35 | Enterprise AI Strategy with Andrejs Karpovs

How do large enterprises navigate the fast-changing world of AI? Audrey Chia sits down with Andrejs Karpovs, an Oracle veteran and AI educator, to explore the challenges and strategies behind enterprise AI transformation. From ROI-driven adoption frameworks to integrating AI agents with human oversight, Karpovs shares valuable lessons learned from the front lines. He also offers advice for professionals looking to build their AI muscle without getting lost in the hype. 

This is a must-listen for anyone balancing innovation with enterprise-scale realities.

Join my weekly Newsletter: https://lp.closewithcopy.co/welcome

Transcript

Audrey Chia: 0:00

Hello and welcome back to the AI Marketers Playbook, where we cover actionable frameworks to help you leverage AI and marketing strategies and your business. I am Audrey Chair, your host, and today I have with me Andres Karpovs. Now Andres is an accomplished tech leader with nearly two decades of experience in driving IT strategy, digital, digital innovation. And transformational initiatives. Now his expertise spans enterprise architecture, infrastructure solutions, and managed services. Andres has implemented AI solutions for big enterprises, partnering with Google to deliver AI and machine learning workshops, and has educated over 8,000 employees on AI even within this organization. Andre, welcome to the show. We are excited to have you today.

Andrejs Karpovs: 0:46

Thank you. Thank you. Good to be here.

Audrey Chia: 0:48

You seem like someone with huge experience and expertise in the world of ai. Could you tell us a bit more about your journey, your background, and you know what you're currently doing right now? I.

Andrejs Karpovs: 0:58

Yeah, so I actually started, as a database administrator. So I'm, by nature, very technical, and I did a lot of, administration and, operations for applications, databases, infrastructure. So I kind of, Grew, together with, technologies. And then, basically we have implemented all the modern things like, public clouds, automation, and, every technology has matured. So, so did we. So we kind of, piloted into, modern things and, yeah, I've been doing, Also a lot of, infrastructure architecture things. Enterprise architecture things, leading different teams. And, yeah, that, that's, that's basically, my, background. But, the main, Vendor or the main technology that I have been working for most of my life is Oracle and Oracle Database. But, Oracle has a large suite of enterprise, products, ranging from databases to SaaS applications, crps, integrations, et cetera. And now they also have obviously generative AI service as well, which we are targeting too.

Audrey Chia: 1:56

That is super cool to hear. You seem like someone with huge expertise in this space, but I also know that you are a huge fan of being an AI generalist. Now. Tell us what does that mean and what does that encompass?

Andrejs Karpovs: 2:09

Yeah, so, I've been fan of, generalists and T shape, T-shaped the professionals, for all my professional life. there are different opinions about it. some prefer and, actually. Think that, so,

Audrey Chia: 2:22

so you might have to repeat that line again. There was like that weird sound. I'm not sure if you heard it. No, but just, I'm sorry, just, just two seconds back. Some, I think,

Andrejs Karpovs: 2:31

yeah, so some, I think that specialists are, much better when you're focused on something. but I've been kind of wearing different hats and, doing it quite successfully. And actually I feel that I'm developing more, when I look. From the hol holistic approach. and, yeah, what, what I immediately, when I saw, gen AI technology maturing and, the advantages it, it gives to people, I, I, I, I immediately, immediately thought like, Hey, if generalists, those who were generalists before Gen AI came into picture, if, if, if those guys would be. Utilizing the AI and, being very proficient with ai, I think that could give so much advantage to them. And, there is a lot of, talks now about, one person, founders and one person, like billion dollar companies, unicorns. I think that's exactly where actually the generalists maybe, or Gen ai, realists, how I call them, may be the ones to try such things in future.

Audrey Chia: 3:27

I love what you said, the gen AI realist. Right? I love that play of words. Now, can you tell us what exactly about AI makes it make sense for a generalist, compared to a specialist? Because I think AI is still very new to many people and a lot of people traditionally are stuck in specialist roles, right? So there is already the evolution of people going from specialists to generalist roles, but what about AI makes it suitable, in your opinion, for the generalist?

