Leveraging AI

110 | A detailed framework for mapping, analyzing, and implementing AI solutions with Svetlana Makarova

• Isar Meitis, Svetlana Makarova • Season 1 • Episode 110

Share episode

Unlock the true potential of AI in your business with our comprehensive step-by-step guide to mapping AI opportunities. Whether you're a C-suite executive, entrepreneur, or business leader, this webinar is designed to equip you with the strategic framework necessary to identify, evaluate, and implement AI solutions that drive success.

Join us for an engaging session led by AI strategist Svetlana Makarova, a expert with a track record of transforming businesses through AI. Svetlana will walk you through a structured approach to uncovering AI opportunities, ensuring your investments are strategic and impactful. From ideation to execution, you'll gain valuable insights into how to seamlessly integrate AI into your operations.

In this webinar, you'll learn:
•⁠ ⁠How to identify and prioritize AI opportunities within your organization
•⁠ ⁠Effective frameworks and tools for strategic AI planning
•⁠ ⁠Real-world examples and case studies to guide your AI journey

Svetlana Makarova is a leading AI strategist with extensive experience in both the educational and practical applications of AI, Svetlana has helped numerous businesses navigate the complexities of AI implementation.

About Leveraging AI

If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!

Isar:

Hello and welcome to another live episode of the Leveraging AI podcast, the podcast that shares practical, ethical ways to leverage AI, to improve efficiency, grow your business and advance your career. This is Isar Mehtis, your host. And we have a really amazing guest and an amazing show for you today. That's going to answer. One of the biggest questions that most businesses have right now, when it comes to implementing AI, before we dive into that, a quick update from us to all of you who are joining us, either on zoom or joining us on LinkedIn or listening to this on the podcast afterwards, we just opened our registration to our July session. Of the AI business transformation course, this course, we have been teaching, I've been teaching personally since April of last year, we've been teaching two courses every single month. So hundreds of companies has been through that course and are transforming their businesses with AI based on what they're learning. And we've teaching most of these courses as private courses. So we get invited by different organizations and groups and companies to teach them specifically. And so we open. Public courses once a quarter. So we've done one in October and then one in January, and then one in April. And now the next one in July. So if you want to change your career, or if you want to transform your business and you're looking for a structured learning process that will teach you all In eight hours. So it's two hour sessions once a week for four weeks and it's the summertime. So work is not crazy busy and you probably have time for that. And you want to come back out of the summer break with a lot more oomph in your business. That is an amazing opportunity to do that. And in addition, because of the. leveraging AI 100 episode celebration, we are running a$200 off promotion with the promo code leveraging AI 100 or uppercase. So this is valid until the end of the month, which is at the end of this week. So if this is something that you're interested in, grab your seat right now. But now let's dive to the topic of this episode. So I told you most companies today. Already understand that they need to implement AI in their business. And they're even willing to invest resources into it. But the reality is most of them don't have a clue where to start or how to start or how to create a process that does it. So what most people in companies are looking for is a framework, a system, a process that will allow them to map AI opportunities in their business and Analyze them in a way that will allow them to prioritize it and then assign the right team or the right people to actually execute the plan. we are in great luck because this is exactly the area of expertise of our guest today. So Svetlana Makarova and Svetlana has spent most of her career as a product manager. She's even been an AI product manager of a large corporation way before AI became cool. So back in 2021, and she's done this for several different years. What does that mean? It means that she has a very deep understanding of identifying user needs accurately and then defining how to address them with AI. That's what she's done professionally for a very long time. And she's taking that knowledge and pouring it into her current company spark change in where she helps companies identify and map. AI opportunities and develop a plan on how to implement them. So since, as I mentioned, this is maybe the biggest question of companies today. I'm really excited to have her on the show today. Svetlana welcome to leveraging AI.

Svetlana:

Thank you so much for an awesome introduction. Thank you so much. You've done a really great with, the research and kind of the, You love showcasing some of my experience, so thank you for that. And I really am excited to talk about AI and how you could leverage it in the business and share what I know. So I know we only have a short amount of time and it's, as you packed this information into eight hours. As you've said, I'm trying to do as much in the time that we have. I'll maybe jump into some of the slides I have, slides prepared and please pause me if. There's anything, that comes up All right, so just to get started, and I'm not going to go into the emphasizing and trying to sell anyone that is here to stay or why we need to do it. But one thing I did want to emphasize is that about 85 percent of people, even before the hype with generative AI came to be 85 percent of people in us already have used aI in some way. Think about Amazon, think about your smartphone, think, if you've ever gone to, like Walmart's website and everything like that, yeah, Google, Spotify

Isar:

maps. it's,

Svetlana:

