Why Everyone is OBSESSED With Claude Code

So, CL code is a black and white texton coding tool that only runs in an intimidating terminal. So, how did this

intimidating terminal. So, how did this happen?

Code three skills.

I used code bar code and became my second. To

second. To be honest, I've never seen a piece of software get this many people this obsessed this fast. And most of them are not even developers. There are so many other products that obsess about

perfectly crafted interactions, amazing design, ease of use, and they all end up failing. So why did they fail and this

failing. So why did they fail and this end up winning? This is a hard to install terminal only black and white text application that was never built to be optimized for retention or for

getting millions of people hooked. But

it did. So how did this happen? Well, I

looked under the curtain and now it all makes sense because once you understand the free key psychology and design principles behind Cloud Code, you'll see why this tool has become almost an addiction to so many people. Oh, and

literally the day I'm recording this video, the source code of the entire app leaked. We're going to also take a look

leaked. We're going to also take a look at exactly how they made this work behind the scenes. This video is sponsored by the friends of the channel, Advanta. More about them later in the

Advanta. More about them later in the video. But to understand this, let's

video. But to understand this, let's start simple. I'm sure you recognize

start simple. I'm sure you recognize this or this or even this one. And you

also know that these are very addictive, but not the social media type of addictive, but that's one more game or one more turn or one more whatever this is kind of addictive when you don't even

realize it's 3:00 a.m. and you're still here grinding. Well, using cloud code

here grinding. Well, using cloud code feels exactly the same. People are

actually losing sleep building the last feature, tweaking the new app or new system you just built. And that is by design. Because if you look closer, just

design. Because if you look closer, just another prompt feels kind of the same as just another turn in a strategy game.

Except these are beautiful games built by game designers and artists exactly for that purpose. And this is a black and white coding tool that runs in your terminal. On the surface, they look

terminal. On the surface, they look completely different. But this is

completely different. But this is actually the first reason why Clot Code got so many people hooked. Because by

complete accident, they nailed the most powerful game design principles out there. This is the art of game design.

there. This is the art of game design.

Basically the Bible for anyone building games. 113 principles for what makes a

games. 113 principles for what makes a game impossible to stop playing.

Principle 63. The feedback loop must be short. In a game, you might be

short. In a game, you might be completing a level which takes a few minutes or you're in an intense quick battle or yes, sometimes you need a little patience. But a great game is

little patience. But a great game is made up of many small loops of action and reward. And now take a look at this.

and reward. And now take a look at this.

You tell Claude what you want and reasoning, actions, file editing start happening. And a short while later you

happening. And a short while later you have your result. Action and reward. So

then new tweak, new prompt and repeat.

Each prompt is a short feedback loop. So

no, what makes it addictive is not how intelligent the AI model is. There's

something else that makes this kind of tool different than a CHBT or even just regular plot. Number 64, juiciness. This

regular plot. Number 64, juiciness. This

is a game design conference in Sweden in 2012. And this looks like a plain boring

2012. And this looks like a plain boring game. Just rectangles bouncing off of

game. Just rectangles bouncing off of one another. But then they started

one another. But then they started making one small change at a time. They

add a bit of color here. some sound

there, made the screen shake a little bit. Give the rectangles some wobbly

bit. Give the rectangles some wobbly eyes. The game stayed the same. The

eyes. The game stayed the same. The

rules stayed the same, but the juicy version felt more playful and fun. And

this doesn't seem like something that relates a lot with this. But look

closer. First of all, you've noticed if you ever use cloud models versus HGBT, they're much more friendly and conversational in tone. They don't feel like this. Or take a look at this. When

like this. Or take a look at this. When

cloth code is doing its thing, there's these unnecessary but fun pieces of text. And now that we have the actual

text. And now that we have the actual source code that leaked, we know that there are exactly 187 of them. And here

they are. It might seem like a small thing, but you don't need to go over the top to give your product just enough personality to make it juicy. And then

principle 55, visible progress. Every

game shows you some progress bar, some XP filling, a map revealed. In Cloud

Code, that's your entire project. You

start with an empty folder and then files start to appear. Features light

up. The file tree grows in real time.

And here's the thing, it's an open map game. There's no natural stopping point.

game. There's no natural stopping point.

If you ever played the strategy games, you know the feeling. Any moment there's multiple things about to happen and new technology in two turns. A building is ready in four. There's always some progress and something that just about to happen and something you're just

about to reach. Principle 36. Chance.

