Most people still think building software is too hard. That was true until last year. I am not a programmer and I wanted a claw code tutorial that normal people could actually use. So I made this. If you've never written a single line of code, you will still go from a blank screen to something real in one afternoon in plain English. Whatever role or industry you're in, you can now build the exact tool your work was missing. Once you understand this simple four-step system in this video, you can build almost anything. So, let's get started.

If you look at claude code, it will feel intimidating at first. Everyone thinks you have to be a techie to build cool apps. They're wrong. Imagine tomorrow you have to negotiate an important deal in Beijing, but you don't speak a word of Mandarin, but you have a worldclass translator who converts it into Mandarin to your partners in Beijing. Now you can sit there, have a conversation, you can negotiate and close the deal without ever touching a single word of that language. software programming and building apps in Python or JavaScript. You know, they all feel like Mandarin if you've never seen them before. You can start building products, projects, and internal software tools without any coding today because Claude Code is your translator. If you can speak in English, you can get the deal done without knowing a single word of any programming language. Claude chat helps you think. Claude code work helps you do and claude code helps you build. So if you think that ah man coding is not for people like me, I want you to test that assumption in the next few minutes.

What would you need to build your first simple app? Not a single line of code. Now this is the desktop version of clawed app. At the top you can see claude chat and code. Select code. By the way, this is the universal icon for code everywhere. Down here is where I tell Claude what I want. You'll see words like repositories sometimes or GitHub. Ignore them for now. For this first build, let's use clot code locally. Pick a folder where you want the project to live and then type build me a beautiful 25inut focus timer with start, pause, reset, and a progress ring. That's it. And watch what happens.

Cloud code starts figuring out what it needs. Creating the files, writing the code, putting the pieces together. And just in a few minutes, there will be an actual application built for you. Let's open it. Here's the timer. Looks good. Start, pause, reset. Okay, all the buttons are there. So, that's the first moment when you start feeling, wait a minute, I built that. I can build this thing. You don't need to know what programming language Claude used or what all those strange messages that scrolled on the screen meant. Now look at the final product. Do you like the way it looks? If not, you can change it by just prompting. It's really that simple. That's the first big shift of how you use clawed code. You don't have to manage the code. You can manage the outcome. Now, we cheated a little bit here. A timer is deliberately a simple app. What happens when the thing you want to build is slightly more complicated? Well, let's go there next.

We'll take our first bite. The best way to build software tools and apps is to stop thinking about software. Imagine you're the director of a $200 million Hollywood blockbuster. You show up on the set and while the entire crew is waiting outside, you tear up the script and spend 3 hours rewriting the entire scene on a paper napkin. That would be so chaotic. But that's exactly what happened during a movie called Joker 2. And the movie was a mess according to many people.

Now contrast that with Mad Max Fury Road. The entire movie is like one big car chase through the desert. So, the director, George Miller, and his team map the entire movie through roughly about 3,500 storyboard panels. Each scene, each stunt was meticulously planned. And the funny part is, it's one of the most chaotic action movies ever made, and ironically, one of the best planned ones. So before you start building on any of these vibe coding tools, plan something, build something small, test it, and improve it. I like a simpler version that I made. I call it bite. You don't eat the entire watermelon at once, right? You take one small bite at a time. Bite is your map.

B is briefing. What are you trying to build? For whom and why. I is implement. Build the smallest useful version. T is test. Use it and see where it breaks. How do you improve it? And E is evolve. Good products evolve over time. And so do good builders. And by the way, if you like to get such tools and frameworks delivered to your inbox, please subscribe to my email. The link is below and it's totally free. I love writing it. All right, let's use this framework to build something slightly more useful.

Let's build a simple relationship manager for people who matter the most to me professionally, personally, former colleagues, my mentors, my partners, people I want to stay close to, but I can't because I'm running around. So, why not build something that's your own that's focused on just one question? Who matters to me? and how am I staying in touch with them? So, let's get started.

We'll start with B for briefing. AI can build faster than any human team, but at that speed, it needs a lot of direction. In your briefing, answer four questions. Who is this for? What problem are we solving? What does the first version need to do? And just as importantly, what are we not building in this product? Let's do it for real. I open clot code and give it a simple prompt. I want to build a personal relationship manager called Orbit. I just named it Orbit. Here's the problem I want to solve. I want to track how I stay in touch with people who matter to me. For version one, use the sample data I provided. Show each person's name, company, when we spoke last, and notes. Make it look professional, sparse, and Applelike. Use a muted color palette. Don't build any email or LinkedIn integrations yet. Before writing code, make a plan and ask me anything important that I have missed. And this is the most important part of the prompt. I am not a techie or a software engineer. Don't ask me to approve or review any technical commands. Make all the technical decisions yourself and only ask me questions about what the app should do in plain language. That's your example prompt. But you get my point, right? You can write your own prompt that is better than the one you just saw.