it's already embedded and grained and like the core part of their experience. So you've, chances are you've already interacted with AI. And one of the things that I want you to emphasize is like the AI progress has been astounding. in the last six, The difference between the six years since AI kind of has become evolved, and now it's accelerating even further. but the quality of, these AI systems have, has improved the quality of images, the capabilities have been improving. So there's a lot of potential, and it's becoming more accessible to businesses, whereas I think previously, Only I think big corporations with big pockets were able to deploy these solutions. But now it's becoming more accessible to small organizations as well. And that's exciting. So how can you leverage? And so what builds a competitive advantage in the business? and how do you actually put AI to use? And so When I talk about AI, there's two ways that most companies think about, and it's not these capabilities. I'll get to them to these things on the slides, but there's, the business model of a business, which kind of, introduces the value that business is providing to their customers, whether it's B2B, B2C, there's some way that business is operating, that's providing and generating revenue as a result of that. And in the same time, they're doing it at according to an operating model that promises to deliver that business value to their customers, or, it outlines how the business operates. AI can help with both, and we'll get to why that matters in a second. But as you look at two of these sides of the business, so the business value you can create, and then the operational efficiencies you can gain. how can you build a competitive mode in, in, in the market comes down to two things, data and capabilities. So data, as we know, lots of organizations are collecting data from their customers. whether it's an online platform, whether it's your CRM system, you have data you've been collecting over time. And the beauty with AI is that it uses that data to create capabilities, or it uses that data to enable your business to create new value for your organization. and the capabilities will spend more time in this session, because I think it's important to really understand how to apply AI. Once you have the right data. So as a company, you have first, I want to pause it

Isar:

just for one second. one of the things that I talk about in, in my course, and if you hear me speak in different places is the five laws of success, in the AI era. And one of them is that there's two ways to win and it applies exactly to that, right? So one way is knowing how to leverage your proprietary data. And the other is how do you gain insights? Efficiencies on your day to day operations in the business. And it perfectly aligns with what you just said, right? So there's two very different ways you can leverage AI within your business. And if you can do both of them at the same time, then you obviously win on both. So learn how to leverage your data on one hand, learn how to do the things that you do anyway, just faster, better, cheaper is another one.

Svetlana:

Absolutely. I think that's well summarized and well said. and, just to dive into it a little bit further to build on what you just said. so as a company, you have a choice to make. so do you want to invest in the car or do you want to invest in the house? So I say this a lot, but I don't think that AI is any different. and I'll explain that a little bit as to what I mean. so there is essentially how you actually solve business problems doesn't come down to Oh, we're going to have to use AI for this. Sometimes there are other alternatives to solve a business problem. but the one beauty in artificial intelligence and why investing in that space as a strategic decision is because it comes down to two different things. So traditional digital tools are built according to this product life cycle, right? So they are they have this initial growth curve and then. The value of those tools does deteriorate over time, because they got to get outdated. So unless you're constantly enhancing them with features, there's a chance that it, the systems or the technology does behind the scenes, gets outdated, and then you have to replace it with something else. So that's like investing in a car because it, depreciates in value over time with AI solutions, which I'm equating it to investing in the house. is something that grows in value over time. And that kind of goes back to this, data flywheel and the data flywheel goes like this. So if you create a product that is, you've created this better product for a business, it invites more users because now they're seeing more value in using that product. It just by the customers using that data. They're contributing to, to the data of your organizations that is used to provide the feedback to your AI solution that further enhances that product further and, attracts more users and it becomes a flywheel effect. So that's where AI strategic advantages is that it doesn't actually depreciates over time, but the more data you feed it, the better of an experience you can deliver to your consumers. Another thing that there's three types. of basically companies and how you could leverage in your business. And again, another decision you'll have to make. Are you a taker, a shaker or a maker? And, this is actually a math methodology. I believe this is by, McKinsey and, what it comes down to is, The takers are usually taking off the shelf tools that are ready. the, you buy a subscription like Chad, JPT, and you roll it off across your enterprise and you let users use it out of the box. You don't make any adjustments to it. You just take it then and you launch it across the business. You are, you can also be a shaper type of company. You invest in the customization of existing tools and you customize it to your data, your processes. You make some customization work. Of existing models, solutions, whatever have you, and you can also be the makers. And then the makers and a great example of that would be open AI. So they took a, they saw an opportunity, a white space opportunity in the market, and then they develop a proprietary model from scratch. And I have experience across both of them, but, the makers are the most expensive, of course, to invest. And then they take the longest time to actually build shape. and launch. So it does require, a significant investment. The shapers are actually the ones that reap the biggest rewards because you're not just taking an off the shelf tools based on how that product was built. You're actually customizing that product to your business, whether it's again, a product, a process or whatever have you. and that's it comes down to your investment. The takers, of course, those off the shelf subscriptions tend to be lesser. In costs, the shape are going to be a bit more and then think of makers as requiring some of the most, highest investments in, into the business. okay, but how can you tap into this AI potential once I've made those choices? It's by learning how to apply AI that doesn't require you to understand. the technological jargon or what these things mean. And so I talk about this as far as capabilities. And you saw that in the previous slide. So AI capabilities are the functions and tasks that AI systems can perform. So I kind of group these AI capabilities into five different things, and then this is across the spectrum. So give me one, any, type of AI solution, I can probably group them into one of these five different buckets. So if you think about AI systems, whether it's computer vision, it's machine learning, neural networks, it doesn't really matter. You can group them into these specific capabilities because it comes down to the one question, what can AI do for my business? What do I need to use AI to do? So if you needed to send something, so maybe see or hear, A customer's interaction. If you're needing it to generate something like create creating text, creating video, creating sound, you could use it to control a specific part of your process. So automating, optimizing different processes. Understanding so uploading a document to something like Chad GPT and you're extracting information or you're extracting insights. So classify, discover, analyze, and then you can also use AI to predict. so you're using existing data to predict some future event or some future metric in kind of the future. basically these capabilities use these algorithms, some machine learning, neural networks, these solutions. That traditionally require human intelligence. and the differentiating factor between going down the path of learning how these systems, gen AI systems work, transformers and everything like that kind of doesn't give you the appreciation for what AI can do. And I think this framework and applying this type of view onto understanding how you can apply AI to business alleviates the need to understand how AI works in, the inner workings of AI. So let's dive into some examples. I'll just maybe skip a few, but I have examples, lots of examples here, but just to give you a sense for what it is. for example, Spotify recommends. So there is a solution that Spotify developed called AI DJ. So it looks at your playlists and it creates, uses these advanced algorithms to recommend a personalized song that you might like. so it uses data that exists in your playlist to recommend, something that you might actually like to keep you on. System for longer. And so I have a video I'll skip, but there is a video that actually demonstrates for how it actually works. you have AI Speaks, and I'm sure you've heard of Syn Synthesia. So it uses, AI to, create an avatar that speaks according to a script. I won't play this again, but you can go to the Esias website and see it in action. So AI can also personalize. And so I like this one a lot, actually. so in this case, AI uses the body type and maybe I can play this. I don't think that you're going to hear the sound, but you can actually customize the body type. You can customize the, Personalize what maybe this outfit looks like, but you can see how the user can actually change the physique or the, ethnicity of a person and then apply different, different types of swimwear to understand, like, how, what would this look like personally? So you could see how a company might step up their game in, using AI to provide more value to their customers.

