Fernandes: 0:00

Maybe I can pay directly to the client and say my customer acquisition cost is 1.5k. Hey, here you have 1.5k and that's my right to optimize your business. You won't even pay me. I will pay you to optimize your business and then you will pay me back in terms of the resource.

Joeri: 0:20

Hello everyone and welcome to the Web3 CMO Stories podcast. My name is Joeri Billast and I'm your podcast host, and my guest today is Kelwin Fernandes. He's the co-founder and CEO of NILG AI. With 15 years in artificial intelligence, he has helped businesses in healthcare marketing and YouTube channel with more than 100,000 subscribers sharing practical AI , and I see the YouTube sign i integrate AI, care marketing and automotive integrate ai. He published over 30 scientific papers and built a n the background. It's following yeah, amazing. Yeah, thank you some. Sometimes I speak a little bit portuguese when when I have Portuguese guests on the show. So welcome, Kelwin.

Fernandes: 1:07

Thank you, you're from Venezuela. I'm living in Portugal now, but I'm actually also a foreigner, oh yeah.

Joeri: 1:14

I forgot. I forgot. Yeah, I think you told me already. But, guys, if you're now wondering who is Kelwin, well, we met at CODA in Porto. You know you're a speaker there. I was also moderating a panel over there. I was on Kelwin podcast and now he's online. So, Kelwin, let's just start off directly, because I'm curious hey, you moved from deep ai research into entrepreneurship, so what was actually the turning point for you to decide to build your company?

Fernandes: 1:47

So I'm from Venezuela, but my family is Portuguese and the Portuguese community in Venezuela almost a hundred percent are entrepreneurs, so it's a very entrepreneurial community. So my parents, grandparents, were always business owners not tech companies, but restaurants and supermarkets and things like that traditional businesses but it was pretty ingrained in our culture. Then, because of all the situation in Venezuela, I came to Portugal and just took a break Doing my PhD. Those were a few years to rest and then, once I finished my PhD, I did what whoever is a good technician does, which is let's start doing freelancing and consulting instead of working for somebody else. Then I got a few big clients and turned it into a company. That was in 2018. But yeah, it was a transition. Let's say, always had the entrepreneurial mindset in my mind, founded a couple of startups in Venezuela and then we moved to Portugal after the PhD, decided to resume my entrepreneurial path.

Joeri: 2:54

Well, yeah, you were quite early in AI comparing to most of us. You know that became like AI users in the moment. But I'm curious actually, because you did early work in machine learning how did that shape the way of approaching, or the approach of, ai strategy for businesses?

Fernandes: 3:16

So when I started with AI. I actually started before even machine learning. I started with planning and scheduling, which is basically like how to create plans for logistics and combinatorial optimization and things like that. So since the early beginning, I started thinking about AI in terms of decision making and not so much in terms of chatbots or in terms of just forecasting. How can we create actionable insights, actionable decisions into a business? So I started with planning and scheduling, then moved to machine learning and as the field has evolved, it has become more, let's say, a commodity nowadays. So using AI today I will say it's even a commodity. You don't need that much of a technical background to do things with AI. You can just plug and play a few blocks and that's it, which means that you have a larger pool of people trying to use AI and people that are on early stage cycles or early stage steps of their adoption journey. So we recognize that we needed to move from just implementation into guiding people on why, where and how they should use AI.

Joeri: 4:24

Yeah and okay, why, where and how they should use AI. Yeah and okay, now lots of people. Everyone talks about AI, but still at you know, companies. I also give workshops for companies. They still treat it as a buzzword or something you know they are a bit scared about. What are the patterns that you see to? You know separate those who create real value and maybe those who are just, you know, doing stuff or experimenting with AI.

Fernandes: 4:51

So of course, there is a lot of fear of missing out, typically those cases you will see on people that are trying to have a very transactional mindset on AI. This is plug and play and it will solve all business problems. Instead of being ingrained into the company culture and how they make decisions on how they approach data and the way they are approaching the project, is this a plug and play project? Or they are looking at it as a plug and play project or they are thinking of it as let's do a continuous improvement on our processes that will lead to a certain gain? So when you see that they have this mindset, it's because they are not into a fear of missing out pattern. I will say, of course, all of us are with a bit of fear of missing out and we are all trying to see how AI could help. I don't say AI is just making a skill right, so the only purpose we use AI is to help us make decisions. Which decisions? That's the right question.

