#268 AI Evolution: Disrupting SEO, Transforming Content Creation, and Redefining Work with Alexander De Ridder

0:00:02 - Mehmet
Hello, I'm welcome back to a new episode of the CTO Show with Mehmet. Today, I'm very pleased to have with me Alexander, joining me from Texas in the US. Alexander, thank you very much for being with me here. The way I love to do it, I keep it to my guests to introduce themselves because I think this is the best way to do it. So the floor is yours. 

0:00:20 - Alexander
Well, thank you. Well, I'm Alexander  De Ridder. I run an AI startup company called SmythOS a SmythOS with a Y and before that, I built some marketing agencies heavily powered with AI technology. We had successful exits on those companies. Before that, I was a fractional CTO and a four hire consultant, and I have been working in the field of machine learning professionally since 2008. And before that, I studied computer science in Belgium, where I grew up and was born. I thought I'd give you the Benjamin Button version backward. 

0:01:11 - Mehmet
You did perfectly well, alexander. Actually, maybe sometimes people see it as a traditional or, let's say, very frequent question, but I love to know also more and understand more. So what brought you to this domain and why did you choose AI now, recently, as your latest startup? 

0:01:35 - Alexander
Yeah, I mean I have big convictions about technology and where things are headed. So the first time I started working in machine learning, it was just because it was the only way to achieve a certain objective at that time. It was for digital kiosks and digital science to recognize the audience that they were interacting with gender, emotions, things like age estimation, things like that to make the experience more secure and reliable. Some of that technology we found in airports. Now and when 2012, deep learning started becoming more popular again thanks to CUDA, which allows you to train neural networks on consumer level graphic cards. Well, for me, that changed everything, because I saw that neural networks could replace I mean, years of my work relatively easy. If you had the data. You could replace all that very advanced machine learning feature extraction that we had developed in a relatively short amount of time. So I got very fascinated by that. Now I'm also quite an SEO expert, not as much in other marketing channels as having specialized in Google, that cat and mouse game with Google that everyone talks about gaming the system. I've been playing that cat and mouse game since 2004. And so by the time neural networks came around well in popularity, because they've been around since the 1960s, then I immediately knew this was an arrow in Google's kit, this that would be too good for them to resist using, so everyone was constantly trying to game the system with backlinks. But neural networks would give Google a new way to evaluate the quality of content and the relevance of content, and so, with Google, it's all about serving links that answer and meet the user search intent. So when I came to that realization, started enterprise SEO agencies and agency had one agency that was not for enterprise, which I sold, but the understanding was critical SEO would completely be changed due to neural networks. And so then, in 2016, google came out with rank brain evaluating. Basically, it's like a reinforcement learning with human feedback loop, where Google takes the data from Chrome browsers, android, google usage and tries to put Humpty Dumpty together if your content is actually satisfying users who search for a certain key phrase in a given context. And so that just led to me developing my own AI to reverse engineer that. I created sparse embedding technologies, a few patents around this, and yeah, and then continue to work in AI, my current startup, and so these have been really successful. 

My current startup is SmythOS, and so the idea around this one came in 2023, this year in March, when GPT-4 was released very first week, we instantly explored its capabilities and the brand new capabilities it had that GPT-3 didn't have at that time was the ability to call functions with JSON, knowing when to ask for help, and that was, to me, that would change the world. I instantly recognized that that would change everything we know about marketing again, that it would change how we do work and it would have huge societal impact. We ended up starting with our engineering team. We started building our own internal version of auto GPT before auto GPT was released to the public in the open source domain and came to the conclusion that it was not ready for primetime. So we didn't take that further, and the reason was that multi-agent systems are only as good as the agents that are in it, and it's the same thing if you make a company from toddlers versus a company made of experienced business people. You'll have very different outcomes. 

So at the time they launched auto GPT, it got really popular really quickly, mostly around the idea that multi-agent systems and agent systems would revolutionize the world. But it was an idea that was not ready for primetime, and so what we did this year is we built a lot of agents ourselves. We were implementing believe it or not our own version of Langchain as well, until we saw Langchain got very popular, but we noticed there were severe shortcomings. It was very hard to. It was not accessible. People needed to know how to code a lot Training and maintenance was just hell. It doesn't work with multimodal systems. It doesn't incorporate some of the best ideas that have been going around, like multi-agent orchestration and so forth. Now here we are. We're now almost December 2023. 

