Here, take this and run with it. So that's part of the evangelism as well because I look at it as even though I operate within a certain team, when people are asking questions, I'm like, oh, yeah. Have you thought about this? Have you... You wanna make yourself as helpful as possible to other people because, you know, there's gonna be times when you're stuck. You're like, hey, guys. I need a hand on testing or I need stuff, like, being done. Can you... The people that you reach out to the broader community are, like, gonna throw their support to help you when you're, like, doing it on your own. And it's a collaborative effort regardless what teams that you're on. You're all approaching the same lenses. Like, you want the organization to grow, mature. And so that's an important aspect of it is helping as many people as you can and being patient. I'm not the most patient person at times, so it is when people ask you stupid questions, you're like, oh my god. We talked about that, like, a year ago. Why is this question coming up? So it... There's times when I have to bite my tongue and... Oh my gosh. But that's why I have to take myself off of the camera so my facial expression says everything. Yeah. You do what you can. Of course. And, like, I mean, also, the larger the organization, there's obviously influx of newer employees. Right? Like, where they may not be privy to or, I don't know, as part of their onboarding process or whatever. How do you approach that, and how do you recommend listeners who are part of enterprise organizations think of it? The good thing is there's always... Within within the group that we're in, There's always teams that are driving and educating, like, on a weekly basis, on daily basis. Things come up. We have channels that we talk about where we're open to kind of exchange ideas and so forth. So I think that's important part of it to having visibility within the core group of people that are the accelerators that are... The community that is actually supporting the rest of the organization. Just open communications, seeing what's what's the newest thing, what's... What what is one LLMM is doing over another one, What's a preferred one? What are some techniques that people are trying? When people have that and they have the moment, they're sharing with the rest of the organization who is sharing to the rest of the organization who are new to AI. I think that's been good support network to actually have that community of... We have about 4,000 people that I'm part of where, yes, things coming in. Hey. This is what's happening. This is what we're recording that we're sending out to those people so those people can share out with the rest of the organization as well. That's an important part of it is even though it gets repeated over and over, it's like, eventually, it will get down to the next person because I think in a smaller organization, you have the luxury of... It happens. Everyone knows it. But with us, you might be thinking that this... We talked about this six months ago, and someone will be like, oh, yeah. This is the first time I've heard about it. So you're like, okay. Do anything beyond. Yeah. And and it's hard because it is really hard because we're we're running at a pace where we're trying to like, hey. What's the newest and what do we can do? And especially all the new tools, what we're tweaking, and... But at the same time, you always have these people that are coming in that are... Haven't even used the tool after you... Even after a year. And they're seeing, oh, yeah. We're just using this for the first time, and they're testing it. So it's... Like I said, it's being patient with those group as well that they're... That are coming in that's new even though we've been doing for over a year and a half. So... Yeah. So, David, we've spoken about business value quite a bit in the early part of this podcast. And one of the things that we notice in organizations at most times when teams start playing their early stages of maturity for LLM adoption, now I'm seeing more and more is where you have internal champions, aka evangelists you could call where the focus is very specific use cases that are much more mature. So they're sort of skipping the early part, getting your hair dirty experimentation burdens, focusing on more mature use cases. I'm just wondering what your experience has been in in that business value. As organizations chase business value, that initial learning and joy in getting your hands dirty and kinda building those capabilities for employees, how does that relate to striving for business value and skipping that early part? What do you guys see, and what is your recommendations for enterprise organizations and how they should approach that with their employees? Now that is a tricky one because the interesting thing is I think everyone should go to the growing pains. Because if you don't experiment, you don't go to the growing pains of, like, what it is. You don't really see... Like, it's hard to kinda skip when you have to kind of go through it. I went through it as well, you know, as part of my experimentation. It took me 40, I think, 40 tries to get something right. But once I learned from that, I was like, oh, I get it now. I know how it operates. So sometimes it's... If you do skip it, it almost handicaps, like, the organization. I think it's healthy for organization to struggle and get people to do that. Because at some point, like, I'll I'll go back to this... The second question you have. It's important for people to understand and see, like, the basic stuff because then they get comfortable with it. If you just skip and go to the other part, then there's this big gap where people don't understand how it works. Right? I think having that... The struggle at the very beginning where you get people onboarded is a healthy thing so that they... People could have their own personal experience. Now the other part of the question that you actually have is the second part, the business value. From my own personal experience, what I've learned is, like, you focus on one desired outcome. Whatever the thing that you want with, focus on that one. Try not to do too many things. If you're a small company, if there's something that you're selling, focus on that one first. Like, work... Like I mentioned, work on the... With desired outcomes. If you think about that one, that should be the primary focus first because that's the output that you want. And if you keep core... To your core principles on whatever you wanna deliver now and work backwards, almost... It's almost like reverse engineering. Right? You focus on what it is. Let's say, I'll just say, I wanna report. Right? This is a report that I want. It has to be this way. And I laugh because this is the scenario that happens a lot. It's gotta look like this. It's gotta be this. Okay. So that's your desired output. Great. We can work backwards from it. I don't need to know everything else because we can work backwards, and that's the same principle. If you're in marketing and you're selling to a customer, focus on that thing and work backwards. The one thing I would say was my own learn... Learning experiences. Think about... As you're working backwards, think of all the potential that you can actually do it. Maybe do a session of all the different ideas and concepts that kinda come from it. Right? And then you go back to the front part where it's like, okay. So now we know what the desired outcomes is. When we first approach, what's the first ingestion that we actually have? Think of a... You don't wanna have different scenarios of all these different things. You want one one thing that leads to the next step that leads to the