Fernandes: 5:50

And you see a lot of chatbots, a lot of people doing just content creation with AI, like the dummy, the naive use cases that are the most trivial to think about when you think in terms of chat, gpt, and I believe those cases will fade away over time because people will realize that maybe the chatbot didn't change the bottom line that much because they just had one customer support person or just a few in terms of their bottom line how it's incorporated. But if you go into the core decisions that the business is making, like the operational ones, the investment you will do, it will be sustained over time, right? So, for example, let's say that I am a manufacturing company. If I'm trying to implement AI in order to automate customer support, probably it won't be a core decision in my business, right? It's just something that I am managing as a support function. So I don't see a long-term commitment in those people.

Fernandes: 6:42

But when you see that they are using AI, for example in a manufacturing case, to optimize their production workflow, how they manage suppliers, how they manage sales, to get supplies, to then orchestrate logistics, then you know that they are thinking about their core processes. And when they're thinking about their core processes, typically a long-term bet that they will keep improving over time. And those cases tend to be more successful, in my opinion, than when people is just going to the naive or to the simple use case. That is when you think about chat, dpt, let's try to have a chat, but that's my like rule of thumb, not the rule to understand if the client's just not trying to go with the wave or if they are actually trying to reimagine their industry, their workflow, their processes.

Joeri: 7:29

Exactly, yeah, that they know that they have a strategy and they do some thinking. And it's not just about diving into the tool Now, talking about chatting, and you know people, it's now when we are recording this podcast. It's August 11th. Gpt-5 launched a couple of days ago, so I'm curious to hear what are your first thoughts. What does it mean, maybe, for the adoption of AI? I'm curious to hear.

Fernandes: 7:55

So we have been trying on the projects we already had ongoing. We are just switching models from GPT-4 to 5 and see how the performing goes. In some cases we are seeing major improvements. In some cases, actually, it's pretty bad in comparison with what we had. Maybe it's because the prompt was super fine-tuned to GPT-4 and maybe we need to fine-tune it again. But we roll back to GPT-4, in some cases Even internal automations that we had.

Fernandes: 8:23

I just switch it to GPT-5 to confirm and then move them back, I see. So my first impression is it's way more natural in the way it talks. So it feels more. I won't say it's thinking more, because it's not thinking, but it seems like the answer is more tailored to your question and the response seems to be more human than before. But I don't know if that's because it just changed the typical style of GPT-4 and then I'm not used to recognize the new pattern and maybe in a few months I will say this is AI generated and it's pretty obvious that it's AI because it has the new em dash. Let's say the new em dash will be I don't know quotes or it will be a semicolon, whatever. I don't know what it's going to be, but it feels more natural. I'm not sure if it is because we are not used to get its responses yet In terms of coding.

Fernandes: 9:19

FRANCESC CAMPOY FLORES yeah, that's interesting FRANCES singer yeah, in terms of coding, I haven't tested it yet for development. The team is working on that, but personally I haven't had a first impression on that part.

Joeri: 9:28

Yeah, yeah, because it's very fresh and you'll see, yeah, it's interesting, and chat gpt came out and you could use it and nobody would see that it was ai. But you know, one year later you look back at those texts it's like, oh, it's so obvious, right. So, yes, I agree with what you're saying. Now you mentioned decision making, that I can really help with that, and actually I was also earlier in ai, but then you would think, because I was into business intelligence and ai was played a role in there. Now, if you are now talking to executives, how do you explain these complex AI concepts to these people in the context of decision making so that they can? I would say, yeah, not fear it, but really, you know, understand that it drives decision making. How do you explain?

Fernandes: 10:22

it. Let me just see here a reference that I always use yes, to explain it. It's a book prediction machines and I love it. I think it's the best explanation. This one, yeah, it's the best explanation that you can get about the eye. It's from this book prediction machines and then the follow-up book, which is power and prediction this one here in prediction machines, these guys say that AI is simply a prediction machine.

Fernandes: 10:53

So the only thing AI is doing predictions okay is predicting at a very low cost, okay, and a prediction is nothing else but a new data point that you didn't have access to, a new piece of information that you didn't have access to. And by having access to that information, you are basically lowering your uncertainty on any decision you're making. So let's say, I need to understand how much product I will buy for my next quarter. If I know how many sales will I have, then I have a lower uncertainty on what is the right amount of product I could buy. If I lower this cost of predicting this piece of misinformation, I have to make good decisions. I can accelerate and improve all my decisions. So, in the end, ai is just lowering your uncertainty on any decisions by running predictions, and that's this book, and the typical gain that we will get from AI is just by replacing a small decision that we already make with AI. So let's say I have a full workflow and then we are using AI just to replace one small piece of that workflow into AI, into a prediction made by AI. What the guy suggests in the follow-up book is that the actual transformation will be when we reimagine our workflows. Okay, so instead of just switching a small piece, a small decision is now that I have this prediction machine up front, can I reimagine my entire workflow so I make it 10x more efficient? And I will give you an example that we are doing now at Nelvi and our company with our podcast Something super simple.

