← The Top 5% Method®
The Top 5% Method®
The Great AI Decoupling Is Here. Your Resume Was Never the Point.
00:48:32
What happens to your executive career when a company with only 3 employees and $6M in funding can seamlessly scale to serve 2,000 clients? In this episode of The Top 5% Method®, host Katheline Jean-Pierre sits down with Luis Rodriguez, CPTO Middle East of Lab 49, who trained neural networks at CERN long before algorithms went mainstream. Luis drops a chilling reality check for corporate leaders: we have officially entered "The Great AI Decoupling"—where business growth and revenue scale are completely disconnected from headcount. If your career value is tied to the size of the team you manage or a static resume hidden behind corporate walls, you are in a highly fragile position. Tune in to learn: The Decoupling Reality: How micro-teams are using AI agents and LLMs to replace entire departments and out-compete legacy institutions. The Enterprise AI Trap: Why 95% of corporate AI initiatives fail (and the exact question leaders forget to ask). The Ultimate Future-Proof Skill: Wh...
Loading summary
Transcript117 lines
- [00:00]
A
I think we're officially hitting a massive decoupling because for the first time in history, the business growth will not be connected to the headcount. Today already you can scale your revenue by 10x. You don't even need to hire a single person, which is a dream of all CEOs, right? They just want to generate more money
- [00:18]
B
and your margin is going to become bigger. Your ebitda. You look like a star. If it works. If it works.
- [00:25]
A
If it works.
- [00:33]
B
Hello, Driving Impact collective, the top 5% method. Today. I'm super excited about our next guest who's joining us from far away. And it's been really a day, a week, months in the making to have Luis Rodriguez, who's the CPTO of Lab 49 and who's been working on AI. Before AI was popular, before AI was a thing. And he's going to tell us a little bit more about it and what's exciting also about Luis. Both Luis and I met on LinkedIn when we were talking a lot about artificial intelligence and the future of work and the future of the world. And he has amazing insights on the platform. If you want to follow him. Without further ado, Luis, please tell me about yourself.
- [01:19]
A
Yes, hello. Thank you. Thank you for having me. So I'm a tech guy by training. I did AI before, as you said before. This was fancy back at university. I was doing neural networks and facial recognition a long time ago and I kind of stopped and I went back and on and off. But AI has been there for many, many years. If you look at what was. People now talk about LLMs and generating nice cat videos and generating Keanu Reeves fighting some other random actor on sea dance model a few weeks ago. But they forget that the world is powered by AI. The how you book a flight is powered by AI. The price of that flight is machine learning, which is effectively AI. How you invest in the stock market is AI. So all of that has been there for a long, long time. I've been doing it for a long, long time also.
- [02:10]
B
So you also studied at cern, which is the center for European Nuclear Research. And it's funny because you're the second guest of the podcast.
- [02:19]
A
I watch your other guests. It's very, very cool.
- [02:22]
B
So tell us a little bit how it impacts you today as being the CPTO. So Chief Product and Technical Officer at Lab 49, which is implements AI driven solutions, correct?
- [02:33]
A
Yeah. So CERN is a magical organization. It has, at the time I was there, more than 10,000 people that have access to the. To the side in my experiment I worked at Atlas, we had like, 3,000. And the whole objective of what we were doing back then, and they're still doing, was to prove that there is a particle called Higgs. There is a movie that calls it the God particle that can give mass to stuff. So stuff has a mouth, right? Has people call it waves.
- [03:03]
B
Wait a minute. You talked about the movie. Are you talking about Lucy?
- [03:06]
A
No, no, I'm talking about the angels and demons.
- [03:09]
B
And it talks about the God particle.
- [03:11]
A
Particle, yeah, the God. The God particle. It's. I think it's that one that is absurd, because they're two. And I always keep them. I always confused them in any way. It's from the stories, from the book. And there are two books, and one of them is at certain. And there. There's this particle, right? It's a magical particle that gives mass to things. So when you say something weighs something. Well, effectively in physics terms is. It's a mass. It's their mass. And this particle is the thing that makes. It makes things have mass. And it was actually found during. When. Well, I cannot say that it was me that found it. Clearly not. I was just one of the, I don't know, 3,000 people or 6,000 people that were working in experiments that, at the time when I was at cern, it was when it was announced. So there was a massive celebration back in 2012. It was a very interesting time. You know, you have hundreds of nationalities there, a lot of people. It really marked me the way. The way that I've been, you know, communicating and working with people around the world.
- [04:07]
B
So, Louis, I need to triple click on the gut particle because I have no. Like, I saw a couple of movies, I think. Are you talking about the super computing power of the human brain? What are we talking about?
- [04:19]
A
So it's called the Higgs boson. And there is a bunch of. When you. When you talk about matter, you have atoms and atoms. Normally, people think about protons, neutrons, and electrons, but there's a lot more stuff. So when you go into the atom, the neutrons and protons are composed of other stuff, other bosons. And one of the particles that exists there is this Higgs boson, and this H boson appears in very specific situations. So this massive experiment called ATLAS and another one called cms, and at opposite ends of the LHC and LHC is this tunnel that is a huge tunnel built 100 meters on the ground in Switzerland and France below CERN. And there you accelerate particles or accelerate hydrogen particles at close to the speed of light. And when they crash, they create all these small things, right? Imagine if you get two stones and you crash them together and stuff comes out, right? So imagine you take a bunch of clouds of particles, you crash them together and then you'll get a bunch of other stuff coming out. And that's by doing this millions and millions of times, you're going to find a bunch of other particles. And throughout, you know, the years of running these experiments, they found the, they found proof that this particle exists.
