Hello, and welcome to another live episode of the leveraging AI podcast, the podcast that shares practical ethical ways to leverage AI to improve efficiency, grow your business and advance your career. This is Isar. Mate is your host and we have an incredible episode for you today. Now everybody, literally everybody is talking about building agents and how they're going to change the world and so on. But the reality is very few people actually build agents and even fewer people build agents that actually do something meaningful in their businesses day in and day out. Our guest today, team Keck here has built over a hundred agents that work in his company regularly and do different functions in his business. And through this process, he became so familiar with what's good and not good, that even fired more than 50 of the agents that he initially created. So he's been evolving this for a while. Now to make this episode even more interesting, we are going to talk not only about how to build the agents and show exactly how to do this, but we're also going to look at it with a very interesting platform, which is cloud code. Now, from an adoption perspective, everybody knows chat. You can see some people know Gemini because they get it through their Google membership, but a lot less people are actually using cloud, which I personally really, really like. And out of that small people that uses cloud, a much smaller percentage uses cloud code and the vast majority of people who use cloud code are coders. But the reality is they are huge benefits that you can get by using cloud code, not coding. And this is not exactly what we're going to look at today is how to use cloud code, not coding because I don't know how to code. And I'm still gaining a lot of benefits from it. Now, the reason most people don't do it yet is because either they don't know that it exists or it has code in the name and people assume that's the only thing it is good for, which is not the case. And this is one of the reasons I'm really excited about this episode because you're going to learn a new tool that most of you probably do not know, and you're going to learn how powerful it is and what can you do with it without writing code. Now, the other reason a lot of people are not using cloud code is because until not too long ago, you had to go through a weird install, which we're going to show you how to do in order to get it to actually up and running. But the reality is as of about a month ago and Tim correct me from wrong afterwards, you can now run cloud code on the web, which means you don't have to install anything. There's still benefits of installing it, but you don't even have to do that anymore. So in today's episode, we're going to explain to you the benefits of running cloud code and how to use it to create incredible agents that actually do functions in your business and provide value. Now, our amazing guest, Tim, brings an really huge variety of experiences that makes it highly relevant for our episode today. He held top leadership positions, including CEO and COO in many different companies, which means he understands how a business actually runs and what he takes to run an effective business, but he was also a professor into separate university, which means he's really good and passionate about teaching. And he is currently the chief AI officer and the founder of AI operator, which is a company that helps businesses implement AI in an effective way. So this combination makes him literally the perfect person to teach us about cloud code and how to create agents not being a technical person. So you don't need any coding or technical skills to do this. And that makes me very, very excited. So Tim, welcome to leveraging AI. In the next three years, AI technology will change our world dramatically. Whether you are a business executive, trying to catapult your business forward, or just somebody who refuses to be left behind and want to advance your career, this is the show for you. I'm your host, Isar Matys, a serial entrepreneur and an AI enthusiast. You'll hear invaluable practical tips from innovative business leaders, AI practitioners and some of the brightest AI minds in our world today. On how you can leverage AI in ethical ways to advance your career and grow your business. Well, thank you so much for the kind introduction. It's an absolute pleasure to be here. It's our thanks. You know, you and I spoke several different times. I'm a big fan of your YouTube channel. We're probably going to talk about that in the end where people can find you. But you're, you're really a magician when it comes to doing these kind of things. But let's really start at the more higher level. What the hell is cloud code? And why do people need to care about it? Yeah, I think that there is one issue about cloud code is that they called it cloud code. Yeah. It's better than GPT, at least it has a name that means something. Yeah, yeah, it's just a code. I mean, cloud is great. But is they call it cloud code and then everybody like, I tell them, I'm like, how have you tried cloud code and they're like, I don't code. I'm like, well, you're missing out. Well, maybe it's good for me because I use it and I'm not others do. But no, I like to share. I want everybody to use it. I want everybody to get the productivity gains and the efficiency and the incredible. I guess you call it magic. I call it just obsession about obsession about AI and what it can do. And so cloud code was basically a way to interact with the cloud models in the CLI, in the command line interface, which is the terminal in a computer. And that's very scary, right? And because there is no UI, there's no user interface or the user interface just takes. It's a dark screen. It looks scary, right? It looks absolutely terrifying. I even sometimes, oh, shit, am I working here again? Like, I just worked 90% of my time now in the terminal, you know, and it's like, it's so funny, like we've done such UI designs and so on, right? And then now we're going back to using the terminal. So cloud code is really an incredible, let's call it product to, to tap into a large language models that are bi-antropic. So cloud opus 4.5 being the latest one. And what's, why is it so powerful? Because it runs on your terminal, it can create files, folders. It can run scripts and it can really control everything on your computer, right? Yeah, from including browser to the applications to the settings of your computer itself. And because of that, it takes the, you know, a level higher in, in, in control and efficiency and being able to achieve things, then a chat GPT, right? The chat GPT is great. You have to connect things. You know, you click on connectors, close itself, you click on connectors, but with cloud code, it's in your computer. It's, it's literally everything that you have in your computer that you can run. And in a nutshell, it looks scary, it sounds scary because of the code, if you don't code. But what I realized with cloud code is that it's right now the most powerful product for building systems, business, operational running systems. And, and, and I'm still mind blown. You know, when I speak about it, I'm like, what the f is it? You know, it's just, it's still incredible what we're living. And it's just, you know, the beginning of December and, and cloud code is, it's just mind blowing absolutely amazing. So, so let's really show people, first of all, how it looks like. And then we can start diving into examples of how you are using it to create these systems and processes. And again, people, yes, I agree with everything team said, like, what is the terminal? How do we even get to it? And it's really simple. You have both on Mac and in