Isar:

I want to pause it just for one second. First of all, for those of you who are not watching and just listening, it's a website that's basically selling clothing or in this particular case, swimwear, and you can upload your image and then basically see how different types of, swimsuits are going to look on you versus trying to guess based on the model that is usually there and Oh, is this Kind of going to be like me or not, but the personalization thing is huge. And not a lot of people understand where we're going with this. And I want to add just my 2 cents on that. This, okay. A lot of people are thinking, this is, I don't sell swimwear. I don't sell goods online. in this particular case, I have services that I sell. How does that even apply to me? And the reality is, at the end of the day, a business is about increasing the cost of acquisition and increasing the lifetime value of your customers in order to make more money. And. Preferably doing this while saving money on operations versus increasing that. And one of the ways to dramatically reduce the cost of acquisition is by personalizing your messages, which means regardless of which business you're in, which niche you're in, which industry you're in, if you can talk to existing clients and potential clients in a personalized way at scale, 20 clients or 2 million clients, but you can address them personally, you will gain more business, more revenue. For longer, meaning you are reducing your customer acquisition and you're increasing your lifetime value. And this was almost impossible previously, unless you were a Google or Amazon or somebody like this who had access to this technology. And now we can personalize things down to messages on our website, the emails that we send, text messages, whatever means we have a communication with clients and prospects, we can per every company now can personalize based on a lot of data points, just using AI. Yeah.

Svetlana:

Yep. And that's a great example for, different ways to use this capability of personalization to actually, again, as we've talked about using data that your organization already has, and then extracting value from that to do more of what it is that your business does. So in this case, as you mentioned, this is the clothing company. So this is how they apply personalization to, to their use case. so you can use AI to organize, kind of data. So there is this app that actually integrates with lots of different technologies and you're able to apply AI to extract insights across these different systems. AI prediction. This one's actually a really cool one. AI, the progressive has this device that you can actually install inside your vehicle and actually tracks your driving behaviors. And based on your tracking, your driving behaviors, it predicts your premiums, insurance premiums. So if you're a good driver, It tends to abide by laws and you're not speeding and that is integrated with the phone's kind of mapping system. So it'll know whether you're speeding or not in different areas. And so your premium would be dependent on your driving behavior. So it is used to predict. the premiums in this case. AI can use to curate, curate information. This is a common one. Adobe has launched this kind of app. Now it's integrated into the Acrobat where you can basically chat with the documents that you're looking through. So I'll skip this one because I think it's a common one. a summarization again, uber uses, some of the summarization of customer or user communications to improve interactions of their customer representatives. so they can provide better customer service. And I'll skip that one. And then AI creates. This is a huge one. And I think we have lots of these different use cases. Now, again, I'll skip, I think a lot of them, but we've all seen Sora. we've seen, I think it's, Get the name of this, but this is a premier pro, I'm sorry, but you can actually use, generative AI to fill, different components of, let's say your video, with Jenny, I, at this point, so this is a great video to like show, how AI can be used. So hopefully it gives you the appreciation that AI can be a superpower. but the question is okay, that's all nice and great. We see what it can do. And it's. It looks like these big companies are applying it, but how can I use these capabilities and how can I, what framework can I use to actually apply it? So it comes down to these three things and I'll elevate this and I know it's written in fine print here, but you always want to focus on your business objectives. And I have an example, we'll walk through for what that looks like, but what is it in that, in your business that you're trying to improve, are you trying to improve, your customer retention rates or revenue, top line revenue. Are you trying to improve your operations? Because you see the, there's some bottlenecks and you're not able to deliver value to the customers as much as you can. So you look for first, the objective, that you've set either for this year, your five year objectives, and you focus on what matters to your business, then you would look across your business to find AI opportunities where AI can add value. You would plan how AI can address those bottlenecks and pain points, and then you would implement. So the identification step is all about creating this heat map of, different pain points across the organization. So think different business verticals, you're looking at each one, and you're trying to see, okay, Things in this part of the business, let's say marketing could have, could be done better considering that there's, AI can now create, AI can now summarize and suggest or do research or automate certain tasks. Maybe this part of the organization can be improved with AI. So you're looking across the organization, maybe a specific vertical, or if you have your organizations, it's smaller and you do one thing, let's say it's construction project planning, You look across all of your business processes and say, okay, can any, do any of these, business processes have pain points or bottlenecks that could be potentially enhanced with AI understanding the capabilities we would then apply to you. Again, I want to, yeah.

