that. We'll take that code and then we'll go over to p5 js over here. In some ways, Claude is already winning because it can actually preview the code directly inside the browser when it's using p5 js. But let's plug this in. So, we got the code right here from the game. We'll compare...just It's weird that this has got a spinning camera. And also, I just think the UI looks a bit nicer inside Gemini. However, Claude did it faster, right? So, Opus coded a lot faster. Now, one of the ways the Claude wins, I would say, versus Gemini is that Claude can actually control your browser. So, for example
Поиск по транскрипциям
using claw code is it has a browser agent. So you can actually open up the page inside our browser and then test out and see how it performs. You can also use this inside the terminal as well. So if you actually go to like new terminal here and you prefer like the terminal field of claude...school community, which is great, right? And we literally just built that in one prompt. Even creating a nice little logo in the top left there using Claude Code and Visual Studio Code. Works really smoothly, super nice, makes it easy. Honestly, I think I still prefer anti-gravity, but if you want an alternative to anti-gravity
moving towards harnessing and then making skills generally available just as part of like their their LLMs. And so here you could see Anthropic has a page on their claude code docs that say extend claude with skills. These are the guys that sort of made skills up first. But because of popularity, a bunch of people um that were...developers and stuff like that started talking about it and asking for it. And so they've created their own skills format. It's very close to the cloud code skills format. And then also over here the Gemini CLI, you can see you can add agent skills. Uh basically no matter what framework or LLM or platform
very, very easy to understand from just a single source. So what I've gotten in the habit of doing is I will either on Claude AR or on Claude code set up my MCP servers to connect to all these various locations and I'll just ask it, "Hey, here's a couple of examples of past project updates...that the success rate on that is much, much higher than I originally thought. - Am I reading your status updates then in your Slack and those are all Claude-generated? - Yeah, I'm never actually writing anything anymore. It's just all Claude. - It was just you reading Claude the whole time. Okay, good to know. How about you John
model the ability to spawn up to 100 parallel agents that work together. So what is agent swarm? Think about it this way. Traditionally, when you give Claude Code or any AI coding assistant a complex task, it works sequentially. Step one, then step two, then step three. It's like having one really smart developer working alone. Agent Swarm
apps was like Hey all of them are great but Claud 3. 5 Sonet for code generation that's where it's at now we have 3. 7 Sonet and is much better now I should say that like you know since the release of 3. 5 Sonet people have been talking about 01 and 01 Pro and deep seek...curser or one of the other apps I keep throwing up my hands a lot because I'm like this is big but it is so that's what Claude code is okay working with it as simple as uh we pointed out I maybe let me just round out this blog article so we can close it and leave
this is pretty important if you didn't previously already have a habit of constraining your prompts. You can also add similar logic to your claude. md file for Claude Code. So you don't always have to write this. Another important point is that Opus
prepared pardon me. So I've pre-prepared a uh agents. mmd file here. So if we drag this in over here, this uh when you're working within claude code, it's called a claude. mmd. When you're working within Gemini, it's called a gemini. md. But when you're working within codeex or openclaw
make sure to leave a like, subscribe, turn on notifications. All I do is create amazing videos about AI. I also launched a Vibe Coding Academy. Has a full Claude Code Master Class in that and weekly calls of me. Check that out. Link down below. Let's keep adding new features here though. I think it would be cool...what are a few different ideas you have for implementing this system. So this is another thing you really should be doing as well. You should be using Claude as a partner, a creative partner just as much as you use it as a developer, right? You don't have to only just give it commands and prompts
case anymore. In fact, my second point in this newsletter was precisely an example of that. Many of you would have heard about the intense discussion surrounding Claude Code, and all sorts of Claude-powered agents that are now sweeping the web. So, we're seeing exponential improvement, right, across the board? Well, let's take one chess puzzle benchmark...just last week, Claude Opus 4. 6, five months further on, scored just 10%. That's not to knock Claude Opus 4. 6, I use it all the time, it's an incredible model at coding. And of course, if the AI labs wanted to improve this performance, they easily could. I think GPT 5. 2 on extra high gets
access the internet controls the most valuable real estate in tech. And now everyone's fighting for it. Number five, the coding agent explosion. Cursor $9 billion valuation. Claude Code $500 million plus run rate. And then there's the Windsurf saga. Open AAI tried to acquire them for $3 billion. Microsoft blocked it. Google hired the entire team...billion. launched it as anti-gravity all within 72 hours. That's how hot the coding agent market is right now. Number four, reasoning models changed everything. 01 Claude's extended thinking. Gemini's reasoning mode. These models don't just respond, they think. AI won gold at the International Math Olympiad started deriving new mathematical results humans hadn