next step. It's almost like an orchestration or a workflow. Right? But you keep it very tight in a certain way where it's like, oh, I'm designing for what that desired outcomes, so I want the metadata. And I think that's an important aspect which I think it gets overlooked. When you take AI in for the first time, it's important to actually have that structured data so that you can use that structured data for something else. And I'll use, for example, when we we take something in, I wanna always build some, a structured database on there, whatever I'm taking in. Now I can extract it for a PowerPoint slide. I can extract it for for an executive summary or whatever it is. But that core data that that is part of my output in the first beginning is so essential because now I don't have to reprocess it again. I take that core, and I say, okay. I'm gonna take these things, and I'm gonna spin it off from all these different things. So if... Does that make sense as far as, like, when you're thinking about it? Start off with desired outcomes. The first time you take it through, think about what the structured data, the core intelligence is gonna be, and then you can weave yourself and build this this really great orchestration workflow that you actually have. But you wanna make sure that the second part includes that that core data that is part of the output that you actually have. Absolutely. So a lot of lessons that you are talking about here, which you incorporate in in the work that you're doing at your organization. But do you think the same principles could apply across company sizes? Do you think mid market companies and SMBs? And I would love to get your take on which of these ideas are extendable. And, obviously, there's advantages and disadvantages enterprise market versus SMBs, like speed, like you were mentioning at the beginning, bureaucracy, and those sort of things. Yeah. I think with a larger organization, it's a lot more complex. We have certain things that we have to go through in order to get the buy in, especially if it's something that's gonna be con... Done consistently. It's actually a luxury for some of the smaller organization because you guys can move fast fast and iterate fast. Right? You don't have the same sort of rules that apply for a larger organization. So you can iterate and test things out due to the proof of concept and move to the next thing. What I would say is think about the governance around there, the guardrails that you're putting in place. It's an important thing to think about when you're doing that. You can move fast, but you also wanna have those guardrails in place to make sure that when you're building out those models later on, that you have those guardrails in place to make sure that you're not overstepping and making decision and letting AI make those decisions for you. So it's very healthy to have thinking, go fast, build it, break it as much as you can, but also keep into consideration that there's a certain guardrail and governance that you have to design for. Now as far as the smaller companies goes, it's like... It... Like I said, you guys have a luxury. Like, small businesses have a luxury to iterate and move fast. Also, with the organization, you can scale that organization up faster. Right? It's easier to teach 30 people AI skills than 200,000 people AI skills. It's a bigger challenge, and people are gonna move at different speeds. I think having the sessions where you basically have demos, brother, for your team is really important. When someone's learning something, share it out. Create a chat where teams can share information. And here's my thing is, like, I always share, like, the designs I actually have because it's always... It's nothing... You know what? Someone can build something better from what you actually have. So with every presentation you have, AI is not a secret. Share what you actually have. Tell people what it is because they're gonna build... It's like where it says, like, AI is making pizza. Right? Gotta repeat the recipe. You share it with someone else. Someone's gonna throw some anchovies. Someone should throw some pineapple, but hopefully not. The Italians will hate you if you throw pineapple. Like, it's it's like building a better recipe for what it is, and that's the wonderful thing about it. It's like it's never hidden in a box. It's always shared with people. And like I said, going back to prompt engineering, that's really important. Thinking about how to design for desired outcomes. All those things. I think within a smaller company, you have the luxury of moving fast, which is awesome because you can break things, and it's okay to break things. It's it's it's a wonderful thing when you you design something and it doesn't work. Great. You've got that lesson. Now you can just design for something else. That's the wonderful thing about having those smaller companies. You guys can move a ton faster than we can. Yeah. As we wrap up the podcast, your final question, there are so many innovations that are happening. I'm just wondering what is your take, and what are you most excited about as you look at the next year or three years? I think overall, we're really at the infancy of it. And even though people think that we're late to the game, you think about the rest of the world. You and I are on this channel, and we're talking about AI. So we're like early adopters. Everyone that's listening to this podcast is early adopters. There's the opportunity to grow for it. I see sky's the limit on where AI is gonna be. Like, the amount of stuff that's happening behind the scenes, robotics, or just just our daily usage of it, it's growing exponentially. I'm just super excited about just whatever kinda... What's... Whatever's next. Right? It's gonna surprise us. I think the most important thing, though, I say is the one thing that I do wanna talk about is think about workflows. I think it's a missed thing when we talk about all the things that we're doing, but yet we're not even looking at our own personal work and say, oh, I can embed this into my work. Oh, that's a no brainer. So let's start with the stuff that's really low hanging fruit that we're doing and using that to hyper speed the work that you're doing. If you do that, you're gonna be you're gonna be, like, 90% ahead of everyone else. It's less about the tool, and it's more, like, about the system integration. Think it's not just AI, just whether it's Copilot, whether it's Gemini, whether it's Anthropic. They're giving you the tools, but it's... You gotta think about the overall system, like, how they connect to the... With each other. Prompt engineering is really important. Like, it's... But you know what? It's like, you can teach it how to teach you how to prompt. You can teach it how to meta prompt and do the things that you do. Those are the things that's a no brainer. But like I said, it's just such a great opportunity to see what people are doing, all the wonderful things. So it's like, I... Yeah. Sky's the limit. I just think that we're just, like, in a very unique time. So it's awesome. David, thank you very much for making the time. And to all the listeners, you can connect up with David and me on LinkedIn. David is David, d a v I d k u o c h. Kuch is how you pronounce his name. So search for David Kuoch at Citi on LinkedIn. Hope you can connect with him and looking for ideas on how to be the evangelist in your organization. I'm sure David can help you out, and he's got lots of ideas up his sleeve. Thank you, Abhijit. Thank you for your time. Appreciate it. Thanks for listening. Big thanks to Spark Novus and the Marketing AI Pulse community. You can join the community at marketingaipulse.com.