And if you look at the landscape, agents have just been popularized with make your own GPT, but less noticed in the news is Microsoft released their Copilot Studio, which brings agents to all Microsoft Enterprise team chat, which is a game changer, and brings it to 1.4 billion Windows 10 and 11 users as well, which is a game changer. So I have been on record on stage keynotes, podcasts, everything. I have been saying that agents would be the next phase of how we use AI and that next year will be the year of the AI agent. This is going to fundamentally reshape how companies reach their customers. It's going to fundamentally change how we do work in our companies and it's quite exciting. So SmithOS is basically the first enterprise ready tool today that can let businesses start building these experiences of the future. 

0:09:19 - Mehmet
Wow, you discussed many things and I will maybe come to each one of them, alexander one by one. 

0:09:25 - Alexander
Right, so or whatever you're interested in. 

0:09:29 - Mehmet
Yeah, no, no, no Interesting topics, but let's start with where you started. Now some people are saying because I want to discuss this and then move to the other topics I see a lot of people on different platforms, whether on X, previously known as Twitter, or other places, saying that because of AI SEO, as we know, it is going away. Is that true? Is that the case? 

0:10:02 - Alexander
Okay, the truth and short answer is no, but it will be radically different than what you think is SEO today. So let's just start with the pre-agent era of this answer, because there's a pre-AI agent and a post-AI agent answer to this question, right? So in the pre-agent era, ai explodes the content supply exponentially, whereas the demand well, people are not exponentially making babies anymore, so it's easy enough to see that that creates a content gap that is exponentially rising, right? Okay? So let's start with the pre-agent era. What that means in a free agent era is that, as AI proliferates, content supplies reach such incredible heights that it becomes impossible for people in the limit to be found by their audience. Unless you are an incredible brand, an incredible authority and, just like anything else in the world, the 1% ends up owning the vast majority of attention, leaving the rest of us basically almost impossible to break through. That is the pre-agent world and that is simple mathematics. Ai creates an infinite content supply. That means that it is harder for you to get attention because the competition has increased so much. In the post-AI agent world, that changes again. So in a post-agent world, everyone has a Siri or a Google Assistant that is so ridiculously good at helping you organize your calendar, your email, your drive, your meetings, your home internet of things, your transportation, your questions, your shopping. These AI systems are so good that you're going to start interacting more with them than with the traditional Google search page. Now, in this new world, it's not just Google, it's also Amazon, it's also Microsoft, it's also Apple, it's Meta, and all of them have AI systems and there is a just like the browser wars, there's an AI assistant war going on. 

Eventually, that may coalesce again around a clear winner in the market, but for the next, I think, five, six years at least, we are going to see SEO shifting to an omnichannel organic approach. It's not going to be like it was in the last 20 years, where Google was king. You're going to have to reach your organic customers across many, many assistance AI assistance, that is and so the way you're going to do that is these AI systems. Do not have the time or patience to, in real time, go browse your web page, as customers have real-time information needs. Some of it will still index the web as we know it, but most of this is going to shift to brands, hosting, plugins, tools, skills whatever you want to call it that the AI agents can in real time interact with, for example, to know if your electric car is charged. That's not something you can know by simply indexing teslacom. You have to somehow talk to Tesla and that happens over API. 

So this is what we mean when we say that Web 3 becomes executable. We mean that websites can get stuff done dimension to it not just read or write, but also executing things. And obviously people also call it the interactive web. That's what I called it back in spring of this year, because the effect of it's being executable is that you can now have conversations with brands rather than just scrolling through and finding the right page and understanding information, processing it yourself In a TikTok generation. People do not have patience for such an organic future. They want to interact with content interactively, not just spend 10 minutes finding the right pages and then processing everything themselves. 

0:15:01 - Mehmet
Yeah, I discussed it offline with someone because I was saying if people keep doing what they used to do, the same way from content perspective, they will not reach any place. Because you have so much content, I mean relying only on search to find this information will become hard and people would need someone to go and summarize that for them. And this is exactly like it will be the bot or the agent doing that. And to your point about the agents, I remember I was doing solo episodes when I started the podcast beginning of this year, the moment I wrote auto GPT. You know like it, my podcast. 

You know it skyrocketed all of a sudden because it seems like everyone was searching what is auto GPT is about and you know like it was the game changing from concept perspective. But you mentioned that you believe next year it will be the year where we will see more of these autonomous agents. Now the question is how good they would become, because you said like, yeah, still it depends on the model behind. But from all the things and I know like things change from day to day, as you were discussing before we start so but what are you know now, the expectations from, from the autonomous. 