PC. It's there's a button that takes you there and, and you can, and you can have the terminal. And as it, as Tim said, it looks like we're back in 1985. It's like a boss prompt, like environment, but it works. And it is extremely powerful. And for those of you who are like, oh, my God, I don't want it to mess with my computer. You do as long as it's controlled. And one of the amazing things in cloud models in general and specifically, cloud 4.5 is that it's code adherence, which is a professional term to saying it does what you fucking tell it to do. Versus kind of does what you want it to do, it, it, it is very good at that. So it's only going to do what you tell it to do, and it is not going to do anything you do not tell it to do. So the fear of like, oh, my God, this thing can take over my computer in theory. If you tell it to you and you define exactly what you want, it will. But if you tell it to do one, two, and three, create this file, copy the information from here, bring it there, activate this thing and send this message, then that's the only thing it is going to do. Yeah, and I think that it has really good security settings. You can set to the settings so that you can decide on what it can do without asking you each time. And it's also incredible at really asking you when it's necessary and when I don't have enough information. It's like, hey, I'm about to do this, but I'm not sure. Like, you know, what do you think about this plan? And you're like, oh, okay, I like the plan. Going to the plan or let's change the plan, right? The planning mode of it is just, it's just mind blowing in general. I find Claude to be the most collaborative model out of all of them that really makes you feel that it's a consultant and not just an execution arm of your dreams. And that makes Claude like an amazing partner for anything this as well. Yeah, I think that in, you know, if we talk about models very quickly before I show things, you know, I use Claude 4.5 opus, the most at the moment. But if I'm doing a business strategic task like planning, you know, Q1, 2026 from my business, I will still switch to GPT 5.1 pro, right? Which it's only on the pro people, so not the plus, the 20 bucks, but the 200 bucks and people tell me, oh, I have the problem. Like, no, no, no, no, you don't have the, how much you paid 20? Yeah, when you don't have the pro. So GPT 5.1 pro is just incredibly amazing at business, a business logic, business strategy, and so on. But everything else now, I'm, I'm definitely a Claude 4.5 opus fan and it's, you know, if you pay for Claude Max at 20X, which is $200 a month, you don't hit the limits. And I run that on my Claude code. So just if I, for everybody, be careful, you know, I'm going to show you things now. If you write it on the API, the cost can go up very, very quickly. You do want to be on a subscription package, but on a max, Claude Max plan, a 10X or a 20X plan. Okay. So let's dive right in. Let's see how this thing looks like, how to set it up and how to create magic with it. Awesome. Let me share my screen. Can we see my screen, yeah? Yes. Awesome. So this is what I said is scary, isn't it? It's just, it's just a black box, right? So let's take, let's take a step back. Yeah. How do you get this thing? How do you open general and computer? Yeah. So for people who don't know what it is and absolutely. So I run a version of the terminal called I term 2. It's an application that you can download. You can run the normal terminal, the normal terminal doesn't have some of the little gadgets. I mean, you know, now you see these as cool gadgets, because you can see, you know, the GPU and what it's doing, but on the normal terminal, you don't see much. So this is a little bit cooler than just the terminal itself. There's also another option that I was into for a bit called ghosty with double tea, ghosty, which was also a terminal tool, which is really good. But, you know, you just go on your spotlight and your search spotlight. If you're on a Mac or in a Windows, you know, to your search and you go right terminal, right? And if you want a little bit of better experience, I would suggest I term 2 is a great little app. Cool. Yeah, I will say one more thing, not to confuse people, but I would say it anyway. If you're using any of the vibe coding tools, whether it is replete or lovable or whatever, there's a terminal version in there that you can run as well. And it has slightly better user interface than this thing, but overall it does the same stuff. So if you're starting with nothing, then just running the terminal in the computer could be your easiest starting point. Yeah. I mean, I think that's the difference there is I think we spoke about this when we first met, but you know, I was saying that yes, in the replete terminal, you can run clothe code. Yeah, I think you tried it since the six minutes after we close the call, I would be running. It's sort of the best hacks people don't realize is that you can open up replete. And not pay too much for replete tokens and only set up your environment and then run the terminal, clothe code in the terminal inside replete. And then you can see the preview, which is great, but you know, they're really talking about coding and building a product that folder and you want to be in that repository. But here what I really love is that I'm on my computer. So the first thing that I did when clothe code came, and I was like, oh, what's this crazy stuff? I was like, OK, and it was a couple of sleepless nights as usual and anything new come. It's a sleepless nights. You know, I love it. And the first thing was like, oh, OK, you can do files, folders in my computer. And it was years that I've been wanting to tidy up my desktop documents and all that kind of stuff. And I just said, hey, can you strategy and I was like, great. Can you implement that and it changed names of files and images and videos and put them in folders? And it just kind of stopped cleaning out my computer. I was like, oh crap, this is very good. You know, I've been wanting to do that for 10 years and so and I never did it. And you know, I just piled shit on top of shit. And then it was it was putting all the PDF somewhere naming them, reading them, telling me what they are. And really creating different folders for me. I was like, wow, this is really like we're starting to see real agentic agentic support assistant, whatever you want to call it on my computer. So here's here is the terminal, right? I turned to I turned to running my terminal inside my terminal. And it's a it's a black box. It's a black screen. Let me see if I can make this a tiny bit bigger for people. Maybe so it's just a black box and then you have to install code code on your computer. That's very easy. You can actually go on on on Chrome or in a browser and write code installation. It's a it's a very small snippet of some words and some some kind of code that you just input here and you let it run and it installs it. Ask you to log in, right? So I'm not going to do that because I've already done that. It's very easy. So this is five minute process. And then what I'm going to do now is I'm just going to tease you a little bit. And then I'm going to show you what I've built and then we're going to into how to build what I've built and this type of things. If that's okay with you, these are awesome. Sounds like a great plan. So what I'm going to do is I'm just going to launch clothes by saying clothes, right? So I'm just going to say clothes and I'm going to enter. And as we can see right away, I get the question asked if I trust this folder, right? Because clothes code may read right or execute files contained in this directory. This can post security