Andrejs Karpovs: 3:53

Yeah, I think, I mean, I mean, of course, AI helps, equally, good to both specialists and generalists. But what I feel will happen, in future as the models will become more capable and, powerful than some of, the specialist roles, may become obsolete. And, of course there is still a. Huge need. And I, I think there will be also huge need of, subject matter experts in future as well. But it's just like when we realize that we can fully offload some of the tasks to AI without worrying that much about, precision, whether they're still hallucinating or not. So once we reach to that step, when, we are very confident in some of the, maybe it's low level tasks, maybe it's something more complex, but then it. It kind of naturally seems that, those generalists will be able to kind of offload them, them, and then look into something, else. And then kind of now, especially because of AI agents and all these things coming, In the picture, they will be kind of able to orchestrate these things and then, see from the bigger pictures wherein I, I don't think that specialists will be kind of fully, Kind of seeing the same picture, right? So that there will be more, still using ai, but maybe to learn, a topic more deeper and, and kind of pilot into something else. But this is, obviously not a bad thing. So it's just that I feel I. Generalists may kind of achieve the next level with, with this technology.

Audrey Chia: 5:19

It's interesting that you say that, right? Like perhaps more specific skill sets can now be offloaded to AI because they are great at, you know, narrowing down to one specific skill and understanding how to do that. Whereas if you are a generalist, maybe with like a big picture overview, then with that perspective you can also leverage AI to the extent I'm curious to know also. What kind of like use cases have you been seeing or observing in, whether in your company or you know, people that you work with? What kind of like AI use cases have people been implementing right now?

Andrejs Karpovs: 5:54

Well, it's, I'd say there, there are similar things that, that are used across every domain, which is, you know, something that helps driving productivity and efficiency, obviously. So, you know, things like, writing and, and, and doing some planning, brainstorming. I think this is applicable to every single domain out there. But, specifically in my domain, we are also working a lot on automation, which means that. In past, we've been used more traditional automation, you know, workflow based and, and, and, now we are also using, or transitioning more like, intelligent automation where we are also using agents and, LMS to help us, do operations, do infrastructure part and, and. Kind of make, also a lot of, obviously coding is, is one of the biggest use cases, right? And then every it tech company you would have this, this use case. So this is what we are also doing with ai.

Audrey Chia: 6:50

Interesting. Does it seem like, in enterprises, is it a lot harder to make that shift, you know, towards adopting AI or perhaps our enterprises also a lot more, capable of making that change because they have resources. What have you noticed so far?

Andrejs Karpovs: 7:05

No, no. Of course. Enterprise, tend to be slower, right? so it'll take, some time to, to achieve the adoption. obviously there is a lot of, regulation, governance, and, and a lot of processes, which needs to be compliant. So I'd say it's slower in enterprise, but there are obviously, in many enterprises there are some smaller departments maybe, which are allowed to do more experimentation. I. which some kind of, you know, AI startup within, enterprise, right? So there you can kind of, do more experimentation and then show some, some results. And, and then of course, in, in enterprises there is also a lot of governance around which tools you can use, especially in eu. it's not that, Easy to, or it's, it's just, you have to be cautious, right? which tools you are using, which information you're sharing. So that's, everyone in enterprise wants to have their own, proprie, like in-house model basically, or, or something that's running from their cloud tenant, which is not, shared anywhere else.

Audrey Chia: 8:05

Got it. And with that in mind, perhaps you can also share like, what are some of the biggest challenges you've seen enterprises face? Right? Because I know one thing you see is that it's a lot slower to, adapt and adopt because they have to make changes on so many layers. And there could be like data and security concerns. But what are the kinds of like common challenges and blockers that, enterprises have to work with, and how are they overcoming it right now?