Isar:

I want to pause it just for one second. I think there's a, there's two critical aspects of this that people need to understand that connect to everything you shared before. Like every business now has to come in and do three different evaluations. And I'm going to skip one of them because it's less relevant to this conversation, but the other two are very important. one is a strategic evaluation, which is from a business perspective, what is going to change in your industry? In other words, What will your clients, your existing clients and your potential clients not be willing to pay for in two to three years because they, I will do it for them. So that's the risks out of the opportunity of the. Strategic assessment, the flip side of the strategic assessment is what new opportunities will open for my business that were not possible before, either new clients I can serve that I couldn't serve before because the overhead was too high or what kind of new services or products I can provide to my existing clients that I couldn't provide before because I have access to AI. So that's the strategic side, but on the practical grassroots opportunity side, it's really focuses on bottlenecks and pain points. And it's exactly what you said. Just. Your team knows it. if you just ask people to write a list. places where they're struggling today or places where there's bottlenecks and they just don't have enough bandwidth and ask it to write it down. You will have the list very quickly. Now what do you do with it? That's, I will let you continue explaining, but to get the list of what's currently not working smoothly, where are people struggling? And sometimes people are so used to something, they don't even know they're struggling, but somebody else in the organization said, this is not a great way to do that. So just ask your team.

Svetlana:

Exactly. and I have one example, so stick around. I actually have a walkthrough example that we can, you could see it in action, but some of these things conceptually make sense. But now, I'll have an example for how you actually do this. so you would apply these AI capabilities to this process, and you build basically an AI roadmap. You prioritize them based on, when you do the planning part, you prioritize those AI opportunities from your roadmap to those business objectives. Yes. Are these things that we want to put in the road map to tackle next? Do they make the needle move on again, the business model or the operating model, right? Because if you're not improving both of them, your business would not be improving, right? It's all about top line growth, and you can do it one or two ways. you would select then one use case to focus on initially. And I think that's too, that's really important not to try to boil the ocean with all of these opportunities because now all of them look like. Great, things to do all at once, but you want to build that organizational muscle by trying use to, to implement one use case. And there's lots of tactics, to do this, like by forming a specialized team, to do this rapidly and then actually test and validate quickly, which we'll get to. And then you would establish goals for what would you consider that if everything went well and according to plan, what would you consider the goal or that to be a successful prototype or, the successful AI solution? And then what KPIs will you be tracking? And then you would plan the solution for prototype. the key thing is to move quickly with the implementation phase, because I think some, it's really easy to get bogged down by the details. But the idea of the prototype is to demonstrate, and I'm big on this, is to demonstrate the signal that this solution can actually deliver on the promised value. And so you're feeding it data. And you're quickly iterating, you're finding solution, you're, evaluating build versus buy and see how can I get to the solution to this promise result the quickest way possible. And typically that's done in about one to two weeks. Sometimes it's done in sprints, so you can do it in one to two sprints. So it could take up to a month depending on how your organization moves. but the idea is to validate can has what we've planned from what we've planned to the promise result. Can we get there in the quickest amount of time possible? So this proof of concept has to only look at the core components of that solution. So we're not talking about features, that are, nice to have make the UI pretty. Even if it's, demonstrable code, and that's what typically I ask, my teams to deliver is just to show me, can I see some promise of that result if we were to scale that solution further? And then you would make a decision. Do you want to scale that solution to its full scale? which may take one plus months, depending on the complexity of your solution. And then you deploy, launch, kind of everything, the change management and everything that comes with that, you would. Plan around the lunchtime, but that's at a high level. And I want you to, I want to, again, pause

Isar:

you to just add, if you can go back to your previous slide, I want to add a few things to this. First of all, for people who are like, Oh my God, I don't have two weeks to start investing in projects, not to mention a month and a half. I think you touched on this before, on the three types of implementations, right? There's implementations that you can take an existing tool And use it to the benefits of your organization, which could take once you figure out how to use the tool, an hour of work to save somebody in your team an hour of work every week, or sometimes an hour of work every day. So there's. I always like to look at projects and I had years of running software companies. So that's how, why my lens works that way. I always look at a small, medium and large kind of projects. The small projects can take minutes to a couple of hours, and there's a gazillion of them with AI in your organization right now that can be done very quickly. Then you have the medium size, which are these kinds of projects, right? So we're going to invest two or three weeks to create some kind of an infrastructure that can help a lot of people, organization gain a lot of benefits that are not doable. If every person in the organization is just tweaking his own little processes. And then you have the large projects that are more strategic that you're saying, okay, we're going to invest in migrating all our data to a data Lake house and building, AI chats on top of all of it, which is. Definitely doable and beneficial if your revenue is big enough and you have deep enough pockets to do You have to also analyze things through that lens and all three types of projects are relevant to many companies. You just got to identify which ones are now going to go back to the thing that you said, that is the most critical of all of this, what is going to move the needle, right? And moving the needle, meaning either it's going to drive more revenue. Or it's going to cost us less money to generate the same revenue we're generating right now. And if it doesn't, I'm not saying don't do it. I'm saying, put it lower in the priority, which means it's probably never going to get done because there's going to be a lot of other stuff above it in the priority. but yeah, this is awesome.