risks are only used files hooks and bash comments from trusted sources, meaning problem to injection. As we as we know, if this was some files and folders that I didn't know. And then there there was some some kind of hacker trying to get my information. It could kind of do that. But it's my computer. I know it's clean, you know, I've been working on my computer for for really long. Even if I change computers, I cloud, you know, so I know what's in my computer. So I'm going to accept yes, proceed. Right? And then that's it. We get this little, this, this little, you know, I love it. This little thing, you know, and here we go. This is clothes, clothes code is ready. And I have clothes on opposite 4.5 with the clothes max plan, right? As I was mentioning, you don't want to be on the API. You're going to pay a shit, put on money if you do that. And then I'm on my user. I mean, I mean, my whole computer. I'm not even in a folder, but what you could also do, which I'll show later, you can run it in a specific folder. Right? So I do run it in very specific folders. Like I have a sales folder. I have an onboarding folder. I have a customs folder and I run it in different places. And what's great about this is that I can open up multiple tabs and I can run 4 or 5 clothes at the same time. And it's like, oh, so I had an assistant and now I have multiple assistants and each can run multiple agents. And each can have different skills, close skills. It just gets a bit meta. Like I run orchestrators that run other ones and they run each other. And it's just because that what? And then I was reading about hyper productivity and it's scary because it's quite exhaustive. It's like, oh, my God, like I have to, I have the baby said all these agents. And also, so I think it's tough. It's not easy to get to the usage that I am, but I'll show some simple tricks for everybody to start getting there. Because this is the world that we're really living in. But in 2026, I do believe that this is going to be one of the most powerful things is to build systems that can run without you, right? And you wake up in the morning. This is my dream. I wake up in the morning, which I'm kind of doing a little bit of that. But I want to wake up at 9 a.m. I'm not, you know, I want to be on my desk at 9 a.m. I've dropped my kids to school. I'm at 9 a.m. in my desk. I open it. And I have done all of this stuff. Bless you. I have all of this stuff. I've done all of this stuff. These are things I was a bit unsure. You know, I want to ask you some questions. Great. What is what did it do? I did all the follow ups. It sent the proposals. It's, you know, it's, it's created some contracts for some clients and so on. And then it has some clarification questions like, hey, I wasn't sure about this. You know, like, can you check that email? Can you check that document? And you check it. And then you say, okay, you know, glad you didn't send this because here's some feedback, right? And then you give that feedback. But then that feedback, the next day, I don't want it to ask similar feedback because it knows it. So I wanted to learn and I wanted to be iterative learning, right? So that it keeps, it keeps learning every time I give it feedback. So I kind of build something like that. I'll show you in a second as well. But this is the world we're going to live in, guys. And you know, it's just going to wake up throughout your whole sleep. And you went to school, drop the kids and stuff. You came back, you've had multiple agents, you know, 50 hundred agents actually doing work for you. Right? This is definitely happening. I can show you in a second. You know, it's just that I still dream of a little bit more, more proactivity from these agents. Right. I still have to run them. I can schedule them. But the more proactivity is coming. I know that because we saw that with chat. LGBT pulse, right? On the pro plan, we have something called chat, which every morning, it gives me like some incredible things about me, about my search history, about my day, about my calendar, and it kind of is proactive, right? So I think we're going to get this proactive right now. It's very reactive. So I have clothes clothes ready. It's just it's scary because you don't know what to do. It's asking like it's waiting for me. It's like, okay, well, I'm waiting for you. What are we going to do? Right? So let's say what we're going to do is we're going to go and we're going to launch. Let's launch a guide about Kloat code. How about that? Right? So I'm going to write a 10 page guides about Kloat code, right? And I want to write this. So what I'm going to do is actually, I would kind of go to the folder where I have content and guides, but I'm not going to do that. And what I'm going to do, hopefully this is going to work. I'm going to talk to it because I talk a lot to these things now and I don't type. So I'm just going to talk to it quickly. I go to my guides of folder inside my content folder in AR Prader and run my guide system and let's write a guide about Kloat code. How how Kloat code with Office 4.5 and the Cloud Max plan and how Kloat code people think it's for coding, but it's not you can run business operations and you can do a lot of things and so on. You know, do a proper research and make sure to run all the agents and and to run the actual guide system workflow. So you still kind of have to babysit a little bit and you have to tell it like make sure to run my workflow, right? Make sure to find the agents, you know, because if no, it's still wants to please you and it's going to go and do it on his own thing, but I already have a system for this. So what I want to do now, I'm doing shift tab as you can see something is changing on the bottom left and accept it its own so that it's kind of it accepts like you automatically accept some stuff. I'm going to take it to plan modes, right? Because I don't want it to go and actually stop doing things. I wanted to plan first. It gets very powerful when we put things in plan modes, right? So this is going to take quite a while. And again, to switch to plan mode, you just used shift tile. Yeah, so just use a keyboard shortcut. There's no again, like anything fancy to do that. You just switch the mode that it's going to operate in. Yeah, shift tab shift tab and then you get to plan mode. You see it's green on the bottom left. It has like this. This kind of pose button. Is it a pose button? So it's more like, hey, I'm not going to take actions. You know, I'm just going to plan. You know, I'm not going to go change anything. And what we can see, you know, we can open up this whole thinking with control. Oh, which gets really big. And you know, and it looks scary, right? This is where exactly it starts looking scary because like, okay, what is this? Actually, it's not that scary because this is just like, oh, it's finding folder, right? It's going to my documents. In my documents, it found my AI operator folder, right, my company name folder. And then there it found the content folder and it opened the app. And then it's looking at certain things, right? And so on, right? And as you can see, I just asked something, but it wrote it's on, it wrote it's on front. Find the AI operator project on the system. Look for a folder protocol, AI operator or similar within it, find the content folder and a guide sub folder. Look for any guide system and so on, right? So I kind of gave itself some task. And it's like, okay, this is what I'm going to do. And then it keeps thinking, I'm going to close this because it can look a bit scary. And let's go to the normal look, right? So that's like the detailed look of seeing what's inside history on the 60 lines, these three on the 76 line. What