Fernandes: 12:32

Let me give you a example that we are doing now at Neil AI and our company with our podcast Something super simple. Let me give you a very simple use case. If I ask you how would you use AI to make a podcast? Most likely you will say I just pass the guest to chat GPT and I ask it to generate 10 questions. That is the point replacement, where you are typically building a script from scratch and now you're replacing it with ai, making the script of the questions that we will ask. What we are.

Fernandes: 13:00

What we did internally is, if I have an ai model that tells me who is a, who is a good guest and who is not, then let's rebuild our funnel for podcast. So what I do now is I follow the podcasts that bring our ICP. We see who they are inviting. I have an AI model that checks how good is that guest, if they are aligned with our vision of AI, if they are not, et cetera. So it scrapes all their social media, qualifies the candidate, ask a human if they agree or not. If we agree, it already prepares the questions and try to generate how the other person will respond and if that episode will be engaging or not.

Fernandes: 13:38

So when I even consider to invite a person to our podcast, it's already fully qualified by an AI model. We already have the potential questions. If they are going to be engaging, they are not going to be engaging. All of that right. So that is rebuilding the workflow. How can you rethink your entire workflow now that you have AI, instead of just doing a small substitution of one step in your pipeline with AI? And at that point is where you get the real gains, because then now you can do 10x more than you did before, or you can do 10x better than you did before, just because you can manage your exploration, just because you can anticipate a lot of missed paths that you will do in a regular way yeah, very interesting example actually.

Joeri: 14:23

So I also try, of course, to use ai to get the best out of the podcast. You know, on the side of the guests, on the side of the questions afterwards for the repurposing. So you're really focusing on your podcast. On your YouTube channel you also speak because I've met you there at Coda. You are very ambitious, I know, with NIL in Portugal and Europe. What are maybe other things that you are doing for the visibility or the marketing of the company, or is the focus the podcast?

Fernandes: 14:55

Yeah, so in terms of so we do our projects right, we do the strategy and development for companies, from small companies to pharma, big pharma, big techs, et cetera. So we have a vast range there on the development part In terms of marketing. So you focus more on marketing and we are very ambitious on the marketing part. I like the approach of the 10x rule. By what is his name, the 10x? Who was the author of that? Grant Cardone maybe? Yeah, grant Cardone, the guy who Grant this guy yeah, Grant.

Fernandes: 15:30

Cardone, this guy and he has an approach which is the 10x rule, which is, whatever goal you have aim for, 10x more, and then, whatever you think you need to achieve that 10x ambition, apply a 10x on the actions that you will make. So we wanted to be the top referencing AI in Portugal. So we said that to accomplish that, we need whatever number of brand, brand events, activities I don't know how to say it in English and we 10x that and we made it to 140 this year. So this year we have 140 events speaking events, podcasts, lives, et cetera and I thought that was already ambitious, but we already achieved that in June. So we got almost one event every day of the year. So we are doubling on that. And then we 10x on the actions that we needed to accomplish that and to 10x the actions. It meant that we needed to automate a lot of it, like how we find the right events, how we submit our proposals to those events, how we submit our proposals to those events, how we repurpose content massively, et cetera. So that's how we achieve AI, how we use AI there, but then, in terms of our marketing strategies, it's actually to be top of mind. So if we want to be top of mind. Where do we need to be?

Fernandes: 16:46

And then we started mapping the journey of our ICP. Like our typical ICP is a business manager already in a senior level. So they go to school. So we go to private schools to talk about AI, because then those kids will talk to their parents, right, and then their parents will say who said that this guy? Okay, so let's talk to that company. Then we are mapping, for example.