- [05:44]
B
What does the God particle do? Like for the common moral? What does it do?
- [05:49]
A
Imagine so it is there somewhere. And I'm not a physicist, right? So I'm a tech guy. So for all the physicists hearing me, please, you know, take, take it to the grain of salt.
- [05:59]
B
Don't cringe too much.
- [06:01]
A
Exactly. But look, imagine that is, imagine all the, all the, you have a bunch of particles. Imagine it like a, a bunch of people together. And then if someone moves through them, that's like with glue, like glues them together, right. And makes people not move. This is kind of, this particle is something that gets things together and does not allow them to move. So it gives them effectively the weight, the mass of the, of the particle.
- [06:27]
B
Okay, and then how does it tie back to your work and what you do with artificial intelligence?
- [06:34]
A
So back then, what I was doing was a very small part of this experiment. I was doing data processing and high speed data processing. So we're talking about petabytes and petabytes of information. So when people think, oh, Google has whatever this amount of data of the world. CERN from these experiments generates a lot more data when things collide, right? Imagine 10 to the 10 particles colliding and things just popping out and then different again. Take two stones, crash them together and you'll see lots of stuff coming out. And then imagine doing this million times per second. And all this data needs to be handled. So I was in one of the teams that handled all that data. So it was more about data processing than artificial intelligence. Was all the part of handling all this data and making sure that the physicists could get access to the data and could see how the experiment was performing. So it's really, it was just a effectively support function, making sure the software for the physicists would run because the physicists are who matters at cern, because they are the guys doing the real work.
- [07:35]
B
And how would you describe yourself? So you said you're not a physicist, you studied at cern. Are you a computer engineer or how do you define yourself? Without your job title.
- [07:44]
A
So by training I'm a computer engineer, so I focus on the intersection of business and technology. So for me, technology solves a business problem. It's all about solving well at the end of the day, solving problems with people. And that's what I've been doing for many years. So when we look at computing, computing is a tool that enables you to solve problems. And that's what I do. So I'm an engineer turned business guy entrepreneur and now in a consulting company.
- [08:12]
B
That's amazing. Well, welcome. So now I want to jump in a topic that's a hot topic. When you and I spoke a couple of months ago and we talk, I want to talk about the future of work because this is what's coming top of mind in multiple business conversations right now. And there's a lot of different statements by the gut f of AI who worked at Google by other different scientists as well. So I want to have your perspective. I remember back in Covid days there was this corporate jargon sentence that was we have to learn to do more with less. And my, I remember my reps at the time when I was working at Google. I don't want to hear about doing more with less to justify the the reduced spending or the reduced this or the fact that people had to squeeze in. Now, in the world of AI where like let's call it AI powered universe, but especially in the future of work, do you really think AI is going to allow companies to do more with less? And what does it actually mean in practical terms for an organization?
- [09:12]
A
Yeah. So I think we're officially hitting a massive decoupling and because for the first time in history, the business growth will not be connected to the headcount today already you can scale your revenue by 10x. You don't even need to hire a single person, which is a dream of all CEOs, right. They just want to generate more money.
- [09:35]
B
And your margin is going to become bigger. Your ebitda, you look like a star. If it works. If it works.
- [09:42]
A
If it works. But it is working. Look, there's many examples. There's an example. There's a company here in the region where I am close to where I am in Israel. There is $6 million a couple of months back. They have three employees. They have, it's a SaaS B2B. They have 200 clients. They want to grow that by I think 10x and not hire a single person. And there's three people. Right. And they were invested by Y Combinator and one of These, you know, many of these American funds. So it's coming, it's coming. It's not only about the cat videos, right? There's a lot of stuff that today you can solve with AI, for example, support. Today you can. There's a few companies here in the region that 50% of their support requests are solved in one second, costing around 1 cent of dollar with better NPS for. Right?
- [10:33]
B
So that's like customers are satisfied people. I might not be in that pool of customers because I don't know, I feel like a lot of the chat bots and agents are still going in
- [10:43]
A
circles, but keep going now they're badly done. They're simply badly done. Because it is possible today to solve One thing that AI. The AI, let's say LLMs. One thing that LLMs are massively good at, it's extracting information, process that information and transforming into some other stuff. So if you tell me a problem, you're right. Look, what's the status of my invoice? It's extremely, extremely easy for an LLM to understand what is your query? And then okay, status of invoice, easy. I check my invoicing system and I know the status of the invoice. So it's something I can answer sub second but I need to delay it a bit so that people feel there's somebody on the other side. Obviously we still need to fake it a bit. Interactions. But you can solve it. That one you can solve and it's proven. I knew a few companies doing this. Half their support requests are solved instantly, which is amazing.