Unpredictable rewards are more addictive than guaranteed one. That's why slot machines work. Of course, Entropic tries

machines work. Of course, Entropic tries to minimize Claude's mistake. That's

normal. But the fact that it still makes some and it's not perfect, it's weirdly playing to their advantage here. At

times it comes up with something brilliant. Sometimes it's just okay and

brilliant. Sometimes it's just okay and 10% of the time it almost work, but it doesn't. And that's the slot machine

doesn't. And that's the slot machine moment. Because now once you're so

moment. Because now once you're so close, it's just another prompt, just another turn. And the game similarities

another turn. And the game similarities and game mechanics don't stop here because even watching the machine work is the reward. Factorio, Satisfactory,

Cookie Clicker. Researchers at Harvard call this the labor illusion where you see work being done for you, you value the results more. And that's exactly what clot code does. You describe what you want and then you watch it. You

watch it read files, write code, run tests, see errors, fixing them, running them again. And that's because while a

them again. And that's because while a generic AI tool for consumers would have hide this under a generic spinning wheel, you see this happening and that's the magic. Now, no one at Entropic sat

the magic. Now, no one at Entropic sat down and said, "Let's make this similar to games and let's make it addictive."

But this is where it starts to get interesting. Because the second reason

interesting. Because the second reason why everyone got obsessed with this is even weirder. Take a look at this tweet.

even weirder. Take a look at this tweet.

All of these products were built to be beautiful, friendly, designed to make AI approachable. And all of them are either

approachable. And all of them are either dead or dying. And then there's Open Claw or Cloud Code. Tools not built for regular people. They are not hiding

regular people. They are not hiding complex terminal commands or what bash is. They're hard to install. No buttons,

is. They're hard to install. No buttons,

no cute on boarding. And still they became a huge success. So what is going on here? This is copy paste. Brain dad

on here? This is copy paste. Brain dad

symbol. Two key combinations. You know

how it works. What makes it work is not that there is a copy paste for images or one for text or one for Windows or one for Mac. It just works everywhere. It's

for Mac. It just works everywhere. It's

powerful. So it goes in here. But in

order to make it powerful universal, your computer is managing clipboard formats, converting rich text, handling images, communicating between separate apps. Apple and their ecosystem has it

apps. Apple and their ecosystem has it working even between different devices.

So this is not really simplicity but perceived simplicity. This is called

perceived simplicity. This is called Tesla's law. Complexity can never be

Tesla's law. Complexity can never be destroyed. It can only be moved. And

destroyed. It can only be moved. And

yeah, Larry Tesler was the inventor of the copy and paste, by the way. Now,

many of the products that come out to date are actually simple and welldesigned and easy to use, but they're just not powerful. They're here.

Okay, this gadget looks sleek, but what do you mean the only thing it can do is book an Uber? What Cloudco did is genius. It started as a coding tool here

genius. It started as a coding tool here in the chart, but it moved itself over here. See, what makes it simple is not

here. See, what makes it simple is not that it never makes mistakes or that it has polished design, but the fact that it solves problems all the way back to being simple on its own. Let me explain.

See, I write these videos with a combination of notion for organizing things and cloud code with a set of custom skills I created. So, I asked it to synchronize the script of this video that I'm writing from notion so it can

read it and help me do some research.

So, it goes ahead, it tries to find an MCP server. And now that it's not

MCP server. And now that it's not working, it's like, "Oh my god, not working. Let me try the API." It goes

working. Let me try the API." It goes out on the internet, finds the documentation online and makes it work for me. The thing I did was very

for me. The thing I did was very complicated, but I perceived simplicity.

It's autofix in itself. You reach the maximum nirvana level when you ask cloud code to debug issues with cloud code itself. And this is basically Tesla's

itself. And this is basically Tesla's law on steroids. I got what I needed. I

didn't have to read documentation or know what an MCP is. It just did it. And

if I'm stuck, I just ask it to unstuck me. That's why it's so powerful, but

me. That's why it's so powerful, but also so simple. And this chart explains so many of the products you use every day and why they succeeded. Some

products fall here. They're simple and sleek and beautiful, but they can't do much. Tools like After Effects and

much. Tools like After Effects and Blender are incredibly powerful, but good luck learning how to use them. And

reason number three that makes you obsessed with cloth code, it's also what made it go from here to here. And now

that the source code leaked, I can just show you. But before we get to that, if

show you. But before we get to that, if you use clause code, you're likely a builder. Maybe you're a founder, CTO,

builder. Maybe you're a founder, CTO, designer or engineer. And after all this building, if you do everything right, your product starts getting traction and big customers come through the door.