0:16:23 - Alexander
I have a different, I have a different view on this. I believe a GI has already been achieved. But here's the thing the internet is the AGI. So you think about it, the internet is always learning, always iterating, evolving. It adds new functionality organically, and the functionality it already has keeps getting better. So what is lacking in the world is orchestration of that intelligence, the harnessing of that intelligence. So there's this movie my grandmother used to like to watch, called Black Beauty, but a black horse. But it could not be tamed, nobody would be able to write it. What good is it, that beautiful, executable web, if you can't orchestrate it, if you can't write it, if you can't harness it? So the the starting with GPT for and actually retroactively function calling now even works with 3.5, but really starting with GPT for the world was first introduced to this idea that intelligence could be orchestrated. 

Now I'll just go really quick on a philosophical, biological tangent to understand this. Okay, so when we are born or conceived, rather, we're just one cell, and that one cell contains the information for every single part of your body, the DNA, right? So the epigenome is like the orchestrator that plays piano on a keyboard. You have all the keys, your DNA. But then what notes are being played? That software? Okay. So same way, that same way. The internet needs to be orchestrated and be used and harnessed in different ways. Our cells are basically genetically identical. They're like clones of each other, but they specialize completely differently depending on how the music is played. Okay, so what we're basically seeing in the human body right, is a collective intelligence. So all of our cells are collectively combined, working together to create what we say is the human experience. It is not like one monolithic entity, like a brick, like this. Human is a brick. Every cell in my body changes over time, it duplicates its split, and so I am a new person all the time, but I'm also still myself. So the being yourself comes from that orchestration, that collectiveness. And then let's zoom out. Let's zoom out from that and let's just look at humanity as a whole. Even if you were to think of a human as a single entity, right, not a collection of intelligence, but as a single intelligence. We all know our brains are neural network, right, they're neurons connected to each other, right? But even if you would think of a human as a monolithic intelligence, let's just say that we have, I don't know, the scientific understanding of the 1500s or 1400s of the common people, right, and we just see humans as this, like solid entity. 

Well, if you would take a human outside of its environment, into a jungle, let's say, without tools, without a network, without people to rely on, what could that person achieve in their lifetime? Would they ever ever build a skycraper? Would they ever figure out how to get an airplane off the ground? I mean, it's unthinkable, it would never happen. So, because they would be too busy surviving, right, they might not even survive a week. If you take me out of my habitat, I'd be pretty weak In this habitat, where I rely on people making food, people building things, and I can just be, you know, a keyboard monkey all day right. I am strong in my habitat, but take me out of it and on my own, without this collection of humans to support me all around the world, I am nothing right. And so the same can be said for Elon Musk and Steve Jobs and Jeff Bezos and everyone that we so praise as like this, some genius humans. Without their support networks, they would have never achieved what we praise them for. 

So, even if you see humans as a monolithic entity, we're still nothing in terms of intelligence, we say, oh, look, what humans have achieved compared to animals. Well, we just have been able to orchestrate our intelligence better by collaborating, and obviously, individually, we have a stronger set of hardware, so to speak, for that intelligence. But collectively, that is really where we see amazing things happen. And so the same thing is true, I believe, when it comes to artificial intelligence. Right, imagine that the next version of GPT or Google's AI or something, let's assume that it achieves artificial general intelligence. Okay, great, you've now got an intelligence that's as smart as my mother, let's say, and sits in a box somewhere by herself and can, I don't know, write poems, draw, like that has no impact on society if it's isolated from everything else in the fabric of the world. 

The only way that becomes valuable is if you allow the AGI to use all the tools that humans have made and you retrofit them so that the AI can use them, which is the executable web, this web three. You give them access to all the AI models, the APIs, all our data. You showed them your workflow, the recipe, how Coca-Cola is made, so that they can actually go make Coca-Cola. So what you do is orchestration. It's how intelligence really works Without orchestration. You know this is called AGI. In a box on its own it's really like just a novel curiosity. It's like a human in a jungle without support network. So the real impact from AI is not just the foundation models, it's that next level of abstraction that comes on top of it, the way that they can orchestrate intelligence. 

0:23:38 - Mehmet
So the question, alexander, that usually people ask is what we will be doing if we have and you are saying we have it so how this will affect the way we work, the way we do tasks. 

0:23:54 - Alexander
Yeah, what's waiting? 

0:23:55 - Mehmet
us. 