- [11:32]
B
So basically the expression of doing more with less, you're saying that right now, in 2026, it's already happening. You're with companies, you hear from companies as well, where they've been able to implement that. Do you see that there's a difference between like the SMB, the small business, the mid market or the large enterprise? Which one of these segments you think is able to do more with less, more effectively? And where have you seen the most breakthroughs?
- [11:55]
A
So the breakthroughs are in startups, are in technology companies. This company I'm talking about is, is a, is a technology company. If you look at big banks, which an area I work a lot, they're not there yet. They, they are not there yet for several reasons. Because they're scared of regulation, because banks are designed to avoid fines. They are not designed to treat their customers nicely. Right? That's the, that's the reality.
- [12:15]
B
They're designed to be compliant and they pride themselves for Their compliance. Listen, I was the managing director of financial services at LinkedIn, working with some of the biggest brands, banks and investment management firms in the world. And it's true that the fintechs were quicker to adapt, faster to move, while they were always pivoting their business model. But I think that's so important because they would pride themselves by being some of the most compliant banks globally. So for them it's very important, like analyzing the LLMs and how the models are running and privacy, et cetera. So you're right there on the money. So when you Talked about LLM vs AI, I just want to pause and make sure that we describe and define it for everybody who's listening to us. What's the difference between an LLM and an AI?
- [12:58]
A
That's a great question. So AI is an umbrella term. It was coined in the 50s and it's everything that is connected with some sort of algorithm that solves some sort of problem. And there's many things inside AI. You have the machine learning and machine learning are all these algorithms that we've been using for the past, I don't know, at least 20 years, when you can forecast cash flows, you can forecast occupancy on an airplane, and then you can price the different seats on an airplane. So this is machine learning. These are machine learning algorithms that have been around for a while. And then some years ago, when I say some, it's always more than 10, 20 years, deep learning appeared. And deep learning is neural networks. And neural networks is this concept that you try to mimic biology. So you try to mimic the human brain or a brain to some extent, where you have neurons, and these neurons, they connect to other neurons. So back at university, one of the projects I did a long time ago was using neural networks for facial recognition. So this has been around for easy 25 years. And it does.
- [14:09]
B
You're doing it probably like other people do it with like voice recognition, like with nuance, like types of companies, etc.
- [14:15]
A
But that's all the same. So that's all, that's all deep learning. And then inside deep Learning you have LLMs, large language models. And large language models, again, use neural networks, very complex neural networks, huge neural networks to process text, but you also have networks for image generation, you also have networks for video generation. So all of this are neural networks.
- [14:42]
B
It's funny because I see your explanation in bubbles and charts in a perfect.
- [14:46]
A
Yeah, follow me on LinkedIn. For whoever is listening. It's one of the posts that normally
- [14:52]
B
performs well, I think that's Amazing. So thank you for clarifying the future of work. Now I want to talk about some of the biggest executive misconceptions, because there's a lot of misconceptions happening right now because the world is shifting. We're entering into a whole new era. So what do you think? Are leaders getting wrong about AI adoption right now? So leaders in companies.
- [15:13]
A
It's a great question because I see it a lot and I keep telling people it's never about the tools. Right? And I'm a technologist and I speak against my class to some extent because we always think, no technology solves problems, let's use the technology. But it's not really about the technology. It's about what problem are we solving. And after you know what problem you're solving, what you want to optimize, what KPI you want to impact, you look at the technology and what's happening now with these mandates of AI from the top. Somebody told the team, the board told the CEO, you need AI. The CEO told the management team, you guys need AI. And then they do AI. And what means do AI? Nobody knows. Then you see, you hear from McKinsey 95 or whatever fail. Of course they fail because as you know, cloud migration failed before them. They don't fail because the technology does not solve a problem. It fails because of the wrong expectations. Expectations are not being, are not the lines because one people think it does something that it doesn't do. So that's a problem of knowledge. And the other one is people did not focus on the right KPI. So you maybe are trying to solve. Chatbot is a great example because chatbots
- [16:22]
B
work customer service, right? You want to improve the customer service, improves customer success.
- [16:26]
A
That works for sure. And you see, as you said, so many examples that fail. But why? Because they want to solve, they want to boil the ocean. This thing of boiling the ocean is, I see it a lot right in my day to day, you try to boil the ocean, but you cannot boil the ocean. You should pick a part that works and use that part. Again, like this company here in the Middle east that 40% of the query, sorry, 50% of the queries are handled and then the rest they go to a human. Don't try to solve everything today it does not solve everything, but it can be a massive, massive business upgrade. Other things that work and people don't use segments of one. Today I can generate content for my clients individually. I can generate an individual marketing email with an offer for all my clients. Let's say I have I don't know, 100,000 clients. I can generate an offer for each of them and people don't do it.
- [17:14]
B
How do you do that?
- [17:15]
A
So you know who they are. You know exactly who they are. You run. So you'll have multiple products, right, that you, multiple offers. And you know what are the needs of the people? You can just go through all of them, understand what are the real needs of them. And per each one of them, use an LLM, you know, use ChatGPT or whatever, a clause and you generate an individual email that says hello Luis, I've seen that you're, I don't know, buying hamburgers, whatever. Don't you want a discount? You know there's a very famous soda. Soda, yeah, but look, that's what Tesco did, right? Tasko is famous for this for proposing. It was 2000s and it's again AI, right? It was proposing to ladies that were not, that did not tell anybody that they were pregnant to propose them pregnancy products, right? And they got huge, huge, huge complaints that they were spying on people and they were effectively award they were using this data. But back then you had to send a bunch of coupons to a bunch of people. Today you can do those coupons individually per each person and that's extremely powerful.