is it doing inside these? You just go control. Oh, it gets even more scary than what it is. So you close it. Yeah. Oh, my God. Let me not see that. But it's good. I do recommend, you know, when you get comfortable, I do recommend to read these things. And you know, if you want to understand what the model is doing behind the scenes, and by the way, it's the same thing in like all the regular tools. Like if you're in chat GPT or Claude or Gemini, it says it says thinking, I'm like, okay, what are you thinking about? If you click on the thinking blinking thing, it's going to show you exactly what it's doing. And then the benefit of that is you understand how the model works and how it thinks. And that usually makes you a better promter because you know the steps that it's going to do. And you can guide it just like Tim did now and like, oh, go look for this folder, go look for these agents. Why is he doing that? Because he looked previously and said, okay, it's kind of like struggling with that. If I tell it what to do, it will know what to do. So it's looking at this actually gives you a better understanding of how the model's work. And then you can be a better promter because you can know how to guide it. Yeah. And I don't, it's not just about being a better prompter. It's about this thing that we see in the news. It's like saying that humanity's getting dumber, right? They're getting dumber and kids are getting dumber because they let chat GPT do all the work. And they just get the output and then they go with it. But the second you open the reasoning, the thinking, right? The thinking I always say it was not it's the reasoning and you open that up and you start reading it and in every model that you're using. If you can read that, you get smarter. Yeah. You get absolutely smarter. I see things that these things are thinking or like reasoning. I'm like, whoa, I didn't think about that. Or let me, let me bring more information. Let me bring more context, right? It becomes really a partner, you know, and not just like God did this work for me. And then I, yeah, it's okay. Let me go. No, it's like, well, we work through things together and I'm learning. And the next time I know how it's going to kind of thing. And I was going to reason and I start understanding different models. It's, yeah, it's very geeky. But yeah, I'm obsessed about all this stuff. So as you can see on the bottom, what happened. This is quite a new UI user interface kind of feature is, you know, it did one, two, three, four things. Like a project type audience use cases pricing for these like a tabs. And then I can submit. And it's great. It's asking me clarification questions now, right? So, so okay, do I want to stand alone, deliverable? Like top level folder, like AI safety starter pack, which I have already built up before. So it's like, do you want a complete package for distribution? Or do you want to research the guide pipeline, right? Do you want just a, just a quick guide about code? Or do you want something for distribution? I potentially want to stand alone, deliverable, right? So I'm going to click that. So I entered that. So, so basically tick that so you can also tick things. If you want to see him, you can do a space box, which I've already done. So no, but here, who is the primary target audience for this cloud code guide? Right, business operators. I think entrepreneurs founders, you know, I think I would go, I'm going to go for entrepreneurs and founders, right? Because those are the people that I kind of, I kind of talk to. So I'm going to go, yeah, entrepreneur founders. What specific business operations or use cases should be emphasized beyond the coding, right? Here, I would potentially spend a bit more time, right, and tell it to the things, but I'm just going to do something quickly. I'm just going to tell it so that it kind of finds my systems. Make sure to go read my systems. I have a this guide system, which is about content creation. I have a sales system, which is about, you know, running a CRM and so on and how to sell in my company. I have a onboarding system after I close the deal. How do I onboard the client? And then I have a training system. How do I train people and create the training program? The custom training program for our 12 week AI first mindset training program, review all of that. And then as a bonus, because I'm on a very important call right now. Also, can you highlight the system building system? No, thanks. All right. So I kind of said that because I wanted to go and check these things, right? So for those of you who are not watching the screen and just listening, the way this looks like, again, think about the poorest user interface you can imagine. So boss level. So there's just this level. There's different rows and you can pick from them by scrolling to them with a keyboard and selecting them. Or you can go to the bottom one, which is just like a open line and type whatever you want to type. What Tim is doing is what I'm doing as well. We're voice typing everything just because it's a lot faster. It requires a decent level of talking skills, but not much because these tools are pretty good at understanding what you want, even if you didn't structure it as well as Tim. You get much better at this. The fact that Tim has a YouTube channel, and I have a podcast makes it a lot easier to say what we want in the first time around in an organized way, but you will figure it out as well. And it's just significantly faster than typing. And to be fair, I find that I express myself a lot more with more details and color when I speak versus when I type. And so that's another benefit because then the AI gets more context. But now that you've done this, let's see what the system does. Yeah, I just want to add something to that is because I'm still kind of practicing the speaking to the machines instead of typing to the machines because guys like every, you know, guys and girls and everybody, you know, like get ready because we are going to speak to the machines more rewind by voice, you know, the new open AI product hardware is going to come is definitely not going to have a screen, right? So we understood that. So it's not going to have a keyboard. We're going to speak to these things and you're going to have a robot in your house. You're going to speak to it like these things are going to happen. So I think it's a great time now to practice speaking to these things. So I'm still not very good at explaining what I want and sometimes I go, but it's okay because it understands and it's somehow cleans it up, you know, and I think by practicing, I'm getting better at it. So I'm going to accept what I just said there. I'm going to say, okay, this is this is the good stuff. And then it said, pricing should the guys focus specifically on cloud max? No, all right. Altier is comparison, right? So we want Altier's comparison. I'm going to submit all this information. So I've submit all this information. It was a four clarification questions. We answered them. And now four or five, right? And it's not going to do the work, right? And you'll see the minutes is spending and the tokens that it's spending, but I'm not paying per token. I have a subscription plan. And you know, if you're listening and not watching, it kind of gives you its plan. So it's exploring the operator systems because I told it, go, look at my systems, you know, about the guy, the sales day on board and the training on my system building system, which is the brand new one. I just done last weekend. And I'm still working on it. And it was like, I was building so many systems. I decided to build a system that build systems. Like I was like, you know what? Kind of like the