0:23:56 - Alexander
Yeah. So today you need to and can. With Smith OS, you can do it yourself. With code, it has a lot of drawbacks and can take a long time. But today you can take your workflow, use a foundation model as the orchestrator of that workflow and pull again the different AI models video, text, audio, images, everything tie it to all your data sources your CRM, your documents, your images. You put it all in a database accessible for the agent. You bring in all the APIs that it possibly could use to accomplish the task and so you map the workflow out just like you would onboard an employee. 

Same way, and with Smith, you do that without codes. You do that with you know, and with Smith, you do that without codes. You do that with you describe your task with English and you drag and drop all the workflows together, and so the AI is constrained to do what the workflow that you have designed it to do, which means it is safe and auditable. We call this constrained alignment. It's another approach to the alignment problem, not from a monolithic AI foundation model perspective, but from a orchestration, collective intelligence. That's why I wrote the atturing test for collective consciousness around that, to measure how those systems interact and how good they are and safe, and so today you can do that, you can orchestrate this, and so what happens tomorrow is the new foundation model comes out, new API comes out, better processes understood by a company, better data is obtained okay, that's the trajectory of humanity. That's the living, breathing internet that keeps getting smarter right, this collective intelligence that we have, where all the data is there? Okay, what happens? Your agents basically become more capable. It is not going from now on. It's not a light switch that all of a sudden turns on. I mean, this is already here today, and so how is it going to impact you? I made a LinkedIn post this last week and the last seven days or so, so by the time this airs, maybe three weeks for the audience, I wrote this right Boring work will be automated, so you'll just won't have to do it anymore, and exciting work will be collaborative. You'll have help. So that's the future we're going to live with. Now. 

There's plenty of people who don't love doing boring work, and that includes me, and it includes my children and includes my family and my wife and parents and every colleague I've ever worked with, even computer scientists shocker don't love building yet one more login page and authentication. If you can avoid it, why even build it right? What we love doing is orchestrating things. That's why there are platforms like Angular and React, because why would you want to build all that stuff from the ground up every single time you have a new project? It makes no sense. So this is just the next evolution, in that we're just orchestrating intelligence. 

And how is the society going to change Because of that? Well, pretty dramatically. First of all, there's this thing that I have been calling the great reshoring, which means a lot of the offshore work is going to be pulled back into countries. The reason why they offshored in the first place is for cost savings, and now they can bring those cost savings in-house without having to rely so much on external resources for knowledge labor. And then you may wonder well, what happens then to all those freelancers in all those countries around the world? Well, they're also going to have access to such agent technology. That means that now they will have the resources to build a hundred person team to compete with the companies they previously worked for. Well, how are they going to afford a hundred people team? Oh, well, that's simple AI agents. So the world we're going to is one with everyone who has a good idea will now have a chance to implement that. Compared to trying to build a website in 1996, compared to trying to build a website in 2004 with WordPress much easier, much, much, much easier. The barrier of entry has just been reduced dramatically. Same thing will happen to economic access and opportunity for people around the world. Our lives are going to be very different. 

I think the vast majority of people will gravitate to working less days per week, have more time for their family and children, which will be a net positive for the world. If you believe in that sort of thing. Parenting what a novel idea. I also think that we can spend more time on humanities and valuing things that truly matter. 

Work is not who we are. It is something we do, and most of the time we work because of necessity. Most people. If they have a choice between spending more time with loved ones versus filling out spreadsheets or making PowerPoint presentations, I know what they would pick. And then there's a group of people who work because they are truly excited about what they do and, believe it or not, if they have their financial or economic needs met, they would work for free right, because they believe in it because they're passionate about it and, honestly, that includes a lot of startup founders who can't afford to take a check yet right, because they believe in their vision so much. Well, in the future, we're all going to have our basic needs met thanks to this technology and we're still going to want to do exciting things just because we're passionate about it. 

0:30:50 - Mehmet
I'm very glad finally I found someone who shares my same thoughts, I would say, Alexander, because I remember also one time, especially when the chat GPT was still new and revealed, a lot of people were saying, hey, stop, don't do. And I said, hey, which one is better? Giving someone I call them silly tasks exactly what you mentioned filling Excel sheets, just copy, pasting data from the internet, and even actually, you don't even sometimes need actual AI, you can just need kind of automation to do this. And, yeah, I would rather prefer to give people time to go, as you said, to spend time with the loved ones and do. 