- [18:21]
B
It's so interesting because Amazon does that, right? And then they have to put some layers of privacy because otherwise Amazon knew too much. But multiple companies do that as well and that become a bit traumatic because some people are hiding some stuff and then you go online and you see that people have searched for some information. So there's a line, fine line with privacy. So I think what I'm getting from what leaders are getting wrong about AI adoption right now is they're trying to boil the ocean. They're not focusing on a core problem to solve initially. So how would you start if you're, let's say you're in a mid. Because you said startups are doing very well. Let's talk about the mid market enterprise. Where, which problem would you solve first? How would you structure the problem solving at the enterprise, the mid market segment for a company. So the mid market, well, the ocean. What like where the core of the company, the, the periphery of the company.
- [19:13]
A
So it's whatever process you have and it's case by case most of these problems that we have today or that we can use AI, it's not even, you know, the fancy LLMs is robotic process automation is you take some terrible process that you have there that takes hours and hours and hours and you convert that process into using agents, LLMs, whatever to improve that process. Maybe it's something as basic as using something like N8N or make.com to build a pipeline of actions that are executed for a specific process. You try that process and you'll see that that process is going to work. After that process works, you can try other processes. And it is a great opportunity to revamp the processes because whatever company is out there, they have processes that are designed. Some of them were designed for pen and paper. Then you move them into the world of Word, Excel or Google Docs and they're still there with this process. Now we are putting it into a massive machine of automation that can automate a lot of things. So ideally you should look at the process and clean it, but do one by one. Don't try to do a bunch of them. Again, the easier ones are what we spoke about, customer support, the marketing part. All of those are extremely easy. Acquisition also very easy.
- [20:31]
B
One problem at a time and then again, double click on this. If you're an organization, what would be the ideal project team to solve that problem? Talking about A, B and C level, talking about cross functional teams, et cetera.
- [20:44]
A
Yes. So first there needs to be buy in from the management, from the leadership. Because if there's no buy in and if the leadership does not understand that they need to do it, it will not work. Right. And it's not only AI, it's anything else. Because this needs resources, then it's cross functional teams. You need people that understand the business, the business process, whatever is there, that is what generates money for the company. And you need technologists that understand what are the limitations of all of this. And someone, if there's nobody in the middle that can speak the language and you need some product person in the middle that can actually speak the language of the technologists and the language of the business stakeholders. But the teams need to be cross functional. There's no way that in this day and age you can operate successfully without a cross functional team. There's too much complexity in all of those things.
- [21:27]
B
Yeah, yeah, no, I think it's. And what about the end user, like the first the employees and then also the external customer. Would you integrate them in the process?
- [21:37]
A
So for sure, the employees. Yes. There's this concept of citizen developer which is enabling everybody in your organization to develop the tools that they need. And today the magic of, you know, this lovable, the replit and all these tools is that easily someone in an organization, if you give them a tool that is sanctioned by it, by security that is connected to the necessary data sources internally. They can build their own tools. So I see a lot now business people going into a tool like lovable and building their own page that does analysis of, I don't know, invoicing or revenue or something that you were not there before. So there's a massive opportunity here to distribute the work of developing tools to all these teams to do that, obviously you need to be very careful how you deploy these tools and you need to train people because you cannot expect people to start using this technology that is potentially the most advanced technology that we've created in the last hundred years because it enables you to do everything and magically learn it right. It needs. People need to learn. You need to, you need to.
- [22:45]
B
I have a follow up questions on that because there's this whole vibe coding movement and some of those tools, the replay, it's the lovable, etc. It enables you to vibe code. But can you really like start with these types of solutions and interfaces and go deep into creating a fully pledged, fully fleshed product that's workable, etc. In your opinion?
- [23:08]
A
It depends what means the fully fledged product and depends on who is on the other side of writing these tools. If it's a developer, a person that is a geeky to some extent and likes doing Excel pivot tables and whatever else, or if it's a person that does not understand technology at all, a person that does not understand technology at all, I think is going to struggle. There's like the basic, you know, generate a page and put some buttons and some stuff. Yeah, everybody can do that. But it gets to a point today still you need to understand code or at least you need to understand the logic of how these things interconnect.
- [23:40]
B
The expected behaviors, the databases and the outputs.
- [23:44]
A
Okay, you need the basic stuff because if you just wanted to solve, you give it one line and expect miracles. It's never going to work. If you're a bit technical, it will make your life a lot easier because it's going to enable you to solve basic problems. Imagine just a form to collect some information and that form has some fancy validation that you cannot do with Google forms today. You can go and build that. Look, for example, I didn't look at the code. I built my qualification form for my newsletter. In the in case with cloud code,
- [24:17]
B
we got to talk about cloud code because everybody's like, ah, it's changing the world. So what's all the hype about cloud code? Like you can do everything now.