clawed scale that builds, clawed skills, right? It's the same. Exactly. Yeah. Exactly. Yeah. I took that approach. I was like, you know, why am I keep designing, you know, these systems? I should have a system that asks me question and goes and designs the system from the first four or five systems I've already done, right? And I just finished that. And you know, I'm still working on it, but I finished kind of the first draft. And I'm like, oh, this is getting better. This is getting better. I even tag lighting and tropic on one of my LinkedIn posts. They didn't come, but I was like, guys, you got to see this. I'm taking a bit next level now. Yeah. So what it does, it gives you the tasks and it has four tasks here. It's going to explore. Then it's going to design the guide structure. Then it's going to write final plan for quote, quote, quote guide. And then execute the guide workflow with all agents, right? So even before going to my guide creation, a system, it's going to do these three tasks, which is the explore, the design and the right, the final plan for the quote guide, which is great because this is not AI slop. This is exactly what I do. These are my systems and this is my context. And so the guide is going to be about exactly what I do. And people are like, oh, yeah, but you got AI to do your guide. I'm like, yeah, well, AI looks at my context, ask me questions and it's going to ask me more questions and I build these systems for the last, I don't know, a couple months or whatever, right? It's definitely very, very custom content to the context of, of Tim, the AI operator and my systems. Let me ask you a positive answer question. You're sending it to all these resources, like your systems and your agents. Can you talk a little bit more about what these are, like what are these systems and what are the agents that it's actually using? Because that's a big part of this, this secret sauce, right? It's like the claw itself is actually using these additional resources in order to do the thing that it's doing. Yeah, yeah. So I mean, it is a perfect timing. I just wanted to launch this because it's going to take a few minutes anyways. And I'll change my screen share. I think you can see my browser now, right? Yeah. So I had to close code open for you guys, you know, you just go to cloud.com product, cloud codes. You can find it. This is like the kind of the kind of code that you have to put it in the terminal to download it. And then you know, you can learn a little bit about it. It might be scary, but here's the plans that I was mentioning and so on, which is cool. And now before we get into the system now and close itself, you actually do have here code. So if you click on code now here, it opens another tab, which is code code, but it runs on the cloud, right? It can run on an environment that you give. Actually, it can start now running inside your also system and your GitHub repository, which is another tool where we put code in. So this is quite new. So we know that what's happening is that they've realized that the usage of code and now they're like, okay, well, let's offer it on a browser. Let's all start offering different things. So there is going to be a product that is going to be better in UI and it's not going to be the terminal. So you're not going to have to be scared. But I believe that the biggest power still is in that terminal. So, so you know, I'm still there. So because you asked that question and I had prepared this, of course, it's hard. So this is I've never doubted you Tim. So I prepared this quick thing with cloud, with cloud code. I said, hey, you know, can you build something that shows my systems? You know, and it's like, oh, yeah, of course, I can build your whole thing and then build me a whole beautiful thing that I'm going to talk about. So I have a systems architecture overview. All right. So all of this that what you're going to see runs in my computer in folders, right? They have agents, they have skills, they have MCPs, model context protocols, they have all these things, but it all runs in my computer, right? So this is my new meta system, which I'm very proud of myself, like I'm proud of myself finally, like I'm taking a whole December like almost on just focusing on this, right? So we've closed all our cohort Q1 clients already. I have a few more training to finish before Christmas, but I'm like focused. I told my team and I guys I'm focusing on systems, you know, because it's just like it's going to help me scale in 2026, you know? So I build a system builder. So this system builder is quite simple. It's the meta system to that build the systems. It has five core agents. It has an orchestrator questionnaire, a mature generator, a validator, which we'll look into it. There's five simple agents, you know, the orchestrator actually does the whole work. The questionnaire comes and asks questions to the human, which is me to make sure that it doesn't kind of create its own thing, but it's from my brains and stuff like that. And then it matches things like I created blueprints from the existing systems that are like templates, blueprints. And it matches the new system that I need to a blueprint. And it says, oh, you already have this type of blueprint so that I go even faster. And then it generates the whole system and then it validates it. So these are five core agents that work on these. And then there is like these little factories, right? So I have an agent factory. I have a skilled factory, a closed skill factory, a folder factory that has to create the foldering system. I have a workflow factory that creates the whole workflow that has a different agent so that it knows how to run. I have a script factory so that anything that needs to be code, right? So that it's deterministic and not non deterministic. Because as we know, large language models are non deterministic. And that's why you don't get 100% each same time, each time the same inputs, the same output. Sorry. And that's the problem that we're all facing with large language models. But you can fix that by running scripts and scripts that are already defined and deterministic. And then I have one that creates all the readme files, which is like files that explains the things, right? And then a one a factory for schemas, right? Building schemas of different things, MCPs, codes and stuff like that, right? So and then this is the blueprints that I was saying. They're called the archetypes here, you know, which is one as a pipeline, one as a journey, one as a production, one as a program. So the pipeline is my sales pipeline. So I made that a blueprint. The journey, the journey is the onboarding is the onboarding one. The production is the content one, the guide one, and the program is my trainings program. And for for archetypes, I think it's quite good because most things are going to be like kind of a pipeline or kind of a journey, kind of a production, kind of program. If something doesn't match, I might create a fifth one, you know, but these are my blueprints at the moment. And I have, you know, a three-phase process. The score is about 20, 30 minutes. And then so it's not this, you know, people think, oh, I'm going to build an agent, it's going to be, you know, and if you do that, you're going to get crap agents because you just build something that is very generic. But I spend, I spend hours on building these systems. And then people are like, yeah, but you spend so many hours, yeah, but I spend 15 hours building that training system, or maybe now about 30 hours that I've spent. And it's, you know, I've served about, you know, about, was it 250, 300,000 dollars of clients with that system, you know, and it's okay. I'll spend the 