Andrejs Karpovs: 8:32

Yeah, I mean it obviously you have to show, ROI, like return on investment. If, if you're doing something with ai, you can just experiment or, or. You know, kind of watching the fancy demos. But then, when it's used on, proprietary data, it's not always, black and white and it's not, not, not even close to, you know, the demos that we are seeing. So when you start really testing something, there's a lot of, bumpy roads over there. So in, in, and then in order to move something to production, you have to to have the clear ROI and, you know, how this either, attracts, more revenue or, or it drives some cost savings. So I'd say that this, this, this is one of the biggest ones, because, you know, in a AI startups, let's say you have a lot more room for experimentation and. that, that's basically, a nature of many AI startups. Now, when they're exploring like fully autonomous AI agent for coding or something, and they, you know, they spend months to come up with some demo and then, you know, someone tests it and then they find some, you know, drawbacks, right? So in enterprise, you can't spend that much money on testing without bringing any result. You have to kind of cautiously, come up with your use case and make sure it has, clear and then, and, good ROI

Audrey Chia: 9:50

Got it. So from what you're saying, it seems like you come up with a use case. You come up with that, you know, potential KPIs you want to hit, and then you measure the ROI at a small scale before implementing it on a larger scale.

Andrejs Karpovs: 10:02

Yeah, yeah. Something like that. Yeah.

Audrey Chia: 10:03

I think that is a great framework for businesses to implement. Even if you are not an enterprise, even if you're running like a small, medium business, this is still a good framework, right? Instead of just using, different AI tools and just seeing which one sticks. I think an approach like that is a lot more structure and measure. So, Andres, you also hold workshops, you know, for folks and you teach them different AI skills right now in these workshops. What are some. Perhaps foundational concepts you tried to cover?

Andrejs Karpovs: 10:30

yeah. Just like showing them what the tools are, what the state of the tools are, what the state of the capabilities are. Because like when you are in a sort of this, AI bubble, it seems like everyone around you knows the same amount of information, if not more. Right. But then when you. Meet some people for the first time who have never heard about, chat, GPT and likes. you know, you, you, you get surprised and then you k kind of, you can go back to the basics and just show, what these tool, what these tools can do. Right? And I always like to show, I had like one slide. About, how, how the pace of development is going. Right? So, and then many, many people who tried, GPT-3, let's say, when it was released, and, and they, they realized, oh, it's hallucinating, hallucinating a lot. Oh, it's not, applicable in my, you know, use case or role or job. And then, they never went back to it. And now when, you know, year, years have passed, like. And we have GPT-4 0.5. Now, you know how much more capabilities at the moment. And then I also show a little bit, even though I'm not a huge expert in, in, AI for video generation, and then maybe, for images, but I just show like how, how many modalities can you, address and tackle with AI and, and a little bit of, how models as well. So just to show the landscape of, of the tools.

Audrey Chia: 11:49

It's great to give people that, you know, overview of what is possible.'cause the problem is most people don't know what is possible. They don't have a basic understanding of the basic tools are out there. So it's very daunting for them to explore ai. And it seems like this big scary world, but like what you see, if you break it down into a lot more accessible bite-size bits and you share like very foundational stuff, I think you get people interested, right? yeah. But I've also had clients who. Reflected some kind of of resistance when it comes to adopting ai. They're either scared or like they're overwhelmed. Have you also encountered that during your workshops?

Andrejs Karpovs: 12:25

yeah. I actually heard, some resistance and, and, some people are, are kind of, again, they're, let's say, Blaming ai that is not accurate. But when I, but then I ask them, when was the last time you have, checked it? Like, when was the last time you tried it? And they say, oh, maybe one year ago I tried it. And then, you know, it, it was not, precise on my use case, but I. Then I kind of explained that the training data has updated, that the models become, more capable. There are now new reasoner models and so on. So then I just encourage them to retest the same prompt or, or the same use case that they did in past. And then, yeah, then, then obviously there are some people, which I heard us saying that this is another bubble and this is similar to crypto and you know. Web3. So nothing will came out of this. But, even with, so the way I like to say about it, that even if, AI would fully stop, like there would be a hard stop on, on all development today, then it would still take like three to five years to. Even transform with the technology state that we have today, because like I said, there are not, we are still in the beginning there's, there's only small percentage of people that are using it daily and actually using it correctly with the real gains. And then even with the current state, I feel like there is so much more that can be achieved.

Audrey Chia: 13:46

It's very interesting because it's the same tool, right? But whether you are a power user or an amateur user actually makes a whole lot of difference. So I was working with a client recently. they thought they were using ai, you know, to its fullest advantage, but when I looked at their workflows, they were using 15 questions to get to one output, when in fact you could just build a GPT and get it in one single try. again, then there's that get between, okay, I, I know how to use ai. I can ask it for things, but do you know how to use it effectively? I think that is like the next, you know, big leap.