- [24:26]
A
Yeah, you can, but you can also do it in Lovable, because Lovable also uses Claude on the back. So, you know, it's. People get focused on the tools and they. And they forget. It's not about the tool, it's about the problem you're solving and which problem you're solving and then you find a tool to solve the problem. Cloud code is amazing, but you need to understand a bit of code or else you're going to do many mistakes. If you want to use Claude to build a website, just use Lovable Replit or one of those. There's also Cloud Cowork, which is another one. It feels like I'm doing advertisement of Entropic.
- [24:57]
B
I know we have not been sponsored by Entropic. Let's pause.
- [25:00]
A
But if they want to sponsor us, I think they could.
- [25:02]
B
We have to navigate all those sponsors because it's like, oh, I just got sponsored for this. But then it means can I still promote that? I like them all, I like them all. So I think let's just to go back to basics. There's been a lot of hype about CloudCard and we're not trying to promote Cloud or ChatGPT or Grok or Gemini, but I think the idea is that there's been a lot of buzz about cloud co op and cloud coworks. I want you to define for the audience what's the difference between Cloud, Cloud, Cloud Cowork and then other similar solutions that might not be in this conversation right now?
- [25:36]
A
Yeah, yeah, yeah.
- [25:37]
B
We want to create an even playing field because we're not sponsored by them.
- [25:41]
A
So these are all agents, right? And so you have the LLMs, right? You go on a website, you write stuff, it gives you stuff back. Basically you have Grok, which last time I looked at the, at the statistics was the worst one out of the western ones. Then you have Claude and Gemini and Chachi, PT and Mistral. Those are the top ones.
- [26:04]
B
Where do you fit? Perplexity and Abacus, AI.
- [26:09]
A
Yeah, so those are agents that use these models. So when you go on perplexity.com or AI, it's again a chat interface. But you can pick the model that you're using and you can pick Gemini, you can pick Claude, you can pick. Honestly, I don't remember what's there. There's a model from them called Sonar also. And what it does, it searches information on the web, it gives it to a model and it gives you an answer. That's what it does. If you go on Gemini. Gemini also has Gemini the model and Gemini the agent that does deep research. So when you click deep research in any of these things, it's always an agent, so it's always an LLM that does actions for you. Normally searching the web, getting articles and giving a summary of those articles. So for example, producing information about what is the best LinkedIn article I can write about AI for tomorrow they can go and search the web and come back with suggestions.
- [27:01]
B
So the agent is going to organize the world's information for you.
- [27:05]
A
Correct. It can also do more stuff. Right. It can go and connect to your Google Docs and read the Google sheets and get information out and make a summary. Or Manus is very famous for generating presentations. You tell it to go and search whatever on the web. It comes back, it connects to Google Docs or something and generates presentations. So it does a bunch of stuff. Right? Whatever tools they have access to, they can all do. And then you have these tools like Claude's codes and anti gravity and codecs from OpenAI, which are agents that are applications, slash agents that run on your own computer. They're not on a browser, they are on your computer and you normally use them to write code. You can use them for more stuff, you can use them to write articles, you can use them to help you write LinkedIn posts, you can use to help them. Whatever is text generation they can do for you. And you ask them to do stuff and they do stuff for you. So Anti Gravity from Google, they released it a couple of months. A couple of months ago. It's kind of a mix between. And then there's so many tools. There's Cursor for developers, there's Cursor and there is winsurf and these are editors that you can speak with agents and write code. Then Antigravity is kind of one of those, but more advanced because it has a nicer interface. But again generates code. Claude Code is a tool you run on the console, so it's terrible for a non technical person to use that generates code for you. Claude Cowork is a tool that runs cloud code inside a NASA application that can connect to stuff on your computer to do actions for you. And those, these are all agents. These are. When we say agents, people use the word agent too much because agent can mean many things. Agent in theory should be total autonomous AI that does stuff for you. And that's not what we have today. We have like pseudo agents that will tell it, go do something and they come back with a result and then you realize it's correct or Incorrect. And you need to tell it, look, fix this, fix that. And this is how you work, right? If you work with Claude, cloud code, cloud, cowork, anti gravity. I use all of them. And whenever I'm writing code, I tell it, look, this is what I need to write. This is my idea. Go and do it. It goes and does it and then you check what it does. And sometimes it's not good and you need to tell it, look, this is wrong, that's wrong, that's wrong. And then it fixes itself, but you need to interact with it. So it's like a, you know, it's like a junior person on the other side that you're interacting with this person all the time. And this is valid for everything. It's valid for writing code, for analyzing pivot tables, for, for doing financial models. Claude has a lot business plans.
- [29:39]
B
Oh my gosh. Trying to give them a business plan, it starts hallucinating. That's amazing. I want to go back into the topic of like the future of work and skills. So what's one of the hot topics we talk about? And you know, there's a lot of different conversations, but sometimes people talk about the fact, hey, your friends were like blue collars or running H vac companies have more job security than you. So white collar corporate person with a master's degree. My question for you is like, what skills do you think I quietly becoming obsolete right now because of AI?