Actually, my dream is like we reach a place where we go and spend time on problems that need exactly what you mentioned this collective intelligence that we all have, and then we have the tools also to help us. Now, one thing, because you touched a little bit on it for now we're gonna have a lot of people doing these freelancers, for example, that you just mentioned. So for the ones who are non-technical, coming from non-technical background, what are the requirements for them to get involved in this? So what's the simplest way to integrate AI with whatever business model they would develop? 

0:32:21 - Alexander
Yeah, so everyone is talking about, or was talking about, this hottest new job called prompt engineer. Right, you need to evolve that into agent engineering, and the beautiful thing is that it allows you to program with language you can program with natural language and so the question then is what somebody needs to teach the agents your special sauce, your way of doing things. Okay, let's just imagine that I start a waffle restaurant here. I have a waffle restaurant. Now somebody else opens a waffle restaurant across the street from me. Okay, now we're both selling waffles. Why would people come to me? Well, I would have to make sure my waffles are better tasting, I assume. 

And I say waffles because I'm from Belgium, but that's okay. So I have to make sure my waffles are better. Okay, how are my waffles better? Well, maybe it's my ingredients, my data, maybe it's the way I prepare them, my workflow, maybe it's the technique that my workers are using to prepare them, the technique, my AI. Maybe it is the different machinery and tools I have that help me make the waffle. Maybe it's a waffle maker deluxe X2000 model, and so the way I make my waffles is by orchestrating all that better than the competition across the street that right there. 

To know how to do that, that requires expertise in something. So if you're in college and you're worried about AI replacing you, don't be. Even AGI needs to go to school, so to speak, by interning and learning the workflow and the special way you make your waffles better than others. So what skills are necessary? Well, I've got some experience with this because I'm training an entire generation of agent engineers or multi-agent system engineers MACE already, and I have some conclusions, even though it's early days. The people that are most successful at this know something really really well. So, for example, don't ask an accounting person to make and maintain an agent for marketing. They don't know what they are automating. So there's a room for everyone. Second, people that have more experience tinkering and playing around already today with GPTs and prompts and know how to make them do what you want. Those are the best people because they know how to program with language. If by now, you are dealing with employees who have not really used AI in their workflow, have not used GPT, by now it is commonplace, everyone can have access to it, even free versions. If they by now are not yet using it, it shows that they do not really have the disposition towards being successful as an agent engineer, because it requires that curious mind. Then, next up, I have put against each other software engineers, and let's just call them AI hobbyists prompt engineers, passionate about AI tinkering, doing stuff together. This latter group is more successful than the software developers at making agents. It may be shocking to some, but more important than knowing Java or Python. But those things don't. They are not important anymore. Python, javascript, other coding frameworks are just language. They are just language. We have Google translate in AI to translate our language to code. We can evaluate it with human language. 

If your entire specialization depends on a programming language, it does not give you an advantage. With agents. You need to get out of your comfort zone, learn a new paradigm. There's a new level of abstraction. If it does still does not sink in with you, then think about it this way If a client came to you and you insisted on using assembly language, talking directly to the hardware of the computer, to solve every problem, you would not have a job unless your job is to code and assembly, all right. Nobody would ever pay you for that amount of work. They would think are you crazy trying to do that? So it's the same thing with agents. It's an abstraction layer above programming languages. So there is no point for you to go and reinvent the wheel and do everything in JavaScript or Python and so forth unless you need it. But for building agents, mostly, you don't need it as much because it's more about orchestrating solutions. 

95% of SaaS businesses. I could take a challenge. Honestly, I could replace 95% of SaaS business functionality with an agent and with less than one week of effort in SmithOS, just without writing one line of code. By just using different AIs, apis, data sources and my workflow, I could replicate 95% of SaaS businesses today in one week time. This week I created an account-based marketing, personalized content creator that makes landing pages, email outreach, linkedin outreach and a video of my salesperson basically 40 seconds of highly personalized video sales pitch. All of that it took me one day to put together and it replaces about four or five days of manual work that my salespeople no longer have to do and they can now go to personal outreach Took me one day. So the skills that you need is curious, mind, prompt engineering, learning how to use an API. I think that's a valuable skill that everyone has it. If you have software background, that is useful. But yeah, that's the skill. And then some specialization. So go to school. Be good at something like making better waffles than the person across the street. 

0:39:57 - Mehmet
Wow, really, I love this analogy with the waffle business. It was a nice one, fine. So again, I'm asking these questions because people would have them in their mind. People talk sometimes okay, we got it, we need to be having all this, but we are a startup, we have limited resources. How we can get this done? So for startups in general, they are all the time with limited resources. So what are the cost effective way to put their hands on all the things that you mentioned and be inside the world of AI? 