Andrejs Karpovs: 14:17

Yeah. That's like, again, bouncing back to this, pace of development topic, because there are so many things, that have been developed, so many new capabilities that it's even hard to catch up for those who are in this so-called AI bubble. Right. It apart from custom GPTs. There, there are now projects where you can attach all your files so you don't have to build your rack pipelines manually and so on. There are now, I have recently shared this, autonomous data scientist by Google, which you can just run and collab, and then it, analyzes the data and you can ask questions. So, so it's like much better than what, chat GPT had. Advanced data analysis feature. So, and, and then if you, when you explore, right, it's like you are your. Kind of finding so many new things that can be used and you don't have time. You just don't have time to test everything.

Audrey Chia: 15:04

Yeah. And it's so funny, like I, I remember last year, during Christmas chat, GPT, or you know, open AI did their Christmas launch. But then they did like so many updates in a short amount of time that nobody really knew what was going on and had no time to catch up. So that was also interesting, right, because there are so many new functions that appear and then disappear after a couple of months. You gotta relearn everything so you could just stay updated.

Andrejs Karpovs: 15:26

Yeah. I actually still have some, because I also, I was watching this, like live streams by open ai and I still have some. Thing in Canvas in my backlog that I needed to test, and I still haven't tested it. So I just, I just heard something they were explaining and then I, wow. This is, this is for me. I mean, I, I, I would love to use and test it. And it's still in the backlog.

Audrey Chia: 15:46

Yes, there are always a hundred things to test with ai, but what do you think are perhaps some of the important like, skill sets that people need Right, in order to start adopting ai? So if they are a professional and, you know, they, they wanted to learn how to use ai. What kind of skill sets do they need? Like, or what kind of, you know, mentality do they need in order to pick this up?

Andrejs Karpovs: 16:05

Yeah, I think they definitely should be curious and, they should be ready to adapt because this will for sure change the way we work. And, there is no way it's being stopping or, you know, this will just continue evolving and, People really need to experiment a lot. So, and, and, and what I'm saying is also, of course there's also, once you start using it, you should definitely, develop your critical thinking e even further, because I. If you are receiving some answer on the topic where you're not really an expert, you have to acknowledge that there can be some drawbacks, hallucinations, whatnot. So they're not 100% correct and reliable. So you, you need to have that skill to kind of notice that you, you, you need to skill to, double check and, and maybe make a correction, right? So critical thinking and, and then. Of course, experimentation. So you, you have to be ready and you have to have that motivation and inspiration to experiment, right? Because, of course, you know, there, there, there are many things that people are doing. so we always say like, we want to automate things to get to things which are more, more important for us, right? But there are things that. Seems like being automated, which people generally enjoy doing. Like for instance, in, in, in the graphic design, like all these images, like fancy images that are being generated with, with one prompt, with a single click, right? People like really enjoy to. Do videos of cartoons and so on. But, yeah, this just have to be approached with the, with the new mindset that hey, something will change. Right? But I, I, I would generally want to experiment with these tools. How can, how see, how can these enhance my workflow? Right? And then take it to the next level. So there was one, one, one, statement. Was it, Figma c or someone who said it, that with this technology we want to, r raise, the ceiling and, and lower the floor. So basically those who couldn't enter that, some kind of field, like the barrier is now lower or it's, it's becoming like, equal for everyone, right? But at the same time, those who are experts, they can raise their ceiling with this technology.

Audrey Chia: 18:11

It's a brilliant perspective to have. Of course, most people are just think, okay, AI is gonna replace my job. AI is gonna take over creativity. AI is gonna, you know, X, Y, Z. But I think that perspective also helps us to realize how you can use AI to your advantage, whether you're, someone new to the industry or someone who's already like a veteran. Right. I would love to circle back on what you said about critical thinking skills. I think that is incredibly important. This is something that a lot of people don't notice, but when AI gives you content, it always tends to sound a lot more confident. It's also formatted in a report form, which subconsciously you might read it and you might be like, yeah, this sounds like it should make sense. but like what you say, right, that critical thinking is important. What do you think are that, you know, that human slash ai skills that we need to, to balance in this new world?