- [30:12]
A
Yeah. So all the stuff that is, is executing tasks is going to become obsolete and obsolete very quick. You need to focus on, on the, on what drives business value. So it's not about task, it's about what role you have and what, what your role enables. You cannot focus as a technology person. Developers and developers, they will hate me saying this, but you cannot focus on writing the code. What you need to focus on is orchestrating these agents with a nice architecture to write the code for you. Designers, they will not be. And I have friends that are very good, experienced UX designers that today they use UExpilot or Figma make to generate designs. You go there and generate a design so you don't need to do copy paste. You, you generate the designs for a bunch of ideas you have and then you use your brain, your experience to pick the options that AI gave you and to develop them further. Marketing people, you, back in the day you'd have to generate a marketing copy. No, today you ask it to generate marketing copy. You look at it and you do your role using whatever it gave you and you work on it, right? You pick. So what is important is the skills for the future. You need to know how to use AI. You need to understand when AI is getting crazy and is hallucinating. And you need to obviously address that. All the tasks are going to disappear. All the copy paste stuff. When you spend hours, you know, in Excel doing a pivot table or doing an index match to go and copy data from one table to the other table, that's gone. If it's not done today, it'll be gone in tomorrow, it will be gone in six months. So when you say that people with masters have a problem. Yeah, they do, they do. Because we won't need as many people as we need today. And you can see it, right? If you look at the especially, I mean, software, look at Tata, Tata Consulting in India. They're a massive company. They fired, I don't recall the numbers, but I think like 10 or 20,000 employees just a few months ago. And the CEO said, oh no, we're not going to fire anymore. Which is simply not true. Because if you read in the Financial Times, they effectively fired between 50 and 80,000, depending if you account for what they said they fired plus what they did not say, but they were forced resignations. So this consulting model, you see all the consulting companies letting go a lot of people simply because today I do not have to have a pyramid of a lot of junior people and a few partners where all these junior people, what they do is copy paste data from past Excel from other clients and then generate a new Excel for the new client and some new PowerPoint presentations. Because of that, you go on Gamma and Combo does it for you, right? Or you go in Gemini and it analyzes the data.
- [32:53]
B
It's crazy. You can do your profit and loss statement. I mean, you need the judgment of a human. And I think that's where the layer that's interesting is like, how does it evolve over time? What is. What are my thoughts about what it gave me back? They don't have the perspective of how to build a business in the long term. But I think they're just the basic computing of like, here's the data, give me a profit and loss. What are the red flags? What are the ideas? And I think then you have a model to play with. I think that's super exciting. What you're saying in terms of like how people would need to think deeper and even create guardrails and figuring out how things are going to evolve with time. I think, what do you think are the skills that are being Undervalued today that you think will become critical in the next three to five years. Again, we talked about my H Vac friends, right, that have more job security than maybe the junior analysts. So what are the three to five skills you think are going to be even more competitive for people who want to get ahead of the curve?
- [33:51]
A
Yeah. So I was having breakfast with a friend yesterday. He's a recruiter and it's a very tech recruiter is surfing the wave. And he was telling me back in the day, I used to have 25 people in the UK, today I just have my company. He moved to the UAE two years ago, three years ago, and he started from scratch. And now what he realizes he does not need to hire 25 people because today what he needs to do is he needs to go and meet the cto, CIO of big organization and understand what they need. Then he needs to go and find whoever are the candidates. He doesn't do it. He asks an agent. So now he has a nice agent that goes on LinkedIn recruiter and filters for him and gives them a list of candidates. Then he filters those with his, you know, what he knows about recruiting. And then he does the call. So he does the call, but he does not schedule the call. Another agent sends the emails back and forth. Done. Then he does the call with the people because he wants to understand who is the person on the other side. And I cannot do that. AI for more than you can say. I can interview the AI doesn't work.
- [34:50]
B
But don't you think there's going to be some bias? The agent is going to be scraping LinkedIn and some people don't have optimized profiles. Some people have like, I don't know, different nationalities, women, etc. And the AI was going to apply the bias and say, no, these candidates are brushed off and those candidates will be better based on some very biased like fact that most of the people developed AI or don't look like me.
- [35:15]
A
Right, That's a great question.
- [35:17]
B
Because there's sexism, there's, there's all those things. I know because when I compute stuff I'm like, why is my image always very more inappropriate? Why do some behaviors like associated with certain ethnicities And I see it because I try to animate videos and things like that and it's like, really? That's the best you could give me? It's full of stereotypes. So how do you trust the agents to be able to hire based on the real skills and capabilities?
- [35:47]
A
Yeah, no, it's great. It's great. So first, you're very right. The AIs and this is a massive problem that we have in the Middle east that exists in Africa and exists in many countries in the world, which is the AIs today are either Western AIs trained on the ideas and concepts and culture of the west and a very specific type of the west, by the way. It's all these weird components of society. So Western, educated, white, and two other letters that I don't recall what they mean. But most of the AIs in the west are trained on this very specific part of society, Africa, there's no proper models. Middle east, there is a few, but not proper. Then you have China and all the other countries don't have them. And it's a massive problem. I fully agree. I don't have a solution. I think more people will need to train them, but they are very expensive. On this specific case, what he does, it does no bias, because what it says is it's very mechanical. It has something that goes and clicks on LinkedIn Recruiter and tell it, I want to find computer engineers. This was this example yesterday. I want to find computer engineers from Oxford, whatever, whatever university and that have been working for the past three years in something and whatever else characteristics. And then the AI just executes that. So instead of being him, you know, clicking through LinkedIn, doing the filter, clicking stuff and taking, I don't know, two hours going through CVS or profiles on LinkedIn, it's an AI that goes and searches for what he asks. So that's the power of it. Is there bias? It's possible. It is. It is very much possible.