0:40:47 - Alexander
Yeah, okay. So if you give it a little bit of time, all of this will come to you and will fall in your lap. So, okay, I predict the end of Upwork and Fiverrcom. It will be like Blockbuster in the United States. Yes, and the reason is agent marketplaces. If you need something, ask an agent. Predictable outcomes, 24-7 available, lower cost. So you only hire a human who is exceptionally specialized and talented if you need something done. 

Most of complex work, though. When I was a consultant in the years, or as fractional CTO building things for people, you know you take a complex client problem, you break it up in small pieces. I could do almost all client jobs by stringing together 10, 20, 30, 40, fiverr gigs and I orchestrate them. This is no different, but I will paint you a picture of what these marketplaces will look like. Okay, so imagine you have a company and you use Microsoft Teams or you use chat, gpt, and very soon the other ones are all going to imitate this and copy this over and it will be everywhere next year. So on the left side you have this. Well, at least in the west, where we read in this direction, but, like Slack and stuff, the list of people and Discord is on the left for us and where we write is on the right, yes, and on the left there's people and channels. Yes, okay, when we want to ask something to the CFO accounting related we don't go and talk to the person there that is a developer. We go ask the person who specialized in finance, right? So same way goes when you need help with anything in the future. 

Imagine your work environment, your Slack, your teams, whatever it is. Your agents are going to be available there in the sidebar, and if you don't have the talent to build your own, you just go to the marketplace and say, click, click, click, click, click. Okay, I have my team now. And then you're going to talk to your, your orchestrator, your manager. You're going to say now, work with that team to make me a new website. That's how it becomes accessible to everyone. Right, this is where we're going. 

1000% replay this a few years from now. And, in fact, I have been saying the same thing all year. You can go look at all my past recordings. In the meantime, there is a period of transition, of course, and so there are these companies that are very cutting edge. The world is still working under the old capitalist market conditions. Okay, if you come with you know 2040 technology in 2024, you're going to crush everyone, just like if you take I don't know these automatic machine guns and go back to Stone Age. Nope, you're the king of the world, nobody can touch you, right right, yeah so this is. 

This is why myself and innovative companies, why we are taking advantage of this technology today. Because you know we are. We are like gods in the marketplace, gods with a small G, right. So it's. It's it gives you superpowers over over the marketplace. It's an unfair. It's an unfair proposition. 

If you are just a small freelancer right today, you can do so much more just using AI tools, even without agents. If you're not yet doing that, don't even ask me about agents, because you're not even using what's in front of you. Right? I can write 30 page analysis white papers in in in one afternoon now. But learn to use this tools really good, and you're going to, you're going to be superpowered already today. And if you already know how to do that and you want to take it to even, you know, even higher levels of power maybe Dragon Ball Z, next science transformation level or something then agents are the way you do that. And so the way I'm transforming the companies I'm part of is we are going department by department and saying what is the biggest process bottleneck, that takes the most time, and we're analyzing it, we're sitting it down, we're diagramming it out and we're turning those processes into agents and then we're rolling down. 

No, this is a not consolidation of intelligence, this is aggregation. 

So what we're doing is we're making these agents then available to the right departments to help them accelerate their job 10 times, 10 times faster. 

And so if a service business is like 20% profit margin with an office building and only local people, 30% profit margin with people from all over the country, 50% profit margin with near shoring, 60% profit margin with off-shoring, then with agents you go to 90% profit margin. 

You can imagine a company with 10, 20 people doing the equivalent of work, what it now takes a team of a thousand people to do. And so what? What you can imagine, you know, a company that has a thousand people. You could never, ever hope to beat them with a small team of 10, unless you now have the ability to do the work of a thousand people with 10 people, where your competitor has to pay a thousand people to do the work of a thousand people, and so and so these big companies are bureaucratic, political, slow to maneuver, but your speedboat with your small team. So every Fortune 500 or Fortune 1000 company right now can be replaced by your startup if you use this technology properly and you can do it with four or five percent of the of the workforce. Just get you and your friends together and go pursue your idea. 