Svetlana:

Yeah, no, I think that's a fair point. I think you're right. The proof of concepts, I think in the sense are targeted towards that shaper type of, company. but the takers, I wouldn't even, They might take as, as you mentioned, you may not even need to build a prototype there. There's really no development or you could do some minimal development with no code solutions. But this is I think you've called it out pretty well that this is targeting that shaper type organization that is looking to put their data. To use and really personalize and apply these AI capabilities that, will create more value in, in, in their organization. So we need to give you an example for how that might look like. So in the chat, can anyone tell me what company does. Is this from, or what company do you think this is? these images are demonstrating, like what company does this make you feel or remember?

Isar:

So for those of you, again, who are listening, it's black and white images of people working out hard in the gym. Like you can see they're really focused and into the sports that they're doing or the workout that they're doing.

Svetlana:

Right.

Isar:

And Some people are saying Peloton, Nike, another Peloton.

Svetlana:

Perfect. You guys have got it. Peloton. Yep. So this is the example. And so again, as I've mentioned, I can't emphasize enough that you want to look at the business objective. So what I actually did, because I don't work for Peloton, but I did pull out their quarterly, review. I forget the name, but their business, basically reports out to their boards and the investors. And I was able to extract some of the strategic objectives that they've set for this year. and without going into too much details, but they're really focused on product innovation, market expansion. Tech development, community engagement, and operational efficiency. typically in an organization, there's something that you want to make the focus off on across your teams, and you would delegate to different teams to focus on maybe one component of these business objectives. And so one of the things for this, for this demonstration, I just decided to look on, look at the product innovation front. And so to notice is that we talked about the business value, which is the business model and creating new value for the organization or operational efficiencies. And in this particular example, I'm going down that creating new value route and rather than just focusing on how to improve efficiencies in the business. So the first thing I do, I look at their product mix. so what kind of solutions do they have in their business that I could basically provide more innovation in the market? So I'm looking across there, they have the bike, they have the tread, they have the row, they have the digital courses, equipment, apparel, they have a Peloton guide, and they do have an AI solution already in these personalized class recommendations. Not a lot of information on it, but, I am also taking that into account. So what can I do in my product mix to enhance it further with AI? So maybe I'm just considering that probably bike, tread, row, and these digital courses already enabled with AI. Let me look at the apparel. so as I'm looking at this particular part or like this particular product and working with that target team, whoever's working on it with their business executives. So how can I, could I make this product smarter? Could I apply our existing data to improve this? And then how would it interact with other products? So if you think about Peloton, they're all about their product mixes. Could this product be interacting with their app? Could it be interacting with their treadmill. So you're again, like thinking outside of the box and saying, okay, if I was to create more value, what would be that? Improvement or could it be enhanced with potentially? So you're starting to think outside of the box and really trying to, innovate and create these innovative solutions. how do the users interact with this product? So you would create basically a journey map if you're trying to improve, let's say, a process. In the business, you would do equivalently like a process map, and you would look across each different points. You would identify the interaction points with the other product or one way where you could plug AI in. And for example, you could use, you could improve one part of the process. to streamline and instead of let's say a user and I'm just giving, giving you an example outside of Peloton, but let's say if it takes a really long time for you to curate the information from different websites, to be able to summarize maybe, insights from the market, maybe you can automate that task. but by laying out that particular process or, the journey map of that particular, product, it gives you visibility and transparency Into. Okay. I see that this process or the interaction points for this product products are 10 different, steps, which part can be improved with AI. And I'll tell you why that's important. I don't know if you've heard from Andrew NG speak about AI, but AI are what he calls. one trick ponies. And so by also looking at this, at these like smaller, different, like succinct steps, you can apply AI without over promising because AI is meant to enhance and do one specific task. And that's why we call it one capability. So instead of trying to revamp this whole entire kind of process or this whole entire experience, you're focusing on one particular thing. So whether you want, that particular step in the process to be able to. to help a human to summarize or curate information you wanted to automate or you want to help a human to forecast information. So instead of doing all of those calculations manually, looking at that process and really mapping it out step by step enables you to find and spot those opportunities that can be enhanced with a I, And so for, I want to,

Isar:

I want to jump in quickly on what you added and actually, a add a little bit to it, but then also play devil's advocate or what I'm going to say, what you said, one thing, really the first step is to really identify opportunities, right? That's what we're talking about right now. And what's the plan? I did is it said, okay, She looked across the business. And again, this is a very specific example, but you can do the same for your business, look across your business, where in your product or services, there's an opportunity to enhance the offering, what the company's actually doing, going back to the strategic view I talked about before, what new things can we do with AI that we couldn't do before? And as I mentioned, this. It touches two things. Can we service new kind of clients that we couldn't before? And how can we improve how we're serving the existing clients, or offer them more so we can charge more money with what we're doing. So that's the opportunity that she's talking about now, to play the devil's advocate on the small little steps. One of the things I like to talk about a lot, in my five laws that I've mentioned earlier is the law of. switching from a lens of efficiency to a lens of outcome in many cases. And this is not a good example of it, but in many cases you can circumvent the standard steps of a process that we're doing today in a business and go all the way to the outcome or several steps to the outcome. And, a great example is customer service, right? In customer service, you have. Multiple steps to deliver happy customers, right? There's an intake form and there's an IVR on the phones and then there's prioritization and then it's get assigned and then and so on and so forth. there's five, six different steps before customer service actually happens when in reality today, there are platforms and one person that's selling one of these platforms in my audience right now, that's what gave me his idea that actually do customer service. Like it's a standalone system that knows how to do the intake, whether in text or in voice and actually solve customer problems. So instead of doing the five steps, it's giving you the outcome. And that's true in many other, AI use cases. So I encourage people. To do both things on one hand, look for small steps where I can solve the small problems. But in many cases, you got to go back to 30, 000 feet. Look at what the outcome is that you're trying to achieve. And there might be an AI solution that takes you all the way there or half the way there, circumventing three, four or five different small steps in the process.