Andrejs Karpovs: 19:01

yeah, I, I mean, so, so there should definitely be human AI collaboration. I, I think, we. Like, I, I would love that we collectively as humanity, we, we would, I mean, when technology gets to a state where it's becoming even more and more capable, like everyone is projecting that at some point there'll be agi it'll be able to do all, human economic labor and so on. So we, we have to decide also as humanity, like we, maybe we want to preserve some of this. All these tasks, right? And, and maybe, I mean, humans should always be in a driver's seat, so we should be acting as an orchestrators. So, that, that, the, the new thing, which I am, now thinking more and more about, and, and I. Where, where I see the, the kind of, the big important is for human is, is also to develop that, so-called agency. So we are talking about AI agents on the or AI side, but for humans we need to develop agency. And that means that human should be, kind of, should be a doer, right? And, and then he should be able to do things, deliver things, and, and he should be proficient in understanding which. Tasks he should delegate to AI and which task he should do them himself and which task he should do in the collaboration with ai. So there was like this, future of work report from, from Microsoft. And I think there was that image, like when, when a human like, leans on AI where it ask it for help somewhere, it just fully delegates and they do it autonomously, you know, and somewhere it's just like always in the driver's seat. So. I think even like this, human AI collaboration, it requires some experience. And then when you use them and experiment with them on a daily basis, that's where you get the, like, the feeling and the taste and everything, like how to handle all the properly. And then if we think about the future where we'll be having hundreds of these AI agents, right, there will be need of this agency and orchestration from the human being perspective.

Audrey Chia: 20:52

Definitely. I think this also really applies to the creative industry. So I came from the world of, brand advertising, but then I also pivoted into the world of startups where it's more conversion driven. So the way I see it playing out is AI is very good for it, like the conversion. Driven ads, right? So anything that requires data that really focuses on, you know, fast, turnaround times because you're just trying to optimize your click through rates and your cac. But when it comes to a big branded advertisement, right, you might want to consider having that human to drive it because the human storytelling being able to still, leverage AI in the process. But still tell your own story. I think there's a beauty in that and I think that's where you have to decide where you stand and what is your role in this bigger ecosystem and where you want to use ai, in your own workflows.

Andrejs Karpovs: 21:44

Yeah, I agree.

Audrey Chia: 21:45

Yeah, and I think one thing that you also talked about is, you know, being able to have the human in the loop, right? So when it comes to. Enterprises or like bigger companies, what does that look like? Like how should they even factor that human in the loop in this process?

Andrejs Karpovs: 22:00

Yeah, I mean, if, if there are some things that are being, offloaded, or fully delegated to ai, let's say AI agents, right? There needs to be means to, Notify a human being if something is wrong, right? So there shouldn't be the case where something is being done autonomously and it's being done wrong. So there should be also an observability on ai, right? So it's like a, any other technology which can fail, which can have a malfunction, downtime, et cetera. So it needs to be observed. So obviously that observation and then this monitoring, that's something that can be also used as a human to. Be notified about any problem and, quickly jump in. Right. So, you know, I had this, we, we have this, car sharing, company called Bolt, which is, similar to what Uber is, right in, in eu. And then there was one conference and I was talking to, to president of Bolt. And then he was, telling the stories of how they were implementing, AI agents in the customer support. So basically sometimes, People were, pushing claims like, their food was, let's say delivered in wrong package or something. And then AI agent, wanted to compensate triple the amount of the, of that price. Right. And then such this is a very generous agent. Yeah, yeah, yeah. And such things are getting spotted by. Kind of observability monitoring, and then it gets remediated, but then the human agent comes into picture. And then if, if, if they see something goes into the wrong direction, then this can be quickly remediated. But of course, such experimentation also gives, Some, again, some experience, and then this can be fine tuned and fixed in the future. But, but that's, that's like just one example, right? And, and sometimes it can suggest some refund or, or, you know, sometimes it can say that, okay, I will escalate to the manager, but there is actually no manager. So it just hallucinated that it'll do it for the sake of kind of, fulfilling the user, kind of, need, right? Wow. That

Audrey Chia: 23:49

is very interesting. I think, I think like what you said, like I, AI is great at responding to, to queries, you know, or requests, right? Which also means that there is AI probability, at least at this point in time, for it to give the user a respond they desire, which is not always what the company's want. So maybe like what you say, having that human to observe that process is equally important, and especially at the beginning when you're still trying to, implement. These new experiments, that's when you really want to have that human in the loop. Do you think that there will be new AI functions or AI roles in organizations moving forward?