only has it gone omega viral at 42 million views for its ability to automate non-coding tasks, the tool itself was produced within Clawude code powered by their latest frontier model, Claude Opus 4. 5, thereby seeming to justify the prediction that essentially all of the code would by now be written by AI. So wait, if they...their lines of code at the beginning of the year to typing barely any of them at the end of the year. I think of this as the Claude code experience, but for all forms of knowledge work. I also think that probably continual learning gets solved in a satisfying
marketing drives leads, copywriting drives sales, email campaigns drive revenue, social media content builds brands, and Claude dominates content creation. But here's what shocked me the most. Claude also dominates coding, real development work, functional applications, interactive tools. This means Claude can help you build the entire business ecosystem. Content to attract customers, code to serve customers, everything...need in one AI. Plus, Claude's consistency is incredible. Every single test, it delivered highquality results. You can trust it. Grock's word count issue in the blog post test is concerning. If it lies about something basic like word count, what else might it get wrong? Speed is nice, but quality matters more. Would you rather have
building AI agents to handle complex business tasks let's talk about some specific use cases where Claude 3. 7 Sonet really shines first up code generation as I mentioned earlier Claude 3. 7 Sonic is state-ofthe-art for coding it can handle tasks across the entire software development life cycle from initial planning to bug fixes maintenance...valuable for Rich code generation and planning this means it can handle much larger and more complex coding tasks without running out of space another great use case is Advanced chat Bots with enhanced reasoning and a warm human-like tone Claude 37 Sonic it's ideal for chatbots that need
creme unquestionably. In fact, it is so good at coding, it beat all the other engineering candidates that applied for anthropic. They gave the coding exam they give engineering candidates to Claude Opus 45. It beat every single engineering candidate that applied for anthropic. Every single one. And I'll tell you this, the coding tests they give to people...your conversations can basically be infinite length. They found a new way to progressively compact as you go. That is massive. Now you can have massive huge conversations in claude code that don't degrade over time. That is revolutionary and is going to make using AI so much better. On top of that, they
three, connect the front end to the back end. Ask Claude to generate the API integration code. It will write fetch requests, state management logic, error handling, everything you need to hook up the Stitch UI to your back end. And boom, you just built a full stack app in under 30 minutes. No joke. This is how fast...Versal or Netlefi. And now you have a fully functional lead capture system. That's the power of combining these tools. Stitch handles the design and front-end code. Claude handles the logic and backend infrastructure. You handle the vision and strategy and together you build things 10 times faster than before. Now, here's something really cool
then you can see Claude is now beginning to work through the to-do list and it will build it out locally. Now, I honestly think like Claude is the best for coding out of everything out there. And then for building a bigger website, that's why I recommend using claw code locally because you can store the files...allow it to build in the background whil whilst you're doing something else right whereas for example if you have this in the claw code chat directly on claude it's going to get messier and harder to handle because there's more like files stored locally so that's the way that I would
mode. And to enter plan mode, just press shift tab until you see plan mode here. And plan mode basically lets you brainstorm and draft plans of claude without having a code. And doing this can save you hours of going down the wrong path. So let's go ahead and put a prompt into cloud right here, which...just be able to get data from uh this third party tool seems to be the best option. And you can see down here that Claude actually wants to start coding right away. And when you see this, when you're in the planning stages of the product, you should definitely say no. I would not like to proceed because
actual agent I've used. probably like second to open cloud from what I've been hearing. But um for example, if I gave this like claude claude or like cloud code or chat or Gemini, it would probably not fully like watch the video etc. and transcribe it, break it down. Um for Madness, like they would actually...transcription etc. So it's actually doing everything granularly um and doing the task rather than making assumptions which is what you would get with you know something like claude or chat. Yeah, and I think Manis just launched their version of Open Claw. So, basically a hosted version of OpenClaw. I haven't played with it, but I think
remember all of this for later? That's where the next command comes in. It's called slashinit. Now, this is one of the most powerful commands in claude code. What it does is create a claude. md file. Think of it as claude's memory. Whatever you put in here, Claude remembers forever. No more copy pasting the same