Svetlana:

That's perfect. and I will say, and I think maybe to build up on that. So there is basically like this interim. for example, when you're applying it to a particular process when you're looking, I would still I still think that this framework would still work by laying out the process because what you're doing is you're taking Pinpointing one bottleneck in your process. And I think that's really important where it could be enhanced next step is for you to look at the neighboring steps and say, Hey, as if I was to improve this bottleneck, how does this impact the overall, like the neighboring steps and the overall process? And so you're saying, and starting to think, yeah, I think if we were able to do this more efficiently, then maybe we don't need this particular, these other steps. So sometimes through this exercise, you're starting to you start small, but then as you evaluate these other steps that are in the process can go from something that used to take eight steps for you to complete to maybe two or three. and what you're also leading there is like sometimes, yes, the off the shelf products can address. A lot of that, with some customization. So I think that goes into like the build versus buy, solution too, because once you identify that opportunity and you go back and you're constantly evaluating against those steps and say, okay, We thought, this was, we've went down to three steps or five steps from our original, like 10 step routine. And when you're evaluating the build versus buy, which I'll get to, you're like, We don't even have to do all of these things because now we can apply AI to the process, but I'll show you what I mean by that. so when you basically make it to this, does this investment make sense also, cause you have to also level up. it's not about, level set. It's not about just jumping into the technology and investing in the space. If it doesn't make financial sense for your organization. You shouldn't invest 20, 000 if you're only getting 10, 000 worth of benefit in return. So you have to project out the benefits. and I think you spoke about it before. So in the customer service, how much time is the saving, our customer service agents and then what can they do in that time? and you're projecting out those benefits and you're mapping them to, is this going to give me more revenue? Growth or is this going to drive more efficiencies in my organization? So there's two ways that I look at this when we prioritize opportunities and look at the ROI behind it is you would map benefit versus cost benefits versus complexity and you assign value. you do an ROI, analysis, whether it's something that's worth that investment. So you want to make sure you don't just jump into it because, Oh, I think we can improve it. Like you quantify the value, because what often happens is that, you jump and Invest into generate generative AI solutions, especially LLMs, without understanding the true costs of inference. And you're getting these like large bills at the end of the month for without those inferences, let's say, for example, costs, but the benefit that you're getting back for the organization. Would have not made sense for you to invest. Yes. is it driving more value, but it's not worth the investment that the company has put in. So before you actually make the investment of building the prototype, doing any type of a development, you want to level set, go back and take it to ground zero and say Hey, if we were to invest into the solution, then there's ways to. to navigate that, via those benefits versus cost benefits versus complexity. And then the ROI calculations and say, is this something that is worth even investing into? and then, yes, let's say we, we definitely think that this is like a huge value gain for our organization. We'll establish goals and KPIs. And why that's important, is because AI, especially when you're developing new solutions or you're customizing existing ones, it's AI is a probabilistic versus deterministic. So probabilistic means that you can't always promise 100 percent of the results. So by setting goals for what you need to achieve in order for you, for it to make sense of the investment, you need to reach specific goals. If you can't reach those specific goals within, X number of months, it's not worth continuing to invest in, let's say in the customization or even development. Further, so I think level setting your goals and expectations for those KPIs and again, I'm not talking about the off the shelf solutions because you're not maybe customizing those, as much, but I'm really talking targeting that, that chaper segment who are looking to customize those solutions going above and beyond that. You want to establish those goals so that you keep your. Investments in check and that you're not over investing in something that's not going to be ultimately bringing value back to your organization. And then when you plan the prototype, you, you evaluate build versus buy. And what that comes down to is, does this make sense for us to develop proprietary, meaning that you're developing this custom, or is this something that, We can buy off the shelf, or we can take an auto ML solution out of, one of our infrastructure models or, whatever have you, but you take either an existing and you customize or you determine whether this is something that can come off the shelves to solve that exact problem. But you already again, establish that benefit versus cost. You've, you said, okay, if it hits, if this is, the investment that we have to make. These are the goals and the KPIs, and then you evaluate build versus buy, which again, goes back and validates whether that's truly worth the investment. And it also de risks it in a way. And so just going back. I my two

Isar:

cents to that. One of the things that I do with many of my clients, and I also teach in the course, I have this, Google Sheets that I can Joyce can probably share in here. Or if you join us on the Friday, AI Hangouts, we can share it over there. but, which is probably a better way because then I can explain exactly how to use it. But, the What that Google sheets enables you to do, it's a tool selection framework, but what it allows you to do is basically to list your needs. So what a lot of people get wrong, if they start with a tool, Oh, they still can do this, and that. I'm like, maybe 80 percent of that is not relevant to you. Like you're looking for the stuff your organizations need to be able to achieve to align with the business goals that the problems that you've identified, you're trying to solve like everything. that's the plan I was talking about. You start with your needs, like these are the things that will help us solve this problem in our business or run faster in our business or whatever it is, then you go and look whether there are tools in the market that exist that align and provide these capabilities, and then you let people rank these tools. So you actually. Do your research, you find two or three tools that might be able to do it. Let people try it out. You let them give a numeric score and a review to how they do each and every one of the needs that you've identified. And then, whether there is, or there isn't an existing. Available tool to buy, to actually do the things you want to do. If there is one, it's probably going to be cheaper than creating your own one. Definitely when you count in like ongoing support and stuff like that and it requirements and so on. If there isn't. Then you can start looking at the, okay, should we develop this? How much is this going to cost? And so on.