Andrejs Karpovs: 24:25

Yeah, definitely. you know, if you start even from the top, there is a lot of, not hype, but, there is a lot of, discussion whether now all the enterprises and organization need see a. CAIO, right? or, or maybe the director of AI or, or data, you know, and then obviously AI engineers, is, is one of the up and coming. And, I feel like in future when, maybe couple more years, in future, I think. All the software development, it'll be like merging in, in this AI engineering. So basically, AI engineer will be able to do the same, role of software developer, plus implementing some of the AI tools and workflows on top. So this will be like, new kind of modern developer. And then, yeah, roles like AI strategist and, even AI researchers, which is more relevant for maybe big AI labs, but they're also becoming in, in demand. Maybe not really a new role, but, I, I'm seeing that enterprises also start to look into this and then, kind of try to hire and have their own in-house, AI excellence. so that would be more data scientists, engineers, AI engineers, researchers, and so on.

Audrey Chia: 25:40

I think it will be interesting to see how roles evolve. I also know that some people, are talking about the concept of having AI sales engineers. So instead of being like fully engineer or fully a salesperson, they sort of have the ability of both to augment workflows. I think this is something we are going to see, not just. Skills, right? But in like a lot of other functions where you know enough AI to put, workflows in place to oversee the process, but you also have enough expertise, to then bring it all together. So in this case, tying back to what we say at the start, being that generalist, knowing enough of everything, I think is also extremely important.

Andrejs Karpovs: 26:18

Yeah. And I mean, generals is not like, There shouldn't be like an impression that, these guys want to know everything and, you know, strive and, and try to do anything, even if they are not proficient in it. And, and even if you are looking into a couple more areas apart from, you know, your current role, that's already a good, Good direction because I mean, especially when, when you feel like, and when you see that, some of, your role tasks are, are being automated and are quite repetitive and, and they're good, like subject for automation, then you have to think about, what are other things you could pivot to, like if you are. If you are, if you are a service manager, then you know, maybe there's something from project manager management that, that you need to take or maybe something from delivery management. But there, there, there is more like human communication with the customer needed and, and which is not for sure not being uploaded, to AI anytime soon because humans love to talk to humans.

Audrey Chia: 27:14

Yeah, and I think as we go on in journey along this. AI driven marketing world, especially I feeling it's gonna be the humans that stand out a lot more than the content. and I think people are gonna go back more to trusting individuals.'cause there will be a lot of, hyper personalized content out there that is really tailored for you, that really speaks to you and sounds human. So it's gonna be a lot. More noise you have to compete with. Mm. And I think the roles that require that human relationship, that human touch, these are things that will never be fully and completely replaced. So it might be good to start considering how you can expand your skillsets to also then prepare for this wave. I am curious to know, Andres, what got you started on your LinkedIn journey, given that you know, you have a full-time role, doing great stuff at the enterprise that you're working with. What got you started into personal branding and also posting content?

Andrejs Karpovs: 28:05

Yeah, actually I, I mean, I was, I was planning to do it for, for, for some years and. That was always in the backlog. Right. And I, well, the, the reason I, when I, when I wanted to start it is that I felt like I gathered enough experience that I want to start sharing, somewhere. Right? I. Because I used to blog, before when I was more technical and doing some Oracle, related things. So I, I used to have a blog and, and I was also a bit more active on x, or Twitter X Twitter. So, but then, that was like a coincidence that, when I wanted to get this idea to life finally and. That kind of, there was a coincidence that chat, chat, GPT was released, like initial version was released. And then when I kind of tested it, I, I was like, yeah, I, I was surprised to say the least. So I immediately thought about the impact it's going to bring in next years. And then my thinking was like, okay, now I want to share my experience, but. Augment it with the thoughts about future and AI and how technology and leadership will change due to this. So that, that's how I started.