- [37:15]
B
Let's talk a little bit about future proofing. So for the professionals who are like, listen, I'm going to fight this, I'm going to evolve with the times. So if somebody wants to be in the top 5% of their field in the next AI wave, what should they start learning? Like today, right now, Today, they should
- [37:32]
A
stop complaining and putting the head in the sand that many people are still doing and they should just go and start doing it. The basic stuff to do is, I don't know, I don't know what the person is, you know, I don't know. Copywriter, Just go and try the tool. Just try it. Try to do your workflow, Pick a workflow that you have and try to ask one of these tools how to optimize it. I bet that one of these tools is going to tell you, use this tool, use another tool, use another tool to do it. I Am sure. Because these tools, they all, they're all going to, you know, they're going to reference other tools in any case. Right, Go on. Perplexity.
- [38:05]
B
Step number one is optimize your workflow.
- [38:08]
A
Optimize your workflow.
- [38:09]
B
Become better in your workflow. What you become more efficient. What's step number two?
- [38:13]
A
You need to go and search for information. You cannot expect that your company. Because there are many companies that the learning and development departments, they are simply far, far in the past and they did not realize that these things are needed. And you cannot expect the companies to do it for you. They should, but they didn't before and they're not going to do it now considering the speed that this moves. So then again, search for information. And it's very easy to search for information. Go on YouTube, there's a bunch of free videos everywhere. Go on LinkedIn, a bunch of people are posting about AI every day with links to courses. Not trying to destroy the course, you know, the course people. But there's many free ones. So start with the free ones. If you're a designer, Google has a
- [38:53]
B
certification, it has mit, Stanford. There's so many.
- [38:59]
A
I have a friend doing the MIT one, he's very happy with it. It's expensive, so it's a bit pricey, but it's valuable if you're a business person. Like if you're a, you know, C level or senior senior manager in a, in a big organization, go and do these things because you will need them. And if you don't have the time to spend hours and hours per day like I do geeking about technology online, you just go and need to learn. So basically it's. I think those two things are essential is you're. Well, you need to understand how to optimize your workflow. And then second, you cannot. It's continuous learning. People have said for many years we need continuous learning with AI. This is very clear. That's the only way that you're going to, that you're going to survive.
- [39:37]
B
Is there a section of the AI curriculum that. Because it's very, very. It's big, right? So what should they learn specifically? Like navigate creating agents, navig orchestrating re engineering processes. What is the skill set that they should.
- [39:51]
A
It depends. Right? Because if you ask me about what the UX designer should do, I think they should go and learn how to use vexpilot and Figma make and stop doing lines themselves. If you ask me what the developer should do, I'll tell them go and figure out how to build your code with high quality in cloud code or anti gravity or Codex or Copilot.
- [40:08]
B
What about project manager? What should they do?
- [40:11]
A
Project manager if they are doing all these documents, these crazy documents that, those crazy reporting documents that I see my colleagues doing, just start using cloud or one of these tools to help them with those, those documents. Again, it's workflow is you need to find ways to stop your tasks, whatever tasks you are doing that takes you forever, stop those tasks and find a way to, to fix that task and to let somebody, somebody in a sense an agent to do being a perplexity, being it, whatever, whatever else.
- [40:39]
B
So you should be winning time and having more light balance because you have a well, a good team of well orchestrated agents that should be an extension of yourself. So that would be like the aspirational goal for every professional get. So get your workflow so optimized that you're not like working I don't know, 10, 12 hours days, 16 hour days anymore. So. But there's going to be an upfront work into optimizing your own workflow for sure.
- [41:04]
A
But like people if all these professionals that spend years in university learning stuff, learning how to learn because that's what university is for, is to learn how to learn. Because most of the stuff you learn that is kind of useless. So they should, again, they need to go back to it, right? They need to learn how to learn, learn how to use this, learn how to use these new tools and it's for sure going to save you many hours. For me, it saves me so many hours in like writing code. It's crazy.
- [41:28]
B
So let's think about like I know you're not a marketer, but people who are in marketing, right? The CMOs are like we're being asked to leverage AI throughout all of the marketing to make marketing more efficient, increase customer lifetime value, do more with less, less dollars, more impact. So if you were the CMO of an mid market company, how would you organize your marketing for optimal sales revenue margin? All of the above.
- [41:53]
A
I'm definitely not an expert. We did spend hundreds, thousands of dollars per month on ads on Google and Facebook back in my previous life. Look, it's at the end again we go back to processes. Are you generating all this copy and all these landing pages, how are you doing for example online, right? All these landing pages, are you actually doing them by hand? If you are, it's a mistake. There's a bunch of tools that do this for you. So find tools that solves whatever you're doing. The landing pages for. So you're running campaigns, right? And the campaigns in normal you land in a landing page that you optimize today you can generate 10,000 landing pages. You don't need to test two, you can test 10,000, test those 10,000 at the same time. Then you're going to know which copy, which button color, which positioning of the form works best. So that's, that's a, you know, that's a basic one that you can definitely do.