Svetlana:

Great points. yeah, no. And I think that the point here is we're trying to really de risk. and I think that's where people, or businesses become conservative and say oh, it's. We don't know what it takes. I think it does require some work and some like pen to paper kind of calculations and identification prioritization for you to de risk that investment and, And it's okay to hire people

Isar:

who know about that stuff, to help you de risk. You're going to pay somebody 5, 000 right now to potentially save you 150, 000 later. Not a bad idea.

Svetlana:

Yes. Yes. Exactly. And and I think to build up on that, so why develop the prototype or like why to even test it, whether it makes sense. And I think going back to the off the shelf solution, sometimes they have subscriptions for you to test or some of the off the shelf solutions are pretty, pretty cheap. inexpensive, to get subscription to. So to a small set of users to really like test that particular improvement and process and you quantify the results. So that whole idea is for you to further de risk that investment before actually investing into the enterprise type of a solution and then rolling it out across the enterprise. But you're, you test small, you validate quickly. So you, I would spend one to two weeks building and then validating the benefits. And then you decide whether or not to scale. So depending on how the results of that particular test or prototype, whatever you end up doing, that do, and do they show the signal, do they give you kind of some indication of promise that they can deliver on your projected benefits? It's that, that ROI calculation. It's the cost versus benefit analysis. So if you have indication that, hey, yes, it is actually saving us two hours a day, then you would, then decide whether to scale or not. And then I would highly recommend scaling iteratively. So for, off the shelf solutions, it may mean by, Different teams and then there's some change management that involves which requires education and upscaling potentially But you also when you do the scaling even of the developed tools There's still change management and other things that you don't want to just blast out an email to your organization say hey now There's this tool The difference with AI is and I found it works. when you prep and then you make some You invest into educating your organization or your teams and how it works, why it's different and then how to use it. And maybe some guardrails for what you can and cannot do with this tool before you actually roll it out. and then you decide to launch and deploy it enterprise wide, or if it's, if you want to do it across your business, but, that's basically how you continue to, do risk and investment. And then each step you're validating, and then you're getting a sense, okay, that's still worth. And you can pause at each step and kind of reevaluate back instead of like blasting it across the organization and just doing this one big launch, and that kind of brings us to the end of the discussions for Q and A's. I don't know if you have any, anything else to add, Isar?

Isar:

first of all, great stuff. I'll do a quick summary. So the first step was to identify the opportunities, right? and you can look at different types of opportunities within the organization. But the three main things you want to look at is you said you want to look for pain points. You want to look for bottlenecks. Why? Because these two things. Impact. And I use the same terminology. So I'm glad I used it moves the needle, right? If you can solve main pain points, or if you can resolve bottlenecks or expand the bottlenecks, maybe it doesn't eliminate them, but they make them wider. Then you get more throughput through your business. And that could be across anything, whether it's sales, marketing, operations. HR, whatever it is, that's limiting your business from growing. So that's one. Then you go to plan, right? Then you have to build a plan and a few important things that's for Atlanta mentioned why the plan has to have a KPI, what is it that we're trying to improve? What's the baseline that we're at right now? So if we're trying to improve this thing, what's our current score on that? What's the current throughput? What's the current cost? What's the current time of the thing that we're trying to develop. So when you do your prototype. You can actually say whether it's doing it better or not. And in many cases, people try to go for trailing KPIs. Oh, this will improve our sales cycle time by 50 percent by the end of the year. I'm like, awesome. But then you have to wait through the end of the year to actually know whether it's working or not. And if you define some leading KPIs that will allow you to identify you're in the right track. You may continue the right projects and cancel the wrong projects earlier on versus chasing stuff that may or may not happen in the long run. So that's another trick when defining KPIs. she talked about, make or buy kind of decision. And we talked a little bit about how you can do that in a structured way. And then in the implementation, I think the really critical thing that you kept on mentioning again and again, which I think is critical is De risk, right? Validate, validate, validate. the fact you decided to go for it doesn't mean that this The best thing to do for the very long run. And you may need to stop it after a month or two months to see if it's actually making sense going back to the KPIs that we discussed. and the last thing that you said that I really is that it's not just about the tech, right? At the end of the day to implement this in a business, you need. Processes, you need guardrails, you need instructions, you need training. You potentially need some additional it infrastructure. So there's just getting the AI tech doesn't necessarily solve all your problems. And so there's a lot more to getting a successful implementation. The company, they're just getting the tech, And you touched on all these really great things. And now I'll go back to the questions that actually came from the audience. The latest question comes from Deb. Can you point to any resources that distinguish special consideration for an AI project versus traditional IT projects? Much of the, preso seems to be fairly standard from IT best practices. So I think the question is, if I'm trying to really understand what the idea is what's the difference between an AI project and a traditional IT project?