Fernandes: 17:07

Still we are trying to map all potential touch points, all potential channels and to be there massively, whether it's events, whether it's live, whether it's podcast. Business directory is LLMs. So we are optimizing all our online presence to be mapped by LLMs. So whenever we ask best AI companies in Portugal, we get there on the top three in any AMO. So I will say we are trying to be everywhere, but in an intentional manner. It's not just create a post, it's not just be cited on a journal, but actually to bring value in all of those narratives. So we are going podcast events and writing a book. We are considering even outdoors nowadays. So we are trying to go massively. And then people always say in your small company company, how do you manage to be in so many places, and plenty of it is because of ai, because we are automating a lot of it yeah, but still, and that's also what I speak about in your own voice, so it's not like it's noise.

Joeri: 18:07

You know you make content. That is because it's not just like one question and not just one problem. There is a whole system behind and also, I guess the aspect of the mindset that you have to 10x better is also very important, right?

Fernandes: 18:21

And then to have the right team. Also, it's quite crucial to make sure you maximize your impact. Yeah, absolutely.

Joeri: 18:30

Absolutely the right team too. And now this is obviously a great opportunity for marketers using AI. But, in your experience, what is the most underestimated business problem that AI could solve today?

Fernandes: 18:46

It's a good question. I don't know.

Joeri: 18:49

Yeah, it's a difficult question. I wanted to challenge you a bit, Kelly.

Fernandes: 18:54

Let's say I will go with small businesses because small businesses I believe the massive impact on AI won't be on large corporations, it will be on small businesses. We are facing a stage where we will need to understand which path leads to a higher success Large amount of data which corporates have, or new, redefined workflows with AI which small companies can create very fast. Right, A large company cannot readjust their entire business model and rethink their workflows from scratch. Right, they are very rigid, so they can only do minor optimizations. With AI, a small, medium business can easily. Let's just dump everything we have and rethink how we charge which our customers have, which ones are our channels. So I believe the most underestimated use case I won't say it's a use case, but the most underestimated use case I won't say it's a use case, but the most underestimated power of AI is reimagining a business model. For example, I will just give you an example very tied to marketing and sales In consulting, the cost of development and the cost of a consultant is getting close to zero, right, because you can now go to ChatGPT and ask questions and develop code.

Fernandes: 20:18

I won't say if it is the most scalable code or not, but it's certainly getting more efficient, close to a cost of zero, right? So if you are reaching a zero cost of implementation, why the hell would a client pay you to do a project, right? Or are we tying pricing to hours in consulting? Is that a useful business model nowadays that we have AI? Most likely not. And this happens in consulting. It happens on services. It happens on a lot of industries, right? So the business model of I charge per hour or maybe I don't charge per hour, but I link my pricing to effort will be redefined. So, let's say, my cost of production gets very close to zero. That happens. Maybe I can start charging by impact, by results, because my risk on the time I will invest is also close to zero, right? So if I have zero production costs, I can charge per impact, which means that I can start embracing smaller clients, because there is where you will get the highest uplift on the outcomes, which means now you have 10x more potential clients because now you are cost-efficient for smaller companies, which means that now you can offer a higher engagement relationship with them because your operating cost is lower in terms of developing the product. You can engage more with them in terms of strategy, which means, in the end.

Fernandes: 21:55

That and this is something I have been trying to idealize which is we have a customer acquisition cost, right? Which is how much does it cost me to get a new customer in terms of ads, marketing expenses, et cetera? Right. And then all the gain of marketing and sales and production is if I can get enough lifetime value from my customer acquisition cost, that's all the gain, right. But what if my cost of production is zero?

Fernandes: 22:22

If my cost of production is zero, if I can automate marketing, maybe, instead of paying my marketing budget to social media platforms to advertise my services, maybe I can pay directly to the client and say my customer acquisition cost is 1.5K. Hey, here you have 1.5K and that's my right to optimize your business. You won't even pay me. I will pay you to optimize your business and then you will pay me back in terms of the results. That's a new business model. You don't see that anywhere there.

Fernandes: 22:57

But if my cost of acquisition gets close to zero because I can automate and personalize marketing and sales if my cost of production gets close to zero, what matters is who has the best strategy out there which means I can move my customer acquisition cost to the end client and pay him directly. Here's my fee. This is my right of optimizing your business. Here you go 2K. Now I have six months to optimize your business and it's a very low risk because I won't invest time. It's because it's all automated, and then you pay me in return. I don't know if that will happen, but I believe that that's a feasible business model in five years, right? If we managed to automate everything. That's feasible, right? It's a very you know right.

Joeri: 23:44

Yeah, it should be feasible, you know, if there still needs to be. You know the human aspect in understanding how to, you know, put automation into play. But yeah, I think it's. It's also better for both sides, right, also for the client or for the, because they, you know, the risk is less, and for you, you know, because the risk is also less it's.