30. Yeah. Yeah. The ROI is very obvious. I have a couple of questions. Yeah. Just diving a layer deeper to get people to know what it is. So we understand it has agents and it has factories. Yeah. Can you walk us through what is an agent and how you built it and what is a factory and how you built it? I think prototypes are pretty obvious, and I think the phases of production are pretty obvious. I think the things that people are scratching their heads, I'm like, okay, about how did you create this? Yeah. Organization or questionnaire or matching agents. And what did you create them with? How did you create them? And where do they live now that you've created them? Yeah. Yeah. And just before that, maybe I'll just finish off on the system. Yeah. Sure. And then we'll dive deep because I absolutely want to show you that. But I want people to understand that, you know, in the front end in a nutshell, in an overview, it's much simpler than, you know, then we might understand because I think it's quite simple at the end. You're going to see in a second, you're going to be like, oh my god, he just made it sound so cool and it's simple things. They are simple things. You know, it's processes, SOPs, and then job descriptions. You know, as I create processes and all these processes go into a system and that system is like, who's going to run the system? Well, I need to 10 people and those 10 people are agents and each agent has a job description. Yeah. It's that simple. You know, like, I mean, it sounds simple when I say that. So my cell system, same thing, you know, it's about sales. And then, you know, you can see where things live. I'm going to show you the back end in a second, my onboarding system, you know, and then, you know, I know how to do these things and it knows it's affording the engagement journey, the onboarding, my trainings at 12 week, it has three phases. It's all mapped out. The guide system that we're just running, it's, you know, five to 10 page guides, two to three case studies, five linked and posted email templates. It's going to do all of this about that guide that we ask about Cloud Code. And this is only seven agents running this thing. It was about 36 agents yesterday and I kind of optimized this and I fired, you know, 29 of them again, you know, because I added a lot of new ones. I added these ones. I added these five core agents. And then I was like, you know, well, this is all my systems are much simpler and efficient than this was kind of cumbersome. It had 30, I don't know what the agent and it was a bit too much and I was getting 40 page guides, which were very good, but maybe it was, it was overkill, you know, so I ended up. And then, you know, I built that guide system that can build guide systems, you know, this was one of my most exciting things. But I'm going to show you now the backend, right? All right, so let's go in the backend again. But this time, if you don't mind, I'm going to open, I'm going to open a visual studio code. So I'm going to open an IDE, right? So where are people code again, right? But I'm not going to code. And why I'm going to open this is because you can read files a little bit easier, right? And I want us to read some files. So I went into my.clote folder, right? This is a folder that is in my computer. And inside this folder, I have all my code things, right? So I have, you know, I have my guide system that we just talked about, right? And I have all the agents here. So you were asking about the agent. So let's get the first agent that we actually running in the backend right now, right? Which is the guide orchestrator, right? So I just open this up. I'll close this to make it a bit cleaner for us. And maybe let me see if I can make this a little bit bigger. Yeah. So this is, this is an agent. And people are like, well, this is how, how does an agent supposed to look like? Anyways, nobody knows, right? Nobody knows. So I think it looks like this. You know, it, it has a name. It's a guide orchestrator. It has a description. It's a meta agent that orchestrates guide creation with five phases and seven agents, creates focused five to 10 page guides, no fluff, never creates content directly, launches specialized agents. So this agent's job is to be the orchestrator. It has some tools, right? It has some tools that task read the blog. But these are different tools that you can have in your terminal. So they can kind of create tasks and read stuff, you can find stuff, you can search for stuff and all that. And then this actually this one runs on Sonnet. So opus decided that Sonnet was more than enough for this, right? And in some, you might see that I have Haiku. Haiku is a good fast model by anthropic again. And yeah, again, for those of you who don't know every time of. I'm frappe comes up with with a new model. So now it's 4.5. There are eventually they know it always come with them at the same time, but eventually they have three models. Haiku is the smallest cheapest fastest one. Then there's Sonnet, which is the middle one, which is the one they always release first. And then there's opus, which opus 4.5 was just released last week, which is their biggest, most capable, and also most expensive model to run. But because team is running on a prepaid plan, he doesn't care how much cost because as long as you're running to your end of your allowance, then you are fine. And then you can run the opus as much as you want. That's exactly it. That's exactly it. And you know, sometimes I go to these agents for the older and I'm like, well, okay, let me switch this to opus or things like that. But but opus help me, opus 4.5 helped me, especially I'm not in the guide system. Sorry, in the latest system, I build opus 4.5 helped me in this one. I think it was opus 4.1 that helped me or it was Sonnet 4.5 when I did this. You know, and it decided that was Sonnet. I'm fine with it. The output is great. So and then, as you can see, it has a little bit about the description, what it's supposed to do as we said. So this, by the way, this is quite important because this is like the tag the metadata about this agent, right? You know, if you just, if an A, if a model has to read something, it will just read this to understand if it needs to use it or not. Sadly, these gets loaded in the context window, but with cloud skills, which is a new feature, you can have skills like this and you have this metadata that it doesn't read anything else. It just reads this and it knows that, oh, okay, I need to use this skill. Let me go read the other stuff. So it loads. I'll pause you for one minute because I don't want to jump into cloud skills. I probably need to go to a whole episode about cloud skills, because you might meet again. Yeah, I, I, be aware of what you wish for. But, but cloud skills are think about every automation you've built basic automations like a custom GPT, but think about packaging it in a way that it knows how to call it just when it needs it. So you don't need to go and copy the files and send it to your custom GPT. Cloud code says, oh, this requires this kind of data analysis. I have a skill that knows how to do it. And for me as an example, I have one that knows how to query my podcast. If I want to know something, when did I talk about this thing in my podcast, it knows how to do that, which otherwise it wouldn't. Or it has one that does things based on my brand guidelines. It has one that does very specific data analysis for me. And like Tim said, the important thing here is instead of blocking your context window, which means it's consuming part of the memory you can use in order to do the actual ask you what to do. It does not, unless it is actually required. And then it goes and opens, looks, basically opens the book reads what it needs to do. And then does what the book tells it to do. So these are cloud scales and it's very, very powerful. And the fact that you can run them inside of anything cloud, including cloud code is really, really cool. Absolutely. And I think that it's just still the beginning. But we do have the problem of the context window, right? The memory that it can remember. It's very small still. And you know, Gemini is doing 1 million tokens. And, you know, we do have cloud models that are 1 million, but they're not served to everybody. But we're supposed to get hopefully bigger content windows. We have about 40 minutes. So let's go. I will shut up. Yeah. No, no, no, I love it. And we'll give value. We'll give value very quickly now. So this is an agent, right? It has some principles. You know, these are forbidden. Don't not write content yourself. You know, and we don't get the pink elephant in the room problem anymore, right? Because prompt adherence, you know, if you say, don't do that, it doesn't do it. But before if you said, you know, don't do that, I'm gonna go do that. So he talked about it. He probably won't. It's like my four and a half year old daughter. You know, you tell her not to do it. She'll do it. You know, but I think that AI now, like large language models that are now a seven, seven years old, maybe not, you know, they act a little bit more like, they know that they'll get in trouble if they do it. So they don't do it. You know, a little bit better. And so this agent also is aware of the other agents, right? Because it's the orchestration. It's like, hey, these are your buddies that you're going to come in and you're going to use these buddies, you know, the guy, the researcher, the guy that I am and so on. And then it has a phase, it has a total time. I has a whole thing that is like, it's like a process, right? It's a good. So again, to make it simple for people who are not looking at the screen, it's simple English, like there's no magic. And it's just describing in a well structured way that I'm sure you use AI to help you structure to explain what this agent needs to do. How to do it and what it shouldn't do. That's basically what it is. Just in painting, like a job description, like you're hiring someone, you're going to give them a job description, you're going to tell them, hey, what you can do, what you can't do, what are the tools that you have access? Here's the logins, right? This is your job. Go to your job, right? So this is the same thing. And I recommend everybody, you know, when you're getting agents and so on, write a job description, you know, like onboard your your agents, you know, give them the same quality that you would give to a good team member that you've hired for a lot of money, right? That's important because that's that's what sets you for success. And then it has, you know, the actions and so on, SOP, right? So standard operating procedure. It knows exactly how to create a project structure, this creates folders. It has, you know, it tells it's what to say when it's done and some decision trees, some quality focus and some things that I should remember, right? This is an agent question, question about this just to verify. I assume you have a either a cloud skill or something like that that helps you write these in a well structured way. Yeah, so it's not even skills in most of the things in writing. It's my agents do a really good job because I'm breaking things into different agents. So I don't have one agent that will outline and write the whole thing, right? I have an agent that will do the research, will write it to a context file, then the next agent that is going to create the outline would look at the output of the researcher agent, right? And then it's going to create his own context saying, Hey, like I looked at that, here's the outline that I think. And then the next agent is going to take that outline, that's that outline that is the complex. And then the instructions don't get too complicated into cumbersome because we're doing something very much more specific and focused, you know, one is a good writer. And then the writer, I will say this is how I write. Here's some examples, right? And then I'm not eating up on the context window, neither. And by another question, because we're in a ID, if you want to reference a file. So I said, this is how I write. You want to give it three examples. Where do these examples live? Or you just give it a link to your folder inside your computer? Yeah. So in a lot of cases, I do have the past name to the file, because it knows in the computer. But if you also want to put an example, you could put another file and say, here's the file, but also you can, you can kind of embed it inside here. You could say, here is my tone, you know, you could kind of put a section here, the words I like to use here are the words that you can kind of do that. No, but I was thinking more of let's say you create one that's writing proposals, and you want to give it examples of two of your winning proposals, then you'll put a file, give it a path to a file. Yeah, exactly. So there's an agent at the end of the day. And let's just go back a stage, you know, so this is an agent inside my guide system. As you saw, there is a bunch of other agents. This is the enhancer. This is the one that is going to write some case studies and examples for the guide, right? And he's going to ask a bunch of questions. So this is a guide system. It's it lives here has agents has some other things like a knowledge base as you were saying. So, you know, here's my program. Here's my company training program. Here's our blueprint. Here's the company methodology. Here's my style guide. Actually, as you can see, there's a team checker style guide. There's a team checker writing style, right? And it knows because in the agent, you can say, Hey, I do have some style guides, you know, go read them. So you're keeping this in a knowledge base. So the guide system has a knowledge base. It has a memory as well, right? So it writes in the memory. It has some schema, which is empty at the moment. It has some templates, you know, so workflow state templates and project metadata templates. So to write those things, and so on each time the same, these are these are adjacent files, right? And has, you know, here's you should have the year month date, the topic, and so on. So it kind of gives very small, kind of template structure. Yeah, structure. Yeah, that's a good word to put it. Thank you very much. And at the end of the day, the more important thing is here is the workflow, right? So I showed you kind of, you know, the beautiful thing earlier, but at the end of the day, this, it's the five phase workflow using the seven agents, right? So this is why, if you remember, when I spoke, I said, make sure to use my guide creation workflow because then it does. It reads this first. It says, Oh, okay. Well, I have this. Oh, I have an agent that I'm going to first work with, right? And then it's guide researcher, and it's the orchestrator that works with it. Okay, then it works with the outline, then it works with the right way, then the polish, then the mark, then the enhancement, right? And then this is basically the handshake agreement between the different agents, right? That's what makes the content move from one agent to the other is this workflow. Yes, because it knows the step I'm going to work with this guy. Yeah, the step two, I'm going to do this, but I do something very different than a lot of people. And I think people really love their mind. And then the other day I showed is I don't let an agent, as we were just saying earlier, I don't let an agent speak to the next agent because they only speak to the orchestrator. They only speak to the orchestrator, but they also only write files to context files. Right? Because the output will be too much, and then the context window would go. But because they write it to a context file, and then the next agent, instead of looking at all things, it just picks up that context file that was written from the previous age. Yeah. And I'm able to keep much better. I were short on time, but I want to explain what you said because it's important. The context window is your limiting factor, right? This is the amount of data that can go in and out out of a single conversation. And it gets worse. And it gets worse. And if you take, let's say now it's the fifth agent, and he wants to know what happened in the first four steps and has to read everything that happened in each and every one of previous agents, it's basically going to run out of memory before it can do the thing that it needs to do. And what Tim is doing is having the only thing that gets transferred is basically the output versus the entire thing. And this is what he's calling context files. Obviously, okay, here's what I created. This is all you need to know. Like you don't need to know how I created it. What the data that is you don't know anything. Just read this. Here's what I created. Use this as your entry point to do your task, which saves a lot of space and efficiency and tokens. There's a lot of other reasons why this is a good idea. Brilliant. Exactly. Yeah. So that's a little hack that I've been doing. So that's basically one system. You know, it has it has a master workflow. It has agents. It has some memory. It has different things. It has templates. And this runs in this folder. Right? And I have a bunch of these, you know, I have a memory system, which this is basically every agent and every time I run Cloud Code, this is something new that I because I read the article a few weeks ago, and I was like, Oh, my God, I need systems that get better because I can't just keep, you know, working on my system. But instead, if I could feedback this memory system takes memory, takes things and saying, Oh, Tim said this, Tim, you know, gave me feedback on here. And then I'm building this memory system because then the next phase, it looks at the memory system and takes the learnings and does the learnings, right? Yeah. And it kind of updates the agents as well as we go along. Yeah. Yeah. So yeah. So every system, as you can see, has same agents being schema skills, you know, and so on. Right? So they all kind of work on the same thing. All my systems, you know, have similar, similar things and the agents are just there. It's, you know, it's simple. Yeah. So I want to, I want to really summarize this and then and then let people know where they can find and follow you and work with you. But at the end of the day, we went from, I know I want to do something to breaking this down to, okay, if I had to hire people to do this, what would I do? I will hire a person that will be their manager because I need five people and I don't want to manage them. Then I will need a person that does this, a person that does this, a person does this. And then you literally go in and explain what each and every one of them does. Just like Tim said, it's a job description, a very detailed job description because it's going to follow that exactly. It's not going to take any initiatives. So you've got to tell it exactly what it needs to do. As Tim said, the smaller you make it the better because in it's very specific and then hand it off to the next agent. It's not going to get confused what it needs to do and then create the plan of how does the handshake work? Once you're done, who do you hand this over to? And then when you run the process, you ask, quote unquote, the process to use this plan and then use the agents in order to complete the task. And then what happens is then the magic happens, right? Because when then you give it a task, it already has all these job descriptions and all the handshake functions on how to go from step one to step two, step two, step three, step four. So if you think about every process in your company, you can mimic this. If you can clearly describe what each of the steps are and how the steps lead from one another, which makes it so powerful and brilliant because you can apply this to literally anything in your business. And it's all 100% of it simple English. There's nothing here that you cannot do without writing any code. This was nothing short of incredible. I think what you built is mind blowing. But as you said, it's also very simple. And I know that people are going to listen to their first slide. Okay, we lost you in 30 minutes in. Listen to this again. And you'll see that there's nothing here that is rocket science. It's literally simple English explaining what you want that would happen and breaking it and structure it in a very well defined way. If I can add something, I think it's about 100%. It's about zooming out. I was, you know, since since chat GPT, you know, which has been three years and four days came out, I was very zoomed in and I was kind of, you know, doing all these things, prompts and so on. And then I realized if we zoom out and then we start looking at this as you said, it's much simpler. It's just a business with processes, a bunch of process system job descriptions and put them in there. And then you'll give them tools. The model context protocol MCP that I mentioned are like APIs. Imagine it's like a login. You know, I can give it an MCP to HubSpot to my CRM and it can update my HubSpot. It can do these things. And it's like getting a salesperson. I hire the salesperson. I'm like, here's your HubSpot login. So it's the same idea. It's just we, I think, overcomplicate things. But when we simplify things again and AI is an incredible colleague now for you to to build these systems, you just have to very much think about what do I do? Right? Yeah. What does this team member do? How do we do? What do we do? What do we do? Right? Because if you actually start asking these questions and you can write them down, the AI will help you build that anyways. You don't have to, you don't have to be a rocket scientist. Amazing. Tim, if people want to follow you, work with you, learn from you, hire you, what are the best ways to do that? Yeah, I think that's the best way is LinkedIn, you know, Tim Chakir, T-I-M, C-A-K-I-R is LinkedIn or my email,
[email protected]. You know, we do have training programs, we build stuff for people and so on. But we usually work with companies of at least 50 people, so we work with kind of mid SMBs, nothing under 30 or so people. Because that's where we really get see return and investment, right? Just FYI for everybody listening. And my YouTube, yeah, my YouTube is a new project. I mean, it's been a year. It's been something where how I'm sharing what I'm sharing today, I share everything that I know, every failure, everything I fix, I try, you know, and then from there you'll find for my links and you'll find news that there's a lot of free content I put there. I want to I want to teach everybody for free, but I have some clients so that they can pay so that I can kind of also share things for free for everybody. Same thing that I do. Absolutely amazing. You are absolutely brilliant. I appreciate everything that you're doing and what you're sharing and sharing with us today. Thank you so much and thanks everybody who joined us live and was in this in the back end and listening to what we're doing. And obviously if you're listening to this after the fact, come and join us in this live so we can ask questions as well. Thanks again.
PodcastTranscript AI