0:47:56 - Mehmet
I have to strongly agree with you, alexander, and just for fun now, I removed it when. To your point about how, if you leverage the technology, so when I started my own consultancy beginning of this year, so you know, when you put your team so I'm putting my, my, my, my own picture and I put Chad GPT and I called it chief content officer, of course, just for fun, you know. Because people ask me like how come you're doing a podcast, you're doing multiple things at the same time, like how many people you have on your team, and I'm saying yeah, it's me and Chad GPT, mainly because I leverage it a lot. Right, and now to your point. I think, the marketplace idea. I think this is what OpenAI they're bringing. I'm not sure if they're still on track in the app store moment that they are calling it, but it doesn't matter. 

0:48:50 - Alexander
It doesn't matter even if they never released a store you can now create agents and deploy them to your team by sharing the private link. So right, if you have a team today and they log into Chad GPT and they don't already have five, ten, twenty agents to do everything in your company faster, you're already losing to companies like mine who do that. All right. So don't wait. If I mean if you're, if you're not even using the tools you have, then yes, wait. But but if you are on the cutting edge and you're keen to beat your competition, you need to build agents. So the agent builders to you know what is a Chad GPT? I mean, it's a. It's a document with a chat. Boom here, go and talk to a document. That's what it is. 

But when you have actions you can add, you can add APIs there. But did you know those API's could be agents in and of themselves, with complex workflows and intimate knowledge of your process, data etc. That's completely separate and private to you as a company. And so what we do is we have this agent builder. You can build the most amazing workflows, even things humans couldn't do, because you need to compare a thousand data sources at once. A human couldn't do it and we just click import in Chad GPT. One click and boom. The agent I created here is now available here. Same thing with Microsoft co-pilot for enterprise yeah. 

0:50:27 - Mehmet
And when I told someone two weeks back in, when they just announced it, I said he said, yeah, I get it. But you know, like, what's cool about this action is that you know you can do everything. And I said yeah, because, for example, imagine now you're by yourself, you are a solopreneur, let's say. So you can have an agent called accountant. He's not yet a CFO, maybe with Pam he becomes a CFO. And then you have your operation guy, and then you have the, your sales marketing guy, and then you have, you know, so you have your team of agents, and then you know they talk to each other's and then they said, really, I can do this. 

0:51:02 - Alexander
I said yeah, with API's, you know, and Smith helps make that ridiculously easy, so that that account based marketing person I talked about they create all the specialized content, right, okay, so my next task is I'm building an agent that finds the right prospects to create specialized content for well as a separate agent and then after that I'm building another agent that then evaluates critically the content and the people once more and just maybe final refinement, like an editor or proofreader or whatever you would have in a company. 

And then the final agent schedules all that content and HubSpot and so forth and connects to LinkedIn and posts it. And then from there on I just say I want you to do out the reach to 100 people. First agent goes to work and find people for me. Second agent creates content for them. Third agent evaluates the content. Make sure I don't say anything silly, just make sure everything is correct. And final agent goes and schedules everything and then I all I watch is sit back and I watch my calendar fill up with appointments wow. 

0:52:15 - Mehmet
And then I need an agent to get me out of that job too now just out of fun, because sometimes people say, okay, but what if the other person on the other side is an AI? 

0:52:28 - Alexander
right, so yeah, no, but this is, this is a silly argument because it's only a matter of time, because before agents start purchasing from each other and in fact, it already happens, if, if let's say that you are 11 labs and you can make text to speech in a convincing way, right, when an agent calls on 11 labs, they are paying 11 labs money. Think about it. You have only authorized that with an API key, you, true, but the agent is purchasing from another provider already. So, and the executable web? That's a silly question because the executable web more economy transactions happen between agents than between humans. 

0:53:18 - Mehmet
Wow, yeah, you know, when I wrote this people they laughed at me. I wrote in 2017, I said in the future, we will see AI, selling to AI and transacting with AI. And people say you're too much futuristic. I said no, like I'm not futuristic, we're going to see it very soon. But, you know, some people talk to Alexander also about the ethical aspects of things. So, and they say you know what, like, we need to be very careful. There is, like you know, it can be biased, it can have give sometimes wrong information. So, from from, from medical considerations perspective, right and especially for, you know, the sector of people we talked about. So what are some of the pitfalls? They need to watch out when, when, when they start to implement AI. 

0:54:14 - Alexander
Okay. So if you yourself are implementing a solution or workflow, you are ultimately responsible for the output. Okay, back to the waffle place. If your waffles, if your waffles, make people sick, you're responsible, not the machine, not the, not the ingredients. You are responsible, right? So that means that you need to make sure your process is auditable and that it follows your standards, and so, again, that's really hard to do without a solution like SmithOS. And SmithOS we are fully debugging debugging of the of the workflow, so in each step, you can see exactly what each AI is doing and what the inputs and outputs are, and you can write your own test cases and test all the scenarios and make sure the outcome is always a delicious waffle. 