Fernandes: 24:06

I think this is really here I'm thinking in the most utopian case, which is your cost actually goes down to zero and everything is automated. So that's the extreme case. But we need to start thinking on the extreme case to see where we will fall in between and what could be feasible and what could not be feasible. That's my point.

Joeri: 24:25

Absolutely. You always need to aim farther. You know, because then you never know you know what you will be able to do. Move, because then then you never know where. You know what you will be able to do. Now you are, you know you are an entrepreneur, building your personal brand and, like I, you know, building my personal brand. I'm an entrepreneur. I'm always curious to hear from the others, like people like you, what is the most powerful lesson that you have learned in being an entrepreneur and building your personal brand?

Fernandes: 24:54

I don't know if it was from Netflix, slack or Spotify. One of those founders said that first time entrepreneurs focus on product, second time entrepreneurs focus on distribution. So when we start building a business, we tend to focus on high quality product and delivering the best service in the best class, in class service, et cetera. But if you are always in survival mode because you don't have enough exposure, because you don't have enough clients, because you don't have enough exposure, then you will never be able to deliver a high quality product because you are always afraid of losing your next client. So when I recognize that and I move my focus from how can I improve the service into how can I get more traction, more exposure, more clients, etc. That changed the game right, because in the end, you will have a higher peace of mind, which will allow you to provide a better service. So focus on distribution, not on product.

Fernandes: 25:53

And then, if there is the saying that first-time entrepreneurs focus on product, second-time entrepreneurs focus on distribution, there is also an extra, which is third-time entrepreneurs focus on networking right on their ecosystem, and I believe the personal brand is highly connected into that third step, which is now that you have reach, now that you can get to a lot of people. How can you leverage your direct connections, your reputation, all your relationships that you are building actively to your business right, whether those are low effort digital connections that you get through social media, whether those are deep ingrained relationships that you build in these circles, retirements, networking events I know you have also, I believe, a retirement or a mastermind in Sintra right. So I believe that's Retreats, yeah.

Fernandes: 26:54

That's like the third level of entrepreneurship, which is okay. Now I have all this reach now who.

Fernandes: 27:00

I know and who knows me and who can deliver to whom to leverage that capacity into my business? And that's basically what I've been trying to build this last year through all of this outreach out there, which is if it's easier to connect with a person that's doing a brand. That's obviously. Everybody knows Elon Musk. Not everybody knows Neuralink, right, but directly. Of course. A lot of people know Tesla and SpaceX, but Neuralink not so much. But as soon as it's from Elon Musk, you get to know that it's highly technically advanced, it's high tech, et cetera. Same thing with ThinkWork or Steve Jobs or whoever. Once you build a reputation, a personal brand, it's way easier to promote ideas and to position yourself and your services in the market.

Joeri: 27:50

Yeah, yeah, I think, calvin, having your personal brand and combining this with AI, that's the best of both worlds, because you really need that. Well, the time went really fast having you on the show. Is there something at the end of the podcast episode that you want to share with the audience, something, maybe that you are really excited about?

Fernandes: 28:10

Maybe subscribe to the YouTube channel. It's the same name as in our business. I'm writing a book, so you will hear about that by the end of the year. It's an intense thing, but I think you also have a book right, Absolutely.

Joeri: 28:22

Yeah, I put it on LinkedIn, maybe you saw the post. I have a book which is on the background. I have three books, but the one the 5K Challenge for Solopreneurs did really well on Amazon a few years ago. And then I have the two other books that I've written with Mark Schaefer and Joe Polizzi, where I wrote a chapter in there. But now I also have news. I am writing a new book. It's actually 90% finished. It's the future CMO, which will also involve AI, of course. So yes, calvin, it was really a pleasure to have you on the show. You mentioned the podcast already, the YouTube channel I put it in the show notes. If people want to connect with you in another way, where would you like me to send them?

Fernandes: 29:02

yeah, for sure, through Instagram, calvin FC and LinkedIn as well.

Joeri: 29:07

Thank, you okay, perfect, guys. What an amazing episode with Calvin. I was on his podcast. Check that one out, this podcast episode that we just recorded. I'm sure this has a lot of value for people around you, so be sure to share this episode with them. If you're not yet following the show, this is a really good moment to do this. You haven't given me a review yet. These five stars will help me reach an even bigger audience and, of course, I would like to see you back next time. Take care.