They ask you about your security. Are

you SOCK 2 compliant or ISO 2701 compliant? This is where the friends of

compliant? This is where the friends of the channel Advant come in. As a founder or builder, you should focus on what you do best, just building the best product you can. Let Vanta automate your

you can. Let Vanta automate your security processes and maintaining compliance. It helps you achieve and

compliance. It helps you achieve and maintain compliance with over 35 security and privacy frameworks including sock 2, ISO 2701 and HIPPA.

And this does the best thing for you, which is freeing up your time and energy for the things that actually matter in the product. To all the fellow founders

the product. To all the fellow founders and builders that follow my channel, Vanta is offering $1,000 off at Vanta.

Just go to venta.com/enico. That's

vaina.com/enico

for $1,000 off. Oh, and feel free to share the link in the description to your founder friends as well. But now

back to the video because one weird thing about cloth code is that most of its users are actually not developers.

They are curious people. They follow

tech. But these people are using it to book airline tickets, file taxes, monitor tomato plans, come from museum archives. They're choosing to use a

archives. They're choosing to use a coding tool rather than the nice and purpose-built apps that we made over the years. And there's a reason why. See,

years. And there's a reason why. See,

the reason this coding tool became so powerful is because when you really look deep, it's code. Moving a folder, it's just a terminal command. manually

renaming 500 photos and organizing folders. It's a script that Claude can

folders. It's a script that Claude can write. Analyzing a spreadsheet, Claude

write. Analyzing a spreadsheet, Claude doesn't need the Excel UI that you use.

It just reads the data directly and whip up some Python code and you never have to read yourself to do what you need.

Most tax you do every single day are really some form of code just with an interface around it that makes it usable for humans. If you go on Skyscanner and

for humans. If you go on Skyscanner and you start to search for flights, you cannot ask, "Hey, find me flights with a one day layover in Singapore, but then that connect to Taiwan because I might want to go to the USA as well to find me

that the UI is fixed." But cloud code can just figure out that it needs to call all the APIs and get all the travel data and present the solution for you.

All done through code. So, it's not teaching people how to code. It just

removed the barrier that was hiding the code that was already there. And how?

Well, this is where the source code leak gets interesting. See, whenever you ask

gets interesting. See, whenever you ask it to do something, first it looks at your project, read your file, figure out what's going on. Then it picks a plan and starts doing things. Not just one thing, but it has over 30 different

actions and it picks the right one on its own. Can read files, write code, run

its own. Can read files, write code, run commands, and search the web. And if

something breaks, and it will, it reads the error, figures out what's wrong, and tries a different approach. An infinite

loop that keeps going until the job is done. Now, this is both extremely

done. Now, this is both extremely complex. There's like five different

complex. There's like five different strategies to compact the conversation and retain context. fake tools added so competitors have a hard time distilling how it works. But on the other hand, it's also quite dumb. Now, we can see

that they're just literally begging the model to please adhere to the instructions. Although you're logging

instructions. Although you're logging when you curse edit because you're frustrated and using that as a benchmark for how well it's working. But now this addictive loop and the fact that it got so popular and works so well, it's also

a big problem. The highest year is $200 a month. That's what I'm paying. And if

a month. That's what I'm paying. And if

you told me a few years ago that I was paid $200 a month for a piece of software, I will tell you, you're insane. But I routinely have three or

insane. But I routinely have three or four instances running in parallel all the time. One user spent $27,000 worth

the time. One user spent $27,000 worth of compute on a single $200 a month subscription. So they are losing tons of

subscription. So they are losing tons of money every single day. They managed to build a product that works so well that it worked too well. But there's

something else that's going on in the tech we use every day. Have you ever wondered why we have super smart AI chats that have to replace entire jobs?

We're spending billions developing useless products that no one will use.

But still, AirDrop doesn't work.

Printers are horrible to work with. It's

almost that our tech is running at two different speeds. Well, this is not by

different speeds. Well, this is not by chance. It's because of one of the

chance. It's because of one of the weirdest laws that decides how your tech actually gets built. And you can learn all about it in this video right here.

I'm Maro.