Svetlana:

so I would say, so the traditional, solutions and. There's lots I talk about them, and I'll show you a preview in my previous course, but there's just so much information to fit in an hour, but typically, traditional methods use rule based, Programming, if you will. So you're defining and you would, the way it looks like you would define the requirements initially and say, okay, based on if this, then that, and then you would define lots of different business rules for how you want to build your solution. So it is very much rule based types of approaches. You build, different screens that initiate, or like different triggers that initiate basically a chain of reaction, but ultimately it comes down to still rules. AI is quite different than that. You can build these autonomous AI agents that actually use reasoning. And so they don't take rules per se. They actually use reasoning and come up with a solution or those rules that you would do manually in traditional kind of product development. And they would autonomously come up with the steps of reasoning to actually get to the end result. so machine learning.

Isar:

To clarify, and because Deb wrote a clarifying question, she's asking if there are somewhere a list of major AI specific considerations when doing the projects. And I think we can do a whole other episode on that. I will say, one huge consideration is how much data you have that is really proprietary and how well it's organized is a huge consideration. if you have. Lots of good data that is well structured and organized. You can make, you can get amazing results easily because as of right now, most large AI projects, 70 to 80 percent of the project is organizing data, which then if you can circumvent that and saying, Oh, I'm only going to invest a little bit in the data structuring, because it's already great. It's going to save you a hell of a lot of time. I think other than that, I'll switch to other questions because, there's a lot of other people asking questions. A great question comes from Stan Robinson. He's asking, in addition to content creation, what are common pain points, the, for businesses, which are low hanging fruit solutions for AI?

Svetlana:

and that's an easy one. I think marketing and sales. So there's a lot of, Investment, or I think low hanging fruit opportunities in marketing and sales. So Jenny, I has made it much more accessible to poor companies. There's no shortage of solutions that you can apply again, whether it's out of Asians or creative. Tasks in the business in both places. So that's why a lot of maybe the use cases that you hear right now in the industry are around like lead generation, how do you automate voice calls or calls to client, voice agents and things like that. how do you maybe pre populate emails, do like maybe newsletter writing, but, for marketing, for example, so a lot of the generative AI capabilities are like. Very accessible, and I would consider them low hanging fruit for those 2 at least business verticals and they're like the least riskiest. If you will, I would still, consider human in the loop effort there, but there's not as much kind of guardrails and. Things that you have to consider as you will maybe with other functions, like legal, for example. I think those, I would consider those as low hanging fruit. I don't know. How about you, Isar?

Isar:

Yeah, I will probably add a data analysis for better, data driven decision making. So there's a lot of. Easy tools there to analyze any kind of data you have, whether it's HR data or financial data, or operational data or marketing data, like doing data analysis in ways that were only available to really big companies who had really fancy tools. And a big, BI team is now available to almost anybody for almost free. And another one, that also mentioned by Judith is, hr, right? Whether it's onboarding or writing. job descriptions or writing questions for interviews or analyzing responses in interviews. there's so many other things other than content creation. And I talk about this a lot in my course, like a lot of people jumping to content creation, because they see a lot of that and it's easy and it's cool as far as business value, going back to moving the needle. It's probably pretty low on the totem pole as far as the business value that it provides. awesome. I'll take one more question because we're almost running out of time. And, so Albert is asking, what are your favorite workflow automation AI tools?

Svetlana:

So for me, I started using more of make. com. I don't know if anyone's familiar, but I've started to implement that across my different systems and I have a lot of integrations and I love how easy it is to navigate. So it's a great alternative to Zapier. but I would automate summarization or curation of insights from different websites because, time is. Pressures, especially for me, managing multiple things on my personal work fronts. So browsing different websites, curating information, or even, looking at my email and extracting like insights from specific newsletters, is really key. so yeah, I think that those would be like make. com has made it very easy for me to do.

Isar:

Awesome. Svetlana, this was great. There's still a lot of questions going on. maybe I'll try to send them to you and then we can answer them in the newsletter or something like that. We'll figure that out. But I want to thank you so much for taking the time and joining us. If people want to follow you, work with you, learn from you, what are the best ways to do that?

Svetlana:

Yes, so I am, preparing to launch the, this, my US use case mastery course, which covers this information and much goes much more in depth and the actual kind of, the things that we didn't touch on, we like brush through, but it actually goes in very much depth for things to consider. And again, change management, how to build, how to navigate these, These tools, how to actually map out processes, like, how do you apply these capabilities in much more finer, greener detail? I am, I have this wait list. If you would like to join to hear more about it. right now, I don't have an ETA, but, probably in the next month. I will let you know when the course launches. So if anyone wants to join the wait list, please consider scanning this QR code, but otherwise you can find me on LinkedIn, and follow me there. I post, content about AI use cases and things that you should know as you can apply AI to your business.

Isar:

Amazing. Thank you so much. this was really great. I think a lot of food for thought for people who are looking to implement these kinds of things in their businesses. I want to thank everybody. Again, who join us on LinkedIn and everybody who joined us on zoom. And there were a few dozens of people in each and every one of these platforms. So it was great having you around and I, we do this every Thursday. So there's another expert every week that you can come and join us and ask questions like you did this week. And we do our AI Friday hangouts in which we have a more open, like conversation with a community of people. there's again, usually a couple of dozen people who join us every Friday. To have a really cool, open conversation about stuff. We're learning about AI that happens every Friday at 1 PM. And you can find your LinkedIn and ask me how to join that. If you're not on that list yet. And with that, we'll conclude our show for today. Thank you so much, Vitlana. I see everybody either tomorrow or next week.

Svetlana:

Thank you. Bye bye.

People on this episode

Head Shot

Isar Meitis

Host