Audrey Chia: 29:21

Yeah, and I think it's great that you're also bringing your own personal take, on top of leveraging the trends, right? Because that's your own experience that you have gathered over years.

Andrejs Karpovs: 29:30

Yeah. Yeah. And yeah, I mean, I, I just, you know. just, just a joke. I, I felt like I need to share some of my data to train the new models.

Audrey Chia: 29:41

Well, you, you never know, and I'm sure it'll be very relevant. And what, what perhaps was like the one surprising thing or one most interesting thing about your own personal branding journey so far?

Andrejs Karpovs: 29:54

yeah, I was actually surprised. you know, I, I never, you know, I was never a big, user of social media before, so I had meta and Instagram, but I not never was really active. So I, I, I, I don't, I didn't know about the algorithms and how all these things work and then, you know, all these, kind of tricks, to get, more rich and engagement. So. I was kind of surprised how quick I was able to score, you know, a so-called viral post, which got me something like, a 1 million impressions. So I was like, Hmm, that, that, that, that, that was, that was interesting. And then I. You know, you after this, you get an impression that's like, oh, now I can do it and I will repeat it. But then, you know, you as, as hard as you try, you know, the algorithm gods are not with you anymore. So, and then that was interesting just to understand how this thing works and how many people are actually trying to optimize for the algorithm. And, and like someone, some, some people are calling them algorithm slaves. So, but that, that's, you know, that. That was never my, aim to, to do this, so, so now. Just being consistent and whenever you share something really useful, then that, that would get, get, get you enough, in engagement anyway.

Audrey Chia: 31:06

It's true once, I think one thing that beats, the changes in the algorithm or going viral is consistency. Like even though it sounds dry is the one thing that will. Work for you in the long term. And especially if like you, like Andreas, you're sharing such valuable content that combines both, you know, new updates on AI with your own expertise and what you think the future could be. That's a great plan of expertise. That's experience and also, trending topics. Chatting is a great mix. So maybe to wrap things up, Andres, what is perhaps one tip you would have for a professional trying to break into the world of AI but don't know where to start?

Andrejs Karpovs: 31:43

Oh, my, my, my tip would be not to not, not to try to follow everything and everyone not to subscribe to every newsletter out there. just pick a couple of reputable sources, because right now there are obviously a lot of, AI influencers and AI experts, which, in reality, may not have that much of experience. So make sure you pick, you know, the correct, the correct persons and then really try, try to experiment, with it a lot. So we are in, now we are in an era where we actually need to experiment. We should not. Procrastinate any longer with watching endless tutorials, you know, and, and kind of tricking our brains that we are doing something useful because that, never brings any result. So just, take, whatever you have read or whatever you had in your mind and test it immediately. And, and, you know, take it from there. Think about, possible expansions and follow ups on this. Think about, Some interesting use cases and, just do it. Don't, don't, do not, watch the sand, endless demos and tutorials, and do not trust all these demo tools until you actually test it yourself,

Audrey Chia: 32:52

because, absolutely

Andrejs Karpovs: 32:52

there's so, so, so much, you know, sugarcoating right now.

Audrey Chia: 32:56

Yes. So I think like what Andres said, don't just watch the YouTube video of someone gyming. Go to the gym yourself. It's the same thing with ai.

Andrejs Karpovs: 33:04

Yeah, exactly. Put

Audrey Chia: 33:05

in the work, put in the wraps. That's how you actually build your AI muscle and that's how you get Yeah, better. So thank you so much for your time. Andres, where can our listeners find you?

Andrejs Karpovs: 33:15

yeah, just my LinkedIn. I'm not sure if you are sharing any links or something, but, we will pop

Audrey Chia: 33:21

your links in the comments. Awesome. Thank you so much for joining us. It was a thank you for having me. Pleasure having you on the show. And thank you folks for tuning in. Don't forget to subscribe to the AI Marketers Playbook and hit the bill for more actionable marketing insights. We'll see you next week. Take care.