- [42:42]
B
That's conversion optimization, correct?
- [42:44]
A
That's conversion optimization. Then on the, on the communication to your clients, you have your nice newsletters, right that you, that you send to send out. If you're sending one newsletter and it's the same newsletter, you're you're missing out a massive opportunity because if you had qualified all these people that are in your newsletter, you know what they're interested in. You know if they're CEO, if they are CPO or if they're a student, if they're whatever, you in serious should know what they want. So you should send them a different newsletter and different offering.
- [43:12]
B
So what tool do you suggest for customization of the newsletter and then for conversion optimization of landing pages? Do you have any tools that you would suggest?
- [43:22]
A
So I use Kit. Kit is very good. Again not sponsored but Kit has a nice four newsletters, has all these flows that you can, that you can filter by categories and then you can give different information to, to different people.
- [43:36]
B
Do you have an agent plugged into your Kit.com?
- [43:39]
A
no, but I get my data from Kit.com, i get my data from authored up another tool since we are anyway promoting tools and I connect that data in a nice pivot table in Google sheets. I take that into Gemini to do the analysis. So I don't do the analysis anymore and it's amazing.
- [43:56]
B
I love the conversion page optimization for marketers. I think it's very actionable. I like also the one to one targeting that you're talking about leveraging whatever it's kit, mailchimp, whichever solution doesn't matter but making sure and then the computing the analysis for your one to one segment marketing and of course we can talk about CRMs and making sure that every like it's we talked about one on one marketing and I think this is the holy grail of what's possible and then figuring out how to increase lifetime value based on what the one person like marketing to one person. So I think that's great question for you. I want to go into the rapid fire questions which is going to wrap up our beautiful episode. And I'm super excited. Thank you so much for making the time in LA. It's 7am now. We did it. I snoozed for like an hour and a half. But it's my first very early interview because of you. Where are you based now? Are you in Dubai? Are you Israel? I never know where you are. Portugal.
- [44:49]
A
I'm in Dubai. I've been here for. So I lived in Dubai for what, four years now.
- [44:53]
B
Where were you before that?
- [44:55]
A
Oh, my God. I was in Nigeria for a long time. I had the business in Nigeria. I was in Myanmar. I also had the business in Myanmar. I was. I know. I keep seeing that face when I say these things.
- [45:07]
B
That's amazing. You're like a global citizen.
- [45:12]
A
I lived in many countries. Yes. After cern, I lived in many countries. I was lucky enough for my business partners to ask me to travel to many places. So I ended up traveling. So that was positive. Yeah. So I lived around. I was living in Kazakhstan. I had business in Singapore. I've been around.
- [45:29]
B
Oh, my gosh. And it was just a fun question. I was not trying to trick you. It's because every time I talk to you somewhere else. Okay, rapid fire, top 5%. What's the biggest lie society tells you about success in the AI era?
- [45:44]
A
They tell you that you're going to be replaced by someone who uses AI. If you learn AI, and that's not true, you're going to be replaced and you need to find ways to keep working.
- [45:58]
B
So you're going to be replaced anyway. That's what you're saying.
- [46:01]
A
Many, many of us are going to be replaced. And it's not me saying, look. If you look at the people from Anthropic. If you look at I want musk. If you look at many of these people, they are saying it and it's. And it's clear it's going to happen.
- [46:15]
B
Yeah. And then humans are going to go live in the forest somewhere and just
- [46:18]
A
enjoy life, you know, not. Not work so hard. Have agents do it for you, you know, like Wall E type of thing. Maybe, maybe not. It's not too good, but.
- [46:26]
B
But, yeah, I don't know. I don't know. Second question for you. What's this one skill that will compound the most over the next decade?
- [46:34]
A
The capacity to speak with other humans and understand what they need.
- [46:39]
B
It's kind of empathy. Customer success. Number three. What is one lesson you wish you knew when you were 20 about the future of work?
- [46:49]
A
So back then, it did not make a difference. Today makes a massive difference. Don't learn tools, don't focus on tools. Focus on what solves the problem and how to solve that problem, because that's what matters. The tools will change. Especially today, you understanding the problem and what are the KPIs and the drivers of success. That will always work.
- [47:11]
B
It's amazing. Question number four. What's one thing the top 5% will adopt early that the bottom 95% will resist?
- [47:21]
A
It's happening today. You see it. AI. AI is being used by 800,000 active users on ChatGPT and they're not there all the time. So if you think the world has 8 billion people, that's 10%. I bet that less than 5% are using AI on a daily basis.
- [47:38]
B
So jump on it. Figuring out your workflow, figure out the logic. So what's your cheat code to stay ahead of technological disruption?
- [47:49]
A
I'm lucky to be a technologist, so I spend too much time, maybe too much time reading a lot of stuff. I spend my day reading stuff about technology. So I keep up to date, I guess more than average simply because I like technology and it's my life.
- [48:07]
B
Well, thank you so much, Luis, everybody. You can follow luis Rodriguez on LinkedIn. So thank you so much for coming to the top 5% method. I'm excited for us to be able to talk a bit more about this topic online and help educate everybody as we're shifting into a different era.
- [48:28]
A
Sam.