0:55:12 - Mehmet
That's. That's nice one, alexander. Also and you know you made me now feeling I'm scarving, scaring for some waffles. So, as we are almost coming to an end, alexander, like where anything first that you, you, you want also to share anything that we didn't discuss. 

0:55:36 - Alexander
Well, you can always invite me for a follow up call if you have more questions, because an hour goes quickly if you're having fun. 

0:55:45 - Mehmet
Yes, and where we can find more about SmithOS and about you also. 

0:55:52 - Alexander
Yeah, go on LinkedIn, alexander Derritter, my hand list, adr IDDER ADRitter, on both ex formerly known as Twitter and LinkedIn. That's where I spend most of my time online, and then my website, SmythOS, has some information about what we're building. This technology is very powerful, so we're, at this time, quite I would say quite select and who we give access to. But as part of our ethical approach, we are having scholarships for smaller companies and teams who we think are doing worthwhile things with it. So reach out to us anyway. If you think that this is something that you want to pursue and you have opportunity to go make something exciting, we want to want to hear from you. Other than that, I just say treat AI like like, like a bag of toys, with the excitement of a child and curiosity of a young child. It is so much fun, so much fun and the things you can do with it. It's just unbelievable I had when I was a little boy in Belgium. 

I had a friend of the family who came to visit from America. Ever since I was little my youngest memory is a child and she became so friendly to the family we call her aunt, and so whenever she would come, she would bring gifts and I had. You know, I say you're like the fairy godmother and like Cinderella or something you know. So recently I was able to, with AI, make an original song for her, like saying thank you. I was able to, with AI, make animations and video with that and it only took me a few hours, I mean. I mean I would have had to go, I would have had to spend weeks to try and do that. Maybe learn how to write music, I mean, have fun with this. My, my children got a visit from the Tooth Fairy with a letter recently. My, you know my things you could not imagine that you could do Write a book, write a book, make a cartoon, make an entire cartoon series with your own characters and humor and voiceover. 

You can now do it. All you have to do is orchestrate. Write a book, make a, make a movie, make a cartoon. Go and analyze your own transcripts. Go, please. If you do not do this like record yourself every meeting, please Record yourself, copy paste the transcript in AI and then talk to the transcript. How did I do? It's the best business coach you've ever had. I do it every day. What did this person really want? I noticed like my notebook here captures about five to 10% of what they really said. 

0:59:09 - Mehmet
Wow. 

0:59:11 - Alexander
Yeah, you need to do that, so it's so much fun. Have fun with it. 

0:59:16 - Mehmet
Absolutely, absolutely, alexander, and I have to agree with you on this. And for me, you know, it added a lot of value for me because you know I'm you know, not as, as you can imagine like perfect in everything, but I'm a perfectionist. At the same time, I like to do my job as perfect as possible. I know I cannot be 100% all the time, but, you know, leveraging the tools that are available and I think AI is one of the greatest tools that we have in our hands today, so we need to to leverage it to the maximum. I 100% agree with you and this is why, you know, I get excited every time I have someone sharing the same, I would say, passion, the same ambitions of leveraging technology. And I think you know the audience. They need to watch for you know what SimuTiOS will be bringing to us, because it's a really cool technology and this is why I was hooked, even you know, once. You know, I saw the. You know you have on your landing page something over there, so it's something very, very, very promising and I think this is where the future also is going. And, as usual, this is, you know, I will make sure also the links you mentioned. They will be in the show notes and this is how I end my episodes. So for the people who are first timers here, you're passed by. Thank you for passing by today. 

If you liked this episode, please subscribe. If you're listening to us on your favorite podcasting app, please don't forget to subscribe. If you're watching this on YouTube, also appreciate if you can subscribe and give up the thumb up, and if you are one of our loyal audience. Thank you very much. I keep reading every single message you send. I keep reading all the notes you sent me also, and thank you very much for your encouragement. I love to see some feedback also as well, so don't hesitate. If you don't like something or you want me to bring a topic to the podcast, I would be more happy than do this, and also if you are interested yourself being with me to record an episode wherever you are in the world. Time zone is not an issue. I am kind of in a mid-mid area in the world. My time zone is very flexible, so don't hesitate. Also reach out to me and thank you for tuning in. We'll meet again very soon. Thank you, bye-bye.