First, it'll ask you a bunch of questions about what you want to build. It'll clarify if there are any tradeoffs you need to make. And depending on what you were building, it'll ask you different questions. But it won't ask you questions about what kind of programming language it should write in or what kind of database you want to design. You can decide what you want and it decides how to build it. If you don't understand a bunch of things in the plan, you can ask questions. You can make suggestions to change things in the plan. Claude can update the plan anytime you give it feedback. In my case, after I read the plan, I realized I had left out one very important detail in the original prompt. The actual user experience. My prompt said nothing about what the layout would look like and what to show where. So I type in plain English. It should look professional and slick as if it's designed by Apple. Make it a threepane layout. Left sidebar with my categories. Middle column with list of names. When I click on someone, their profile opens on the right side. And after adding a little bit of that, I approve the plan. And now cloud code starts building.

To see what happens next, we move to I for implement. The biggest mistake trying to build everything at once. We'll dig into that important concept once Claude finishes building the first version. But you'll see that right now it starts creating and editing files, running commands, installing whatever it needs. If none of that means anything to you, great. It doesn't have to. Now look at the bottom and see if auto mode is available on your account. And if it is, you can turn it on and it will keep working without stopping for every permission that it needs to ask from you. For anything sensitive, you can slow down and review what it's doing, but usually you can sit back and enjoy your coffee. In my case, it ran for about 10 to 15 minutes. It wrote the entire code. It was a JavaScript code. It checked its own work. It made its own database and then came back with its first version. Now, here's what it looks like. The search works. I can type and see the people. That's good. Three PES just like uh we asked. All good. Let's see. The categories on the left are all accurate. It's good. It's genuinely usable. That's version one of Orbit.

And this is what you should notice about our version one. It is the simplest thing that is still useful in tech. There is a concept called MVP. not most valuable player but minimum viable product and I love this concept because it can be applied to any business any career any output all technology startups are obsessed with this one question what is the smallest version you can put in someone's hands that actually solves their problem because then you can observe how they use it what works what breaks what parts they like what parts they actually use that feedback tells you what to build next. And one sidebar, but a useful trick, I like using the strongest reasoning model for the planning and a cheaper, faster model for the implementation. It's like having a senior person help you strategize and a junior person handling the leg work.

Let's go to T for test. We have our minimum viable product. Now, let's test it and make it even more useful. Your first version works. Congrats. Great. Once you start using it, though, you'll see what's missing. For instance, in this app that we just built live, you realize the the names are there, the data is all there, the dates are all there, but there are no real insights that are obvious or actionable. So, let's go back to the prompt. Put three cards at the top. How many of my connections need attention? How many of them have gone cold? And how many follow up this week? That's a nice prompt. A few minutes later, there they are. Now, I open the same app and I have some kind of a dashboard at the top, not just data. When you're using clot code or any vibe coding tools, your goal is to get to the answer and make whatever you're building more insightful and more useful to you or the actual user. Then after a little more testing, you might find other gaps. For instance, when I reach out to someone, there's no way to tell, at least in the app, how I did it and when I did it. So you can go back to the prompt and add this add a button called log contact to each person's profile. When I click it, let me pick how we talked. Email, call, text, coffee. Then let me pick the date and let me add a quick note. And now Claude Code will go back and change the code. It'll rebuild some of the parts. When it's ready, I click on it. I check it out. Nice. It's working. Now you have started inching toward your own relationship management system. It's not a spreadsheet anymore. It's how you keep in touch with those who are close to you, those in your closest orbit.

And now we go to the final step. E for evolve. The final step is not just to evolve what you build, but to make sure you evolve with it. Anthropic did a study that looked at roughly 400,000 real sessions of people building software apps with AI. And the results were surprising. They found that people who had never programmed before succeeded almost as often as trained engineers and managers actually scored slightly higher than even software engineers. Now, it was just a survey and not a super rigorous study, but it does point to an interesting fact. The whole skill of vibe coding boils down to knowing what you want and describing it clearly to your staff, which is your AI coder in this case. So, the best people managers become the best AI managers.

So from here you can evolve in three different directions. First, the app you build can evolve into a a system. You can reuse the code to make a CRM for your side hustle, a tracker for your kids soccer team, a dashboard for your goals, whatever comes to mind. Second, if you feel like it, you can go a bit deeper. You can learn about APIs and how to connect with other apps, how to use GitHub, and what those Python libraries actually do. And every time you get stuck, don't be afraid to ask clarifying questions. Open Claw Chat or any chat and ask it to explain it like you are a 10year-old. And the third direction is try to learn something about coding. Even if your career right now has almost nothing to do with software because of all the things AI will change, software building is one of the areas where AI will improve incredibly quickly in the coming years. And there are reasons for that. But that also means that anyone with the right tools can create absolute magic. The world is still going to need great software engineers to understand complex systems and architectures and security and much more. But this new way of working and building your digital life is coming at you at 100 miles an hour. So spend a weekend tinkering with claude code or lovable or replet or maybe take an online course because the real goal is to become comfortable building, breaking and rebuilding things. What you build today will tell you what to build next. So go ahead and make a bad version. That's the only way you'll know how to